Georgetown University Round Table

on Languages and Linguistics 1981

Analyzing Discourse
Text and Talk
Deborah Tannen
Editor
 Georgetown University Round Table
on Languages and Linguistics 1981

Analyzing Discourse:
Text and Talk

Deborah Tannen
Editor

Georgetown University Press, Washington, D.C. 20057

 Bibliographic Notice

Since this series has been variously and confusingly cited as:
Georgetown University Monographic Series on Languages and
Linguistics, Monograph Series on Languages and Linguistics,
Reports of the Annual Round Table Meetings on Linguistics and
Language Study, e t c . , beginning with the 1973 volume, the
title of the series was changed.
The new title of the series includes the year of a Round
Table and omits both the monograph number and the meeting
number, thus: Georgetown University Round Table on Lan-
guages and Linguistics 1981, with the regular abbreviation
GURT 1981. Full bibliographical references should show the
form:
Becker, Alton L. 1981. On Emerson on language. In:
Georgetown University Round Table on Languages and Lin-
guistics 1981. Edited by Deborah Tannen. Washington,
D.C.: Georgetown University Press. 1-11.

Copyright © 1982 by Georgetown University.

All rights reserved.
Printed in the United States of America.
Library of Congress Catalog Number: 58-31607
ISBN 0-87840-116-4
ISSN 0196-7207
 CONTENTS

Welcoming Remarks
James E. Alatis
Dean, School of Languages and Linguistics vii

Introduction
Deborah Tannen
Chair, Georgetown University Round Table
on Languages and Linguistics 1981 ix

Alton L. Becker
On Emerson on language 1

Walter J. Ong, S.J.

Oral, remembering and narrative structures 12

Robin Tolmach Lakoff

Persuasive discourse and ordinary conversation,
with examples from advertising 25

Frederick Erickson
Money tree, lasagna bush, salt and pepper:
Social construction of topical cohesion in a
conversation among Italian-Americans 43

Emanuel A. Schegloff
Discourse as an interactional achievement:
Some uses of 'uh huh' and other things that
come between sentences 71

Malcolm Coulthard and David Brazil

The place of intonation in the description
of interaction 9H

in
iv / Contents

Roger W. Shuy
Topic as the unit of analysis in a
criminal law case 113

Catherine E. Snow and Beverly A . Goldfield

Building stories: The emergence of information
structures from conversation 127

Georgia M. Green
Competence for implicit text analysis:
Literary style discrimination in five-year-olds 142

Joseph E. Grimes
Topics within topics 164

Teun A . van Dijk

Episodes as units of discourse analysis 177

J. L. Morgan
Discourse theory and the independence of
sentence grammar 196

V . Melissa Holland and Janice C. Redish

Strategies for understanding forms and
other public documents 205

William Labov
Speech actions and reactions in personal narrative 219

Charles J. Fillmore
Ideal readers and real readers 248

William Bright
Literature: Written and oral 271

Sally McLendon
Meaning, rhetorical structure, and
discourse organization in myth 284

Joel Sherzer
The interplay of structure and function
in Kuna narrative, or: How to grab a snake
in the Darien 306

John J . Gumperz
The linguistic bases of communicative competence 323

Ron Scollon
The rhythmic integration of ordinary talk 335
 Contents / v

Steven Feld and Bambi B. Schieffelin

Hard talk: A functional basis for
Kaluli discourse 350
Fred 'Doc' Bloodgood
The medicine and sideshow pitches 371
 WELCOMING REMARKS

James E. Alatis
Dean, School of Languages and Linguistics
Georgetown University

Good evening. Welcome to Georgetown University, the School

of Languages and Linguistics, and to the 32nd Annual George-
town University Round Table on Languages and Linguistics.
In previous years, when I chaired the Round Table, I was
never at liberty to say just how great a program had been put
together. This year, however, since the Round Table is the
work of Deborah Tannen and her able assistant, Susan Dodge,
I may say, with all modesty, that the program is impressive in-
deed.
In looking over the program this year I was amazed to note
that the pre-conference sessions present as wide and interesting
a range of topics as the conference itself. This is a tribute to
Dr. Tannen's energy and enthusiasm, as well as a mark of the
widespread interest that the Georgetown University Round Table
program generates. I was particularly pleased to note today's
pre-conference session on oral proficiency testing (1) because
it marks the continuation of our joint efforts with the inter-
agency round table and further cooperation between govern-
ment and university, and (2) because oral proficiency testing
is a field of language activity which is of common interest to
professionals in the fields of foreign language, EFL, ESL, and
bilingual education.
The topic of the Georgetown University Round Table on Lan-
guages and Linguistics 1981—Discourse Analysis—is a very ex-
citing one, and appropriate to a coming of age in linguistics.
Now that the tide has turned, it is safe for me to say in public
that, when I was introduced to linguistics, it was billed as the
key to the ultimate understanding of literature and discourse.
I am, therefore, very pleased to welcome you to a conference
that will indeed further understanding in this area.

vii
viii / Welcoming Remarks

I was going to make a few more rousing remarks, but I

understand that I have been upstaged by one of the world's
last medicine and side show pitchmen, Mr. Fred Bloodgood.
That's a tough act to follow, so I will simply turn over the
microphone to Deborah Tannen, with my thanks and appreci-
ation for a job well done.
 INTRODUCTION

The topic of the Georgetown University Round Table on Lan-

guages and Linguistics 1981 is 'Analyzing Discourse: Text and
Talk.' Perhaps a word is in order concerning the meaning and
use of the terms 'discourse', 'text', and 'talk'.
The subtitle, 'Text and Talk', can be understood to refer to
two separate modes of discourse: text as written prose, and
talk as spoken conversation. This is a common use of these
terms (for example, Cicourel 1975). But 'text' is often used
interchangeably with 'discourse'. Indeed, the term 'discourse'
is used in varied ways, to refer to anything 'beyond the sen-
tence'. The term appears in reference to studies of the struc-
ture of arguments underlying written prose (for example, van
Dijk in the present collection), and to analysis of pairs of hy-
pothetical sentences (for example, Bolinger 1979). However,
'discourse' is also used to refer to conversational interaction.
In fact, a book entitled An Introduction to Discourse Analysis,
written by a participant in this meeting (Coulthard 1977), is
concerned only with conversational interaction. Schegloff (this
volume) argues that rather than conversation being a sub-
variety of discourse, all forms of discourse are subvarieties of
conversation.
Discourse, as the term appears in the title, and as it is used
in the papers collected here, encompasses all these. It refers
to both text and talk, and these not as two separate genres to
be compared and contrasted, but rather as overlapping aspects
of a single entity. As the object of study, spoken discourse is
'text', much as words spoken in a speech are commonly referred
to as the text of the speech. In this sense, 'discourse' and
'text' are synonymous.
In a nonlinguistic discussion of what linguists know as the
Sapir-Whorf Hypothesis, Laing (1959) suggests that speakers of
English cannot conceive of mind and body as one, because their
language does not provide a word to express them so. The
best that English speakers can do is attempt to conceptualize
mindandbody, squishing them together but never really

IX
x / Introduction
perceiving them as a single entity. It is fortunate, therefore,
that there exists in English a word that refers to language in
context across all forms and modes. That word is discourse,
and that is the sense in which it is intended here.
Given this unified approach to discourse, it would be in-
felicitous to think of written and spoken language as separate,
that i s , of text as anything written and talk as spontaneous
conversation. The inadequacy of such a division is a recurrent
theme in recent research (see papers collected in Tannen 1982a
and 1982b). Features that have been associated exclusively
with spoken or written language are often found in discourse
of the other mode. For example, Bright (this volume) shows
spoken discourse to exhibit verse markers previously considered
poetic; Chafe (1981) finds spoken ritual Seneca to share many
features with written language; and written fiction exhibits
many features expected in spontaneous conversation (Tannen
1982c).
In their study of all forms of discourse, linguists are con-
cerned with central questions: of structure, of meaning, and
of how these function to create coherence. How do people put
words together? How do particular combinations of words yield
particular meanings? In short, what makes individual words
into discourse?
Discourse analysis raises another issue which is dramatized in
the following personal experience. Recently, my parents visited
me, and my father asked about my work: How do I really know
when I have made a discovery? How can I prove my findings?
How scientific is the study of language? I began to comment on
interpretive vs. statistical methods; that statistics may lie; that
sometimes it is necessary to look beyond what will fit into a test
tube, to understand what is in the world. My voice must have
taken on an intoning quality, because my father (who is a law-
yer) hesitated, looked at me, smiled slightly, and said, 'It
sounds as if you've had this discussion before, but I'm having
it for the first time, and that gives you an advantage'.
It is likely that many analysts of discourse have had this dis-
cussion before, from one or more of these perspectives. How
and to what extent can linguistics claim to be--and does it
aspire to be?--a science? The expansion (or, more accurately,
the return) of our sphere of study to discourse, to language in
context, raises more and more troubling questions of accounta-
bility, reliability, and verifiability; the role and nature of
interpretation, or hermeneutics; and, again and again, the
question of whether linguistics 1is one of the sciences, or of
the humanities, or of the a r t s .
Perhaps the choice is not really a choice at all. In a well-
reasoned argument identifying science as an art, Judson (1980)
quotes Nobel laureate physicist Paul Dirac: 'It is more important
to have beauty in one's equations than to have them fit experi-
ment' (p. 11). 'It seems that if one is working from the point
 Introduction / xi

of view of getting beauty into one's equations, and if one has

really a sound insight, one is on a sure line of progress 1 ( p .
199).
How can science be seen as an endeavor seeking beauty? For
one thing, in searching for explanations, science, like art, dis-
covers patterns and relationships. It seeks to understand the
exhilarating tension of creativity within constraints. Just so,
linguists seek to discover patterns that create and reflect co-
herence. Just so, the linguists whose work is collected here
have discovered the principles and processes underlying co-
herence in a wide variety of texts. Thus linguistics, at the
same time that it is scientific, is also concerned with aesthetics,
for aesthetics is (in the terms of Becker 1979, citing Bateson),
'the emergent sense of coherence*. An aesthetic response is
made possible by the discovery of the coherence principles
underlying a text.
In Christopher Hampton's play, The Philanthropist, a linguist
is introduced to a novelist, who asks him how he can bear to do
such narrow work. The linguist replies that he is interested in
the same thing as the novelist—words. The novelist, unim-
pressed, scoffs, 'But one at a time--not in a sequence'.
The study of discourse means that linguists are indeed inter-
ested in words in a sequence, and in that mysterious moving
force that creeps in between the words and between the lines,
sparking ideas, images, and emotions that are not contained in
any of the words one at a time--the force that makes words into
discourse.
Those who came to linguistics from the study of literature,
and those who came from mathematics, or anthropology, join
together in the study of discourse, seeking to discover patterns
in language--a pursuit that is humanistic as well as reasoned,
that is relevant at the same time that it is elegant, that is theo-
retical and empirical, and even beautiful.
The diversity of work in discourse analysis is reflected in the
papers collected here, and in the range of pre-conference ses-
sions that were organized in conjunction with the Georgetown
University Round Table on Languages and Linguistics, as can
be seen in the following list of session titles (organizers are
shown in parentheses).
1. Oral proficiency assessment (James Frith, Foreign Ser-
vice Institute)
2. Applications of discourse analysis to teaching: Spanish
and international affairs (William Cressey, Georgetown
University)
3. Toward adequate formal models of natural discourse
(Jerry R. Hobbs, SRI International)
4. Functions of silence (Muriel Saville-Troike, University of
Illinois)
5. Pragmatics (Nancy Yanofsky, Georgetown University)
xii / Introduction

6. Spoken vs. written language (Deborah Tannen, George-

town University)
7. Cordel literature in Brazil: Oral or written? (Clea
Rameh, Georgetown University)
8. Writing (Marcia Farr Whiteman, National Institute of Edu-
cation)
9. Linguistics, psychotherapy, and plain talk (Daniel P.
Dato, Georgetown University)
10. Discourse answers to syntactic questions (Flora Klein,
Georgetown University)
11. Discourse approaches to reading comprehension (Ulla
Connor, Georgetown University)
12. Association to cure monolingualism (Joshua Fishman,
Yeshiva University, and Dorothy Goodman, Washington
International School)
13. Discourse analysis and its relevance for translation and
interpretation (Margareta Bowen, Georgetown University)
Proceedings of many of these sessions will be published in
collections edited by their organizers. Papers from Hobbs'
session will appear in a special issue of the journal Text, and
papers from Tannen's session are included in Tannen (1982a)
and (1982b).
I want to thank the organizers of and participants in the
pre-conference sessions, and the participants in the plenary
sessions whose papers appear in this volume. Indeed, there
are many people--far more than I can name--who deserve heart-
felt thanks. First, I am grateful to Dean James E. Alatis for
giving me the opportunity to organize this year's Georgetown
University Round Table. I want to thank my colleagues,
especially Roger Shuy, for their generous support, and the
many Georgetown students who selflessly volunteered time and
enthusiasm. Finally, my deep thanks go to Susan Dodge, who
was at my side from start to finish, and without whose able and
cheerful assistance I cannot imagine this year's Round Table
having materialized at all.
Deborah Tannen
NOTES
These remarks have gone through a number of transforma-
tions, from typed notes to oral face-to-face discourse (a blend
of reading and extemporaneous talk) to typed transcription
(for which I thank Marta Dmytrenko) to revision for print. In
the last stage, I was helped by comments from Alton Becker,
Wallace Chafe, Robin Lakoff, Fr. Richard O'Brien, and Roger
Shuy.
1. Becker suggests, following Burke (1961), that linguistics
may be none of these, but something else entirely: a unique
epistemological realm.
 xiii / Introduction

REFERENCES
Becker, Alton L. 1979. Text-building, epistemology and
aesthetics in Javanese shadow theatre. In: The imagination
of reality. Edited by A. L. Becker and Aram A. Yengoyan.
Norwood, N . J . : Ablex. 211-243.
Bolinger, Dwight. 1979. Pronouns in discourse. In: Dis-
course and syntax. Edited by Talmy Givon. New York:
Academic Press. 289-309.
Burke, Kenneth. 1961. The rhetoric of religion. Boston:
Beacon Press.
Cicourel, Aaron V. 1975. Discourse and text: Cognitive and
linguistic processes in studies of social structure. Versus
12.33-84.
Coulthard, Malcolm. 1977. An introduction to discourse
analysis. London: Longmans.
Judson, Horace Freeland. 1980. The search for solutions.
New York: Holt Rinehart Winston.
Laing, R. D. 1959. The divided self. Middlesex and Balti-
more: Penguin.
Tannen, Deborah, ed. 1982a. Spoken and written language:
Exploring orality and literacy. Norwood, N.J.: Ablex.
Tannen, Deborah, ed. 1982b. Coherence in spoken and
written language. Norwood, N . J . : Ablex.
Tannen, Deborah. 1982c. Oral and literate strategies in
spoken and written narratives. Lg. 58:1.
 ON EMERSON ON LANGUAGE

Alton L. Becker
University of Michigan

'Emerson wrote by sentences or phrases rather than

by logical sequence.. .The unity of one sentence i n -
spires the unity of the whole—though its physique is
as ragged as the Dolomites.1
Charles Ives, Emerson, p. 23
'The maker of a sentence... launches out into the infi-
nite and builds a road into Chaos and old Night, and is
followed by those who hear him with something of wild,
creative delight. 1
R. W. Emerson, Journals,
October 18, 1834.

1. Comparative noetics. For a philologist, the task is to

build a road across time to an old text, or across space to a
distant one, and to try to understand it. It is a Utopian task,
but nonetheless essential, at least to the extent that such
understanding of distant texts is crucial in an age of rapidly
growing, increasingly powerful systems of communication, and
diminishing resources. What does it mean that Balinese watch
Star Trek on television, followed by a propaganda film on the
American space program, and mix the two? There are few
more difficult questions right now. Is there something like
noetic pollution, a spoiling of the noosphere by some cancer-
ous overgrowth? And where might we be--American linguists--
in the sweep of the noetic history we are just beginning to
write?
The term noetic is an old word in English; its turbulent
history is traced in the Oxford English Dictionary. Coleridge
used it to designate a science of the intellect, drawing on the
Kantian philosophers of Germany. Emerson knew Coleridge's
2 / Alton L. Becker

work well, and visited him on his first visit to England in

1833. In Emerson's early work, the mental realm--the noetic
sphere, or, as Pierre Teilhard de Chardin called it, the noo-
sphere--is called Spirit or Thought. Walter Ong has given
currency to the old word noetic. (In Javanese, the term for
reviving old words is jarwa dhosok, forcing--literally 'push-
ing'--old language into the present.) In his essay on the
drum languages of Africa, Ong (1977) defines noetics as the
study of the shaping, storing, retrieving, and communicating
of knowledge. He has also laid the groundwork for a new
kind of study: comparative and historical noetics, or looking
out in space to other cultures and back in time, even within
our own, at others' ways of shaping, storing, retrieving, and
communicating knowledge. Surely the whole process is lan-
guage-ful, which is what Benjamin Lee Whorf has had such a
hard time trying to tell us: that language is involved in the
whole process of shaping, storing, retrieving, and communi-
cating knowledge.
In the histories of particular noetic traditions, there appear
to be a very few powerful laws, among them a one-way se-
quence of changes from orality to writing to printing to second-
ary orality (this latter is Ong's term for the print-based,
electronic orality of our American present). Each stage in
this continuum seems to recontextualize the prior stage, so
that certain kinds of knowledge remain in the medium of the
past: most prayers are still spoken (or thought), many per-
sonal letters are still appropriately handwritten, and most
books are still printed, though photocopy would allow mass
circulation of manuscripts. As Ong and others have shown us,
each of these media entails its own noetic economy, its own
power and authority: the power of the voice, of the signed
document, of print, and of television. Furthermore, major
changes in noetic systems seem to happen initially in a few
places and then spread to other places, across language
boundaries. That is to say, noetic changes for most people
in the world have involved language contact, within which a
noetically more powerful language exports the secret of its
power to the weaker. What increases over time appears to be
the scope of central control, as a language adds media. In
making these statements, one is forced to hedge, with terms
like appear and seem, in acknowledgment that a new myth of
history is being shaped, and like most myths it is to some ex-
tent self-serving. As Sartre (1964) wrote: 'Progress, that
long, steep path which leads to me'.
Yet the myth has its power in helping one interpret changes
one sees going on. For instance, imagine a village, small, in
the hills, where knowledge is stored in the cultivated memories
of a few old people, often blind, often women. They sit in
special places in trials and other village meetings and tell who
is related to whom, and who did what, and who owns what bit
of land, and what the last ruler said. And now imagine what
 On Emerson On Language / 3

happens to them when a young person who can read and write
the new national language returns, after a few years of school-
ing, to the village. The old blind people enter their own new
dark ages. This is a dramatic, but oft repeated, instance of
noetic change.
Or imagine what happens when a foreign colonial language,
always more powerful noetically, replaces--as a source of
knowledge--important functions of the local language. Some-
times a voice across time and space lets us share a feeling of
noetic change. A Javanese poet writes, in a world in which
understanding of the present was shaped by Dutch, in the
late nineteenth century:
Anglakoni zaman edan
Ewuh aya ing pambudi
'(We) walk in an unstable world
Not at home, struggling against our own imagination1
I have written elsewhere (Becker in press) about this remark-
able poem, and the difficulties we have in reading it and
understanding it; the English meanings at the deepest levels
we must abandon, even as we encounter meanings of tense
and person as background cohesion, and the presence of
elaborate focus and deixis--to mention only a few of the more
obvious grammatical differences one encounters.
The paradox of philology (to paraphrase Ortega y Gassett
1959) is that distant texts are always both exuberant and de-
ficient at the same time. I read too much in, and I am unper-
ceptive of what is there, and so I understand only through
successive approximation, by giving up--unexpectedly--various
etic aspects of English and slowly getting attuned to new emic
possibilities of language. Noetic exploration into terra incog-
nita is, of necessity, a very slow and very difficult process of
abduction.
2. Pontification. But one can go back into one's own lan-
guage, too, as Michel Foucault has done in French, to an-
other episteme, another noetic era. Many years ago I began
a lingual biography of Emerson. He left us an abundant
record of letters, journals, lectures, and essays from through-
out his life--plenty of data, and the primary philology has
been done very well. It is all laid out chronologically, well
annotated. Furthermore, Emerson was self-conscious about
language itself and was keenly interested in the ideas and
books the New England ships brought back from India. His
European correspondents kept him in touch with the explora-
tion of Sanskrit. For many generations of academics his 'The
American Scholar' has provided our own version of the Hippo-
cratic oath. However, the task of understanding Emerson on
language is not unlike other philological excursions: it is
slow and difficult.
 I Alton L. Becker

Emerson's own language changes noticeably when he leaves

America on Christmas day, 1832. He was very sick, his 19-
year-old wife had died, and he had resigned his prestigious
Unitarian pulpit in Boston, agitated with religious doubt. He
was sailing to Italy in search of health. Up until then, his
writing was quite conventional. When he arrived in Italy, well
and energetic, the voice we recognize as Emerson's has
emerged, and one begins to read him as the words quoted
earlier suggest, 'with something of wild, creative delight'.
Just one sample: he writes his brother William from Messina:
The fault of travellers is like that of American farmers,
both lay out too much ground & so slur, one the insight
the other the cultivation of every part. Aetna I have
not ascended. (Rusk 1939:364)
Writers in different times and places innovate at various
levels of language, some at the levels of sentences, like Emer-
son. (The reader is invited to parse the sentence just quoted,
in whatever methodology seems comportable.) Some writers
innovate at the level of words, like James Joyce. Some inno-
vate in drawing on new sources of prior text, like the Irish
mythology which Yeats both shaped and drew from. Some
innovate in the language act they perform, like the opening
of the inner newsreel in Virginia Woolf or, again, Joyce. It
is very Emersonian to ask: what are the ways people can inno-
vate in language? In what different ways can we deviate from
the norms we inherit? Emerson, like his literary descendent,
Gertrude Stein, is an innovator in sentences.
His sources vary, and he sometimes uses odd words, but the
language act he performs is constant. Kenneth Burke (1966:5)
has called it pontification:
pontificate; that is, to "make a bridge." Viewed as a
sheerly terministic, or symbolic, function, that's what
transcendence is: the building of a terministic bridge
whereby one realm is transcended by being viewed in
terms of a realm "beyond" it.
As a writer of sentences, Emerson slows us way down. We
academics have come to expect innovation at higher levels and
a certain 'stereotypicity' at the level of the figure of a sen-
tence. We can read rapidly only if the lower figures are regu-
lar. Some see our field of linguistics as one big text we are
all engaged in writing, and of necessity we must agree,
therefore, to certain conventions, and still our individual
voices. In what language acts do personal constraints matter?
If there is one idea popularly associated with Emerson, it is
his celebration, like Gertrude Stein's, of the individual voice
shaping sentences. 'The maker of a sentence.. .launches out
into the infinite and builds a road into Chaos and old Night,
 On Emerson On Language / 5

and is followed by those who hear him with something of wild,

creative delight'.
3. Norm and deviation: The individual voice. The most
difficult task of the philologist is to hear the individual voice.
In reading an Old Javanese text, I have very little sense of
what is stereotypic and what is innovative, what is norm and
what is deviation. That means that an essential feature of
aesthetic response, deviation (however so slight it may be in
a traditional genre), is inaccessible. More often than not, in
a new language one sees everything as unfamiliar innovation.
A few years ago, when I wrote about constraints on the crea-
tion of a Javanese shadow play (Becker 1979), where etymo-
logizing is an opportunity for innovation, where stories have
no climax, where several languages are used simultaneously--
the repeated response of my Javanese friends was, 'What's so
new about that?' Their stereotypes were innovations for me,
and therein lies the odd aesthetic excitement of philology.
Emerson saw the American scholar as a deviation from the
European norm. A new context brought new meanings, a new
relationship with the world. But before he presented his well-
known lecture, 'The American Scholar', he published a small
book called Nature, in which there is a section called 'Lan-
guage'. If we read it as one must an Old Javanese text, not
as something to agree or disagree with but rather first as
something to understand, it makes an interesting noetic jour-
ney for a linguist, back to our own chirographic age, as it is
preserved in print. In his own day, as many people heard
Emerson as read him, even as far off as Kalamazoo or Ann
Arbor. Though the book Nature was prepared for printing,
Emerson's sentences were hand-shaped, first, in his letters
and journals. The journals became sources for lectures, the
lectures sources for essays. The essays are talks to be read,
meant to be heard, with the voice in them, shaped by hand-
writing. The best way to enjoy Emerson is to read him aloud,
slowly. I suspect it may be impossible to read him fast.
4. Nature as a source of constraints. Nature, Emerson
writes, is 'all that is separate from u s , all which philosophy
distinguishes as the not me1. The aim of science, he says,
is 'to find a theory of nature'. x He lists those aspects of
nature not only unexplained but previously thought inexplica-
ble: sleep, madness, dreams, beasts, sex, and language.
Central to the essay, then, is the image of I and Other, a
particular ego and its context. How does an 'I' relate to its
context? Emerson lists four ways:
I use nature as a commodity
I use nature as a source of beauty
I use nature as a source of language
I use nature as a source of discipline
6 / Alton L. Becker

Note the order in the list. I relate to the world as first sub-
ject to my will and finally as subjecting me to its discipline,
with aesthetics and language in between.
The section on language explores the nonarbitrariness (or
iconicity) of the relations between language, nature, and
thought. For Emerson, language consists of signs for Nature
(as he defined it, the 'not me1), while Nature in turn is sym-
bolic for what he calls Spirit, but which we might call, after
Bateson (1979:89-128), mind, what phenomenologists call
2
noema.

5. Iconicity and double metaphors. One sentence from the

section on language of Emerson's Nature first struck me a few
years ago when I was trying to understand the root metaphors
behind the Burmese system of classifiers (Becker 1975): 'Parts
of speech are metaphors because the whole of nature is a meta-
phor of the human mind' (p. 18).
This is a puzzling statement. It seems to require some ad-
justment in our thinking in order to make it appear true. It
is not self-evidently true.
There is some prior text. In a lecture entitled 'The Uses
of Natural History', given in 1833, Emerson expressed a simi-
lar idea (Whicher, Spiller, and Williams 1959:24):
The strongest distinction of which we have an idea is
that between thought and matter. The very existence
of thought and speech supposes and is a new nature
totally distinct from the material world; yet we find it
impossible to speak of it and its laws in any other lan-
guage than that borrowed from our experience in the
material world. We not only speak in continual metaphors
of the morn, the noon and the evening of life, of dark
and bright thought, of sweet and bitter moments, of the
healthy mind and the fading memory, but all our most
literal and direct modes of speech--as right and wrong,
form and substance, honest and dishonest... are, when
hunted up to their original signification, found to be
metaphors also. And this, because the whole of Nature
is a metaphor or image of the human mind.

The deeply metaphoric nature of one's own language is most

clearly seen across cultures, in that 'passage to India' that
motivates the philologist-linguist. Emerson had a mistaken opin-
ion about the language of 'savages': he thought it was like
the language of children (Sealts and Ferguson 1969:15). It
remained for his student, Henry Thoreau, to begin to under-
stand the language of 'savages' by living close to them and
listening. Yet Emerson's notion that parts of speech are
metaphors is strongly confirmed in the work of the compara-
tive philologist, for whom even nouns and verbs lose their
 On Emerson On Language / 7

iconicity, as does the notion of grammar itself. Languages

seem to select from nature one or another pattern--a set of
regularities to build coherence around: temporal sequences,
perspectives from the speaker or from the hearer, the basic
dramatis personae of case, the distinctions of the sexes, or
the division between changing actions and stable things--all
the etic icons that cohesion may be built around. These
regularities perceived in nature are all, in a sense, available
in nature--in the relation of person and context--to build
language around. The deepest regularities are the most
backgrounded features of language--the most iconic. (By
iconic I mean felt by the observer--culturally defined--to be
the most natural--as ordinary speakers of English feel tense
to be a natural fact, not a lingual metaphor, or as Burmese
speakers feel their classifiers to be in nature, mirrored in
language.) To learn a new distant language is, in Emerson'a
terms, to develop a new relationship with nature, a new set
of iconicities, at least in part. 'Parts of speech are metaphors
because the whole of nature is a metaphor of the human mind'.
Or, as Gregory Bateson (1979:17) put it, 'contextual shaping
is only another term for grammar'.

6. A figure for defamiliarization. 'Parts of speech are

metaphors because the whole of nature is a metaphor of the
human mind'. This is a favorite Emersonian figure, the
strategy of the double metaphor. It dominates this section of
his essay, where about 75 percent of the sentences are equa-
tive strategies, simple and complex. A figure is a sentence
or larger unit conceived of as a substitution frame, in which
certain points are more open to substitution than others, for
example, the figure of a recent riddle, 'How many does it
take to change a lightbulb?', in which one substitutes a name
of a human category, such as a nationality, a profession, a
religion, etc. The answers evoke the stereotypes of each
category. (A new point of substitution, and a new impact,
came later, when someone substituted 'government' for 'light-
bulb'.) All language3 can be conceived of as sets of partially
remembered figures.
The general shape of Emerson's figure might be rendered
as: 'X is Y because the A of B is a Y of C It is a com-
plex equative, or identificational strategy, the figure of defi-
nition and of metaphor, of overlays. For Emerson and others
in the Kantian tradition, metaphor is a strategy of reason,
which, as readers of philosophy know, meant just the opposite
then of its normal present meaning: it meant intuition--direct
apprehension, 'first thought'. Part of the difficulty in moving
into another episteme (in time or space) is learning not only
new words but new meanings for old ones, a variety of what
has been called ostranenije 'defamiliarization' or even 'alieniza-
tion', making things strange, 1 * or what Burke (1964) calls
8 / Alton L. Becker

'perspective by incongruity'. Defamiliarization is essentially a

metaphoric process: seeing the familiar in a new way. Java-
nese makes one see English in a new way: it defamiliarizes it.
Javanese is a metaphor of English, and vice versa.
Within the metaphoric figure that recurs in the text, Emerson
plays with three classes of terms: terms for language, terms
for nature, and terms for what he calls spirit or mind. What
are the relations of language, nature, and thought? The cen-
tral proportion, expressed in the sentence we are considering,
is that 'Language is to Nature as Nature is to Thought'.
'Parts of speech are metaphors because the whole of nature is
a metaphor of the human mind1.
7. Linguistics of particularity. There have been hundreds
of books and articles written about this one essay by Emerson,
but only Kenneth Burke penetrates below the macrostructure
to the kinds of microstrategies Emerson uses in order to make
possible higher level meanings. Is there not artistic and
philosophical creation at the level of sentences, a play of
figures and lexical classes at a level below that usually noticed
by the literary scholars, and a bit above that usually studied
by linguists, a level where, as Isidore of Seville wrote, 'gram-
mar is joined to the art of rhetoric' (quoted in Murphy 1974)?
It is here that we encounter what Kenneth Pike has called the
linguistics of particularity, and Paul Ricoeur calls discourse.
For Ricoeur (1971:529-562), discourse differs from language
(i.e. Saussurian 'langue') in that discourse has a particular
writer or speaker, a particular reader or hearer, a particular
time, and a particular world. I would add, any discourse
also evokes a particular set of prior texts for the participants.
A discourse can be understood only in its particularity.
For the study of particular discourse, we need techniques of
textual parsing which include all the kinds of discourse vari-
ables which constrain its particularity--which help shape it.
Such techniques will have to allow us to move across levels of
discourse and discern the different kinds of constraints oper-
ating at various levels: word, phrase, clause, sentence,
paragraph, monologue, exchange.
One of the reasons why I never found local grammars of
Southeast Asian languages, although I looked hard for them,
was that Southeast Asians do not customarily view language,
I recently realized, at the level of the clause and the sen-
tence. In Java, for instance, people have a rich vocabulary
(much richer than ours) for what they call undo usuk, i.e.
the choices of words or phrases in given positions within
figures, (l/nda usuk means literally the parallel wooden strips
on a pitched roof on which one hangs ceramic tiles.) This set
of terms represents a conventional understanding of para-
digmatic choices. One might imagine a traditional Javanese
student of language beginning his or her study of English
with a description of the difference between 'Close the door'
 On Emerson On Language / 9

and 'Shut the door1. That is, the Javanese would begin with
paradigmatics rather than with syntagmatics, with constraints
on substitution rather than immediate constituents.
Returning to Emerson's figure—X is Y because the A of B
is a Y of C--one notices that it includes two equative clauses,
although rhetorically they might better be called identificational
clauses. In general, identificational strategies operate at noun
phrase level, and--as is well known--many languages have no
equivalent of an identificational copula; in those languages,
equational clauses are structurally identical with noun phrases.
Without a copula, the copula strategies are awkward to express.
In the West, copula strategies characterize some of our most
important figures: definitions, syllogisms, generics, even
passives—all our most evaluative figures. In pontification
(i.e. writing the moral essay) copula strategies dominate.
(It is interesting to think of the essay as the reverse of nar-
ration, to some extent: in narration, narrative strategies
dominate, and are evaluated by, among other figures, generic
copula strategies; while in the moral essay, copula strategies
dominate and are evaluated by short bits of exemplary narra-
tive.)
Within copula strategies, Emerson relates three sets of terms
to one another over and over again, offering evaluative in-
stances: terms for language, terms for nature, terms for
spirit or mind. A double metaphor is established, in which
language is metaphoric of nature, and nature--now considered
as text--is metaphoric of mind. The essay establishes termi-
nistic depth, via a sequence of overlays. Each time a term for
nature, mind, or language reappears, it has acquired more
context. 'Parts of speech are metaphors because the whole of
nature is a metaphor of the human mind.'
This sentence, like the fragment of a hologram, projects an
image of the whole essay, Nature. 5 This double metaphor is
at the heart of transcendentalism, where grammar, rhetoric,
and epistemology meet, in a figure 'as ragged as the Dolomites'
(Ives .1970:23). To understand it means to be reshaped by it,
to let it defamiliarize one's world. It means to think and write
for a moment like Emerson, who concludes his section on lan-
guage with these words:
'Every scripture is to be interpreted in the same spirit
which gave it forth,' is the fundamental law of criticism.
A life in harmony with Nature, the love of truth and vir-
tue, will purge the eye to understand her text. By de-
grees we may come to know the primitive sense of the
permanent objects of nature, so that the world shall be
to us an open book and every form significant of its
hidden life and final cause.
A new interest surprises us whilst, under the view now
suggested, we contemplate the fearful extent and multitude
10 / Alton L. Becker

of objects, since 'every object rightly seen unlocks a new

faculty of the soul.' That which was unconscious truth
becomes, when interpreted and defined in an object, a
part of the domain of knowledge,--a new amount in the
magazine of power, (pp. 18-19)
The goal of this short excursion into a few lines of Emerson's
'Language' has been to think about another episteme, another
conceptual world, another noema, another mind. Our difficulty
on so many levels—including the grammatical—in reading Emer-
son's words as currently relevant knowledge, tells us much
about ourselves and the distance—even within our own culture,
within our own language, and within our own field of study—
of the conceptual world of 1836. The task requires, I think,
that we change ourselves as readers, moving from an etic to an
emic understanding, by imagining a world in which those words
could be true: 'Every scripture is to be interpreted in the
same spirit which gave it forth'—words Emerson quotes from
George Fox—might also be the fundamental law of modern
philology, and hence a discipline for one important approach
to the study of text.
NOTES
Dedicated to Marvin Felheim, 1914-1979. Thanks for critical
readings of a prior draft to J. Becker, A. Yengoyan, and D.
Tannen.
1. Quotations from Emerson's Nature are from Sealts and
Ferguson (1969). These passages are from page 5. Page
references for later quotations will be cited in the text.
2. For an overall view of phenomenological methodology, see
Ihde (1977). 'Noesis' and 'noema' are discussed on pages 43-
54.
3. Most medieval rhetorics and grammars included discussion
of figures, or common deviations from ordinary language. The
number of figures varied, but it was common to distinguish 45
figures of diction and 19 figures of thought. Ten of the
figures of diction were called tropes. Later writers described
up to 200 figures. For details, see Murphy (1974). I use the
term figure here to designate a minimal text strategy.
4. This term is fully explicated in Stacy (1977). The term
was first used by the Russian critic, Viktor Shklovsky.
5. The hologram has given rise to a new sense of whole-
ness. See Bortoft (1971:43-73).
REFERENCES

Bateson, Gregory. 1979. Criteria of mental process. In:

Mind and nature. New York: E. P. Dutton. 89-128.
 On Emerson On Language / 11

Becker, Alton L. 1975. A linguistic image of nature: The

Burmese numerative classifier system. International Journal
of the Sociology of Language 5.109-121.
Becker, Alton L. 1979. Text-building, epistemology, and
aesthetics in Javanese shadow theatre. In: The imagination
of reality: Essays in Southeast Asian coherence systems.
Edited by A. L. Becker and Aram Yengoyan. Norwood,
N.J.: Ablex.
Becker, Alton L. (in press) The poetics and noetics of a
Javanese poem. In: Spoken and written language: Explor-
ing orality and literacy. Edited by Deborah Tannen.
Norwood, N . J . : Ablex.
Bortoft, Henri. 1971. The whole: Counterfeit and authentic.
Systematics 9. 2: 43-73.
Burke, Kenneth. 1964. Perspectives by incongruity. Bloom-
ington: Indiana University Press.
Burke, Kenneth. 1966. I, eye, ay--Emerson's early essay
"Nature": Thoughts on the machinery of transcendence.
In: Transcendentalism and its legacy. Edited by Myron
Simon and Thornton H. Parsons. Ann Arbor: University
of Michigan Press.
Ihde, Don. 1977. Experimental phenomenology: An intro-
duction. New York: Capricorn Books.
Ives, Charles. 1970. Emerson. In: Essays before a sonata.
Edited by Howard Boatwright. New York: Norton.
Murphy, James J. 1974. Rhetoric in the Middle Ages.
Berkeley: University of California Press.
Ong, Walter J. 1977. African talking drums and oral noetics.
In: Interfaces of the word: Studies in the evolution of
consciousness and culture. Ithaca: Cornell University
Press.
Ortega y Gassett, Jose. 1959. The difficulty of reading.
Diogenes 28.1-17.
Ricoeur, Paul. 1971. The model of the text: Meaningful
action considered as text. Social Research 38.529-562.
Rusk, Ralph L., ed. 1939. The letters of Ralph Waldo
Emerson. Vol. 1. New York: Columbia University Press.
Sartre, Jean-Paul. 1964. The words. Greenwich, Conn.:
Fawcett.
Sealts, Merton M. J r . , and Alfred R. Ferguson, eds. 1969.
Emerson's Nature: Origin, growth, meaning. Carbondale,
111.: Southern Illinois University Press.
Stacy, R. H. 1977. Defamiliarization in language and litera-
ture. Syracuse: Syracuse University Press.
Whicher, Stephen E., Robert E. Spiller, and Wallace E.
Williams, eds. 1959. The early lectures of Ralph Waldo
Emerson. Vol. I. Cambridge, Mass.: Harvard University
Press.
 ORAL REMEMBERING AND
NARRATIVE STRUCTURES
Walter J. Ong, S.J.
Saint Louis University

1. Of all verbal genres, narrative has the most evident and

straightforward relationship to memory. Narrative is funda-
mentally retrospective. Even a live radio broadcast of a foot-
ball game tells you what is just over with. A science fiction
story cast in the future is normally written in the past tense.
The creative imagination as such has curious alliances with
memory. Wordsworth insisted that poetry 'takes its origin from
emotion recollected in tranquility'. I have heard a fiction
writer from the Midwest explain passionately to an audience
how in creating a story his entire imaginative activity is one
of memory. What he is making up--his plot and his charac-
ters--present themselves to him as remembrances of what has
been. Creative activity is nostalgic. This is why the artist
who deals with the future, as perhaps artists now must, is
under particularly heavy strain.
All narrative, moreover, is artificial, and the time it creates
out of memory is artificial, variously related to existential
time. Reality never occurs in narrative form. The totality
of what has happened to and in and around me since I got up
this morning is not organized as narrative and as a totality
cannot be expressed as narrative. To make a narrative, I
have to isolate certain elements out of the unbroken and seam-
less web of history with a view to fitting them into a particu-
lar construct which I have more or less consciously or uncon-
sciously in mind. Not everything in the web will fit a given
design. There may be, for example, no way to fit in the
total series of indescribable moods that I lived through in the
few moments after I was shaving. It is hard even to distin-
guish these clearly from everything else that was going on in
and around me. So, with almost everything else, such matters
are dropped in favor of more standard topoi. Writers such as

12
 Oral Remembering and Narrative Structures / 13

James Joyce or William Faulkner enlarge the number of topoi--

to the dismay of complacent readers—but the number remains
always limited. The totality of existence-saturated time is
simply too much to manage. 'There are still many other things
that Jesus did, yet if they were written about in detail, I
doubt there would be room enough in the entire world to hold
the books to record them' (John 21:25).
Something other than the events themselves must determine
which events the narrator cuts out of the incessant and dark
flow of life through the density of time and frames in words.
He or she must have a conscious or unconscious rationale for
the selection and shaping. But what the rationale is in any
given case it is difficult and often impossible to state fully or
even adequately. 'Jesus performed many other signs as well--
signs not recorded here--in the presence of his disciples. But
these have been recorded to help you believe that Jesus is the
Messiah, the Son of God, so that through this faith you may
have life in his name' (John 20:30-31). Here a rationale is
stated. It is what we call 'salvation history': the author
picks from Jesus' life what is particularly relevant to human
beings' salvation. But such statement is exceptional. Few
historians can put down so straightforwardly as the author of
the Fourth Gospel the rationale they have settled on for their
selection and structure of events. Even fewer fiction writers
could adequately state their own rationale when they bring
into being the artificial construct that we call narrative.
There is no reason why they should have to. But neither is
there any reason why we should not ask what the rationale is.
The ways of articulating memory, of bringing to mind and
representing the past, are various. They differ from culture
to culture and from age to age. In particular, as we have
become increasingly aware, bringing to mind and representing
the past is quite different in oral cultures from what it is in
cultures such as our own where writing and print, and now
electronic processes, have been interiorized so deeply that
without great learning, skill, and labor we cannot identify
what in our thought processes depends on our appropriation
of writing and the other technologies into our psyche, and
what does not. Often oral narrative processes strike us as
divergent from what we consider 'normal', whereas in fact
many mental processes which seem 'normal' to us have only
recently been feasible at all.
Here I propose a few reflections concerning oral noetic pro-
cesses centering on the way memory and narrative plot are
related in some primary oral cultures as contrasted with the
ways they are related in chirographic and typographic cultures
and electronic cultures, the ones to which we ourselves are
closer. I understand plot in the ordinary sense of the tem-
poral and causal sequence in which events are presented in a
narrative.
14 / Walter J . O n g , S.J.

2. Memory, in its initial role and in its transformations, is

in one way or another a clue to nearly everything that went
oh as discourse moved out of the pristine oral world to literacy
and beyond. Memory is still with us, but it is no longer with
us in the way it used to be.
The retention and recall of knowledge in primary oral cul-
ture calls for noetic structures and procedures, largely formu-
laic, of a sort quite unfamiliar to us and often enough scorned
by us. One of the places where oral noetic structures and
procedures manifest themselves most spectacularly is in their
effect on narrative plot, which in an oral culture is not quite
what we take plot typically to be. Persons from today's liter-
ate and typographic cultures are likely to think of consciously
contrived narrative as typically designed in a climactic linear
plot often diagrammed as the well-known 'Freytag's Pyramid':
an ascending action builds tension, rising to a climactic point,
which consists often of a recognition or other incident bringing
about a peripeteia or reversal of action, and which is followed
by a denouement or untying—for this standard climactic linear
plot has been likened to the tying and untying of a knot.
This is the kind of plot Aristotle finds in the drama (Poetics
1451b-1452b)--a significant locale, for Greek drama, though
orally performed, was composed as a written text and was the
first verbal genre, and for centuries was the only verbal
genre, to be controlled completely by writing in the West.
Ancient Greek oral narrative, the epic, was not plotted this
way. In his Ars Poetica, Horace writes that the epic poet
'hastens into the action and precipitates the hearer into the
middle of things' (lines 148-149). Horace has chiefly in mind
the epic poet's disregard for temporal sequence: the poet re-
ports a situation and only much later explains, often in detail,
how it came to be. He probably has also in mind Homer's con-
ciseness and vigor (Brink 1971:221-222): he wants to get im-
mediately to 'where the action is'. But whatever these further
implications, literate poets eventually interpreted Horace's in
medias res as making hysteron proteron obligatory in the epic.
Thus John Milton explains in the 'Argument' to Book I of
Paradise Lost that, after proposing 'in brief the whole Subject'
of the poem and touching upon 'the prime cause' of Adam's
fall, 'the Poem hasts into the midst of things'.
Milton's words here show that he had from the start a con-
trol of his subject and of the causes powering its action that
no oral poet could command. Milton had in mind a plot, with
a beginning, middle, and end (Aristotle, Poetics 1450b) in a
sequence corresponding temporally to that of the events he
was reporting. This plot he deliberately dismembered in order
to reassemble its parts in a consciously contrived anachronistic
pattern.
Exegesis of oral epic has commonly seen oral epic poets as
doing this same thing, imputing to them conscious deviation
from an organization which was in fact unavailable without
 Oral Remembering and Narrative Structures / 15

writing. Such exegesis smacks of the same chirographic bias

evident in the term 'oral literature'--which is to say, 'oral
writing'. Radically unfamiliar with the psychodynamics of a
given phenomenon, you take a later or secondary phenomenon
and describe the earlier or primary phenomenon as the later
or secondary phenomenon reorganized. Oral performance is
thought of as a variant of writing, and the oral epic plot as
a variant of the plot worked out in writing for drama. Aris-
totle was already doing this sort of thing in his Poetics (1447-
1448a, 1451a, and elsewhere), which for obvious reasons shows
a better understanding of the drama, written and acted in his
own chirographic culture, than of the epic, the product of an
oral culture long vanished.
In fact, an oral culture has no experience of a lengthy,
epic-size, or novel-size climactic linear plot, nor can it imagine
such organization of lengthy material. In fact, it cannot
organize even shorter narrative in the highly climactic way
that readers of literature for the past 200 years have learned
more and more to expect. It hardly does justice to oral compo-
sition to describe it as varying from an organization it does
not know, and cannot conceive of. The 'things' that the action
is supposed to start in the middle of have never, except for
brief passages, in anyone's experience been ranged in a
chronological order to establish a 'plot'. There is no res, in
the sense of linear plot to start in the middle of. The res is
a construct of literacy. It has to be made, fictionalized, and
it cannot be made before writing. You do not find climactic
linear plots ready formed in people's lives, although real lives
may provide material out of which such a plot may be con-
structed. Any real Othello would have had thousands more
incidents in his life than can be put into a play. Introducing
them all would destroy the plot. The full story of Othello's
whole life would be a bore.
Oral poets characteristically experience difficulty in getting
a song under way: Hesiod's Theogony, on the borderline be-
tween oral performance and written composition, makes three
tries at the same material to get going (Peabody 1975:432-433).
Oral poets commonly plunged the reader in medias res not be-
cause of any grand design, but perforce. They had no choice,
no alternative. Having heard perhaps scores of singers sing-
ing hundreds of songs of variable lengths about the Trojan
War, Homer had a huge repertoire of episodes to string to-
gether but, without writing, absolutely no way to organize
them in strict chronological order. There was no list of the
episodes nor, in the absence of writing, was there any possi-
bility even of conceiving of such a list. If he were to try to
proceed in strict chronological order, the oral poet would on
any given occasion be sure to leave out one or another epi-
sode at the point where it should fit chronologically and would
have to put it in later on. If, on the next occasion, hypo-
thetically smarting under the earlier disgrace, he remembered
16 / Walter J . O n g , S.J.

to put the episode in at the right time, he would be sure to

leave out other episodes or get them in the wrong order.
Neither he nor any other poet ever had the poem by heart at
all. Oral narrative poets do not memorize a poem word-for-
word , but only a potentially infinite number of recitations or
rhapsodies of formulas and themes in various configurations,
depending on the particular situation. (The Greek rhapsoidein
means to stitch together song.)
Moreover, the material in an epic is not the sort of thing
that would of itself yield a climactic linear plot in any event.
If the episodes in the Iliad or the Odyssey are rearranged in
strict chronological order, the whole has a progression but it
does not have the tight climactic structure of the typical
drama. It might be given that sort of structure, as might
the real life of a real person, by careful selection of certain
incidents and bypassing of others. But then most of the epi-
sodes would vanish. An epic put in straight chronological
order remains a loose concatenation of individual episodes,
with only very weak climactic progression. There is really
no res, in the sense of linear plot, in the epic, waiting to be
revealed.
What made a good epic poet was not mastery of a climactic
linear plot which he manipulated by dint of a sophisticated
trick called plunging his hearer in medias res. What made a
good epic poet was--among other things, of course--tacit
acceptance of the fact that episodic structure was the only
way and the totally natural way of handling lengthy narrative,
and possession of supreme skill in managing flashbacks and
other episodic techniques. Starting in 'the middle of things'
is the original, natural way to proceed for lengthy narrative
(very short accounts are perhaps another thing). Lengthy
climactic linear plot, with a beginning, a middle, and an end
is essentially artificial. Historically, the classic dramatic plot
is a literate transmutation of episodic procedure, not vice
versa. If we take the climactic linear plot as the paradigm of
plot, the epic has no plot. Strict plot comes with writing.
Why is it that climactic linear plot comes into being only
with writing, comes into being first in the drama, where there
is no narrator, and does not make its way into lengthy narra-
tive until more than two thousand years later with the novels
of the age of Jane Austen? Earlier so-called 'novels' were all
more or less episodic, although Mme de La Fayette's La
Princesse de Cl&ves (1678) and a few others are less so than
most. The climactic linear plot reaches a kind of plenary form
in the detective story--relentlessly rising tension, exquisitely
tidy discovery and reversal, perfectly resolved denouement.
The detective story is generally considered to have begun in
1841 with Edgar Allen Poe's The Murders in the Rue Morgue.
Why was all lengthy narrative before the early 1800s more or
less episodic, so far as we know, all over the world (even r
Lady Murasaki Shikibu's otherwise precocious The Tale of
 Oral Remembering and Narrative Structures / 17

Gen/O? Why had no one written a detective story before

1841?
The answers to these questions must be sought in a deeper
understanding of the history of narrative than we have thus
far had, an understanding beginning from the fact that in
lengthy oral narrative climactic plot is not really central to
what the narrative is 'about', to the aims of the narrator, or
to the audience's participation and enjoyment. Structuralist
analysis by Claude Levi-Strauss (1970 and elsewhere) and
others has revealed some of the organizing principles of oral
narrative as these can be described in terms of binary themes
and parts. But structuralism leaves out a lot of what is go-
ing on. It is unfamiliar with much relevant scholarship,
largely American.
3. Some new insights into the relationship of memory and
plot have been opened in a recent lengthy work by Peabody
(1975). Peabody builds on the work of American scholars
now famous for their pioneering work on oral epic, notably
Milman Parry and Albert Lord and (less obviously) Eric Have-
lock, as well as upon work of earlier Europeans such as
Antoine Meillet, Theodor Bergk, Hermann Usener, and Ulrich
von Wilamowitz-Moellendorff, and upon some cybernetic and
structuralist literature. He situates the psychodynamics of
Greek epos in the Indo-European tradition, showing the inti-
mate connection between Greek metrics and Avestan and Indian
Vedic and other Sanskrit metrics, and the connections between
the evolution of the hexameter line and noetic processes. This
larger ambience in which Peabody situates his conclusions sug-
gests still wider horizons beyond: very likely, what he has
to say about the place of plot and about related matters in
ancient Greek narrative song will be found to apply in various
ways to oral narrative in cultures around the entire world.
And indeed Peabody, in his abundant notes, makes reference
from time to time to Native American Indian and other non-
Indo-European traditions and practices.
Partly explicitly and partly by implication, Peabody brings
out the negative correlation of linear plot (Freytag's Pyramid)
and memory, as earlier works were unable to do. He makes
it clear that the true 'thought' or content of ancient Greek
oral epos dwells in the remembered traditional formulaic and
stanzaic patterns rather than in the conscious intentions of
the singer to organize or 'plot' the narrative a certain way
(1975:172-179). 'A singer effects, not a transfer of his own
intentions, but a conventional realization of traditional thought
for his listeners, including himself (1975:176). The singer
is not conveying 'information' in our ordinary sense of 'a pipe-
line transfer' of data from singer to listener. Basically, the
singer is remembering in a curiously public way--remembering
not a memorized text, for there is no such thing, nor any
verbatim succession of words, but the themes and formulas
18 / Walter J. Ong, S.J.

that he has heard other singers sing. He remembers these

always differently, as rhapsodized or stitched together in his
own way on this particular occasion for this particular audience.
'Song is the remembrance of songs sung' (1975:216).
Creative imagination, in the modern sense of this term, has
nothing to do with the oral epic (or, by hypothetical exten-
sion, with other forms of oral narrative in other cultures).
'Our own pleasure in deliberately forming new concepts, ab-
stractions, and patterns of fancy must not be attributed to
the traditional singer' (1975:216). The bard is always caught
in a situation not entirely under his control: these people on
this occasion want him to sing (1975:174). The song is the
result of interaction between him, his audience, and his
memories of songs sung. Since no one has ever sung the
songs, for example, of the Trojan Wars in any chronological
sequence, neither he nor any other bard can even think of
singing them that way. His objective is not framed in terms
of an overall plot. In modern Zaire (then the Democratic
Republic of the Congo), Candi Rureke, when asked to nar-
rate all the stories of the Nyanga hero Mwindo, was astonished
(Biebuyck and Mateene 1969:14): never, he protested, had
anyone performed all the Mwindo episodes in sequence. We
know how this performance was elicited from Rureke: as he
narrated, now in prose, now in verse, with occasional choral
accompaniment, before a (somewhat fluid) audience for 12
days, three scribes--two Nyanga and one Belgian--took down
his words. For print, the text, of course, had to be mas-
sively edited. How the entire Iliad and Odyssey were elicited
from a singer and doubtless edited to give us our texts in
the complete--though still episodic--forms in which we have
them, we do not quite know, but it was very likely in some
similar fashion.
What the singer remembered in ancient Greece--and seemingly
also in modern Zaire--were themes and formulas, although the
formulas are apparently more obtrusive in the Greek than in
the Nyanga story because the Greek is all in verse, while the
Nyanga is a mixture of verse and prose narrative. For the
simple reason that the singer has never heard a linear plot in
chronological sequence from beginning to end of a lengthy nar-
ration, he does not remember such a chronological sequence—
though he may keep pretty close to temporal sequence in
shorter narrative of a few lines generated out of a theme. He
cannot create a linear plot out of his ordinary resources, since
he is a rememberer, not a creator--in the sense of a creative
narrator, though he is a creator in the sense that he creates
an interaction between this specific audience, himself at this
particular period in his development, and the memories he has.
In this sense, a full linear plot of the Iliad and the Odyssey
in chronological sequence never existed in anyone's imagination
or plans, which is the only place it could have existed before
writing. These and other comparable oral performances came
 Oral Remembering and Narrative Structures / 19

into existence episodically, and in no other way. Episodes is

what they are, however masterfully strung together.
By contrast, the situation was utterly different with Milton
when he sat down to compose Paradise Lost aloud, for, even
though he was now blind and composing by dictating, he was
doing essentially the same sort of thing he did when he
learned to compose in writing. His epic was designed pri-
marily as a whole. Milton could have his dictated lines read
back to him and revise them, as an oral narrator can never
revise a line spoken or sung: Milton was creating, not
remembering. Though he of course used some memory in his
creating, it was not the communal memory of themes and
formulas that Homer had dwelt in. Paradise Lost is not the
'remembrance of songs sung' as the Iliad and the Odyssey and
The Mwindo Epic are.
Peabody's profound treatment of memory throws bright new
light on many of the characteristics of orally based thought
and expression, notably on its additive, aggregative character,
its conservatism, its redundancy or copia (which helps pro-
duce the constant feedback that characterizes oral thought
development and gives it its often bombastic quality), and its
participatory economy.
Of course, narrative has to do with the temporal sequence
of events, and thus in all narrative there is some kind of
story line. As the result of a sequence of events, the situ-
ation at the end is subsequent to what it was at the beginning.
Nevertheless, memory, as it guides the oral poet, often has
little to do with strict linear presentation of events in temporal
sequence. The poet will get caught up with the description
of the hero's shield and lose completely the narrative track.
We find ourselves today, in our typographic and electronic
culture, delighted by exact correspondence between the linear
order of elements in discourse and the referential order, the
world to which the discourse refers. We like the sequence in
verbal reports to parallel exactly what we experience or can
arrange to experience. When narrative abandons or distorts
this parallelism, as in Robbe-Grillet's Marienbad or Julio
Cortazar's Rayuela, the effect is utterly self-conscious: one
is aware of the absence of the normally expected parallelism.
Oral narrative is not much concerned with exact sequential
parallelism, which becomes an objective of the mind possessed
by literacy. Parallelism between the sequence of events in a
narrative and its real-life referent was precociously exploited,
Peabody points out, by Sappho and gives her poems their curi-
ous modernity as reports on temporally lived personal experi-
ence (1975:221). By Sappho's time (fl. c. 600 B.C.) writing
was already structuring the Greek psyche. But there is little
of this parallelism at all in epos—or, for that matter, in other
discourse in oral cultures (Ong 1967:50-53, 258-259). Similarly,
'narrative description in the epos is seldom the description of
an Aristotelian causal chain' (Peabody 1975:214). Philosophical
20 / Walter J. Ong, S.J.

and scientific analysis are entirely dependent on the interioriza-

tion effected by writing.
Thought in oral cultures develops, but it develops with
glacial slowness, for individuals cannot move far from the tra-
dition in which oral culture stores its knowledge without losing
both their auditors and themselves (Ong 1967:231-234). 'The
amount of effort, inventive imagination, and technical skill'
needed for an oral performer to work up in recitation (his
only resource for working out thought) a store of new infor-
mation discovered and organized by himself is simply pro-
hibitive, as Peabody points out. For an individual to work out
with conscious intent truly original thought on any appreciable
scale, 'some time-obviating mechanism like writing is necessary
to organize, formulate, and realize' the thought. Instead of
being analytically linear, oral thought is highly redundant and
echoic: this is the only way it can proceed, by feedback
loops out of and into itself (1975:173-176).
Further details of Peabody's creative extension and deepening
of recent scholarship on orality are too complex and at times
involuted for full treatment here, especially since his most
fecund and wide-ranging discussions emerge from his primary
concern with the sources of the pentameter line as such--
sources which are treated not just metrically, however, but
psychodynamically in full social contexts. One can question
or qualify certain features of his argument, as Havelock has
done in a lengthy article-review (1979), but one must at the
same time affirm, as Havelock forthrightly does, the incontro-
vertible value of a work 'so close to the realities of the oral
situation' (1979:189) and to the mentality that uses the oral
medium in what I have called a primary oral culture. Pea-
body's work shows how profoundly and suggestively our
understanding of the shift from orality can be expanded and
deepened.

1. With writing, and even more intensively with print and

the computer, the operation of human memory is drastically
altered and the noetic processes that mark oral cultures are
transformed. Writing, and later print and computer, enable
knowledge to be stored outside the mind--though of course
only after a manner of speaking, because there is no knowl-
edge outside the mind. Unless a human mind knows the code
for interpreting writing, letters on a page are no more knowl-
edge than random scratchings would be. It is what is in the
mind that makes the letters signify. Writing and its sequels
do not, strictly speaking, store 'knowledge' outside the mind
but rather set up structures outside the mind which enable
the mind to engage in intellectual activity otherwise unavailable
to it. This is why writing and its sequels become more effec-
tive as the human mind 'interiorizes' them more and more, in-
corporates them into itself by adjusting itself to using them
without having to reflect on them, so that operations
 Oral Remembering and Narrative Structures / 21

supported by the technology of writing, and later of print and

electronics, seem to it as normal as its unsupported natural
operations. Today, as we have seen, most literates are totally
unaware that their most characteristic kinds of thinking--those
that organize a school textbook or even a newspaper article,
for example--are unavailable to oral peoples, are not 'natural'
at all in the sense that they cannot be carried on by a mind
unaffected, directly or indirectly, by writing. Chirographically
conditioned or implemented thought is technologically powered,
although at the same time it i s , of course, 'natural' in the
sense that it is totally natural to man to devise artificial tech-
nologies to improve his own native prowess.
A helpful analog for the use of writing is performance on a
musical instrument. Musical instruments are tools, the products
of technology, totally outside the human being who plays them,
as a written or printed text is outside the reader. When
Beethoven composed, he was imagining sounds made with tools
('instruments' is the more commendatory word, but it says the
same thing), and he was writing down exquisitely specific
directions on how to manipulate the tools. Fortissimo: hit
the keys very hard. Legato: do not take your finger off one
key until you have hit the next. An orchestral performance
is a demonstration of what human beings can do with tools.
Some music-making devices, such as a pipe organ, are not just
tools but huge, complex machines, with sources of power sepa-
rate from the player. Electric guitars and other electrified
instruments are machines, too. Modern instrumental music,
whether that of an electronic orchestra, a classical orchestra,
or a piano or a flute, is a technological triumph. Try to make
a musical instrument and see.
And yet, the violinist has interiorized his tool, the violin
cradled in his arm, so deeply that its sound seems to come
from the very depths of his soul. He has humanized the
tool, interiorized it, so that to him and his audience it seems
a part of himself. The organist similarly humanizes, inter-
iorizes a huge machine whose mechanism he by no means fully
understands. This is what human beings have also done with
the technologies of writing, print, and computers. Writing
can be compared to a violin or flute or trumpet, print to a
piano (far more mechanical and dismayingly complex), the com-
puter to the pipe organ or electronic organ.
Human memory does not naturally work like a written or
printed text or like a computer. Natural, orally sustained
verbal memory is redundant, essentially and not by default,
echoic, nonlinear, and, unless supplemented by special inten-
sive training, it is never verbatim for any very lengthy pass-
age. Rather, it is thematic and formulaic, and it proceeds by
'rhapsodizing', stitching together formulas and themes in
various orders triggered by the specific occasion in which the
rememberer is remembering. It works out of and with the
unconscious as much as within consciousness. Peabody has
22 / Walter J. Ong, S.J.

made the point that the performance of the oral narrator or

singer is very little determined by conscious intent: the
formulas and themes--the tradition--control him more than he
controls them. This is the ultimate difference between human
memory and language on the one hand and, on the other, the
retrieval and communication systems set up by writing and
print and electronics. Human memory and language grow out
of the unconscious into consciousness. Writing and print
and electronic devices are produced by conscious planning--
though of course their use, like all human activities, involves
the unconscious as well as consciousness.
We have interiorized tools and machines so deeply that we
are likely--unconsciously--to use them as models for human
activity. We think of retrieval of material from the written or
printed text or from the computer as the model of native human
memory. It is not an adequate model. Human memory never
recalls simply words. It recalls also their associations. Liter-
ate scholars of an earlier age, predisposed to take as the model
of mnemonic activity the literate's verbatim memorization of
texts, commonly supposed that since oral peoples clearly had
prodigious memories, they had prodigiously accurate verbatim
memories. This supposition is untrue. Literates can normally
reproduce a lengthy narrative in metrically regular verse only
when they have memorized a preexisting text. Oral performers
can reproduce a lengthy narrative in metrically regular verse,
as Lord has shown (1960), by quite different techniques--not
by learning verbatim a string of words but by expressing
themselves with vast stores of metrically tailored formulas,
accommodated to the narrative themes and forms they know.
Out of these formulas they construct metrical lines ad libitum
which tell the same story in perfect metrics but not at all
word for word. Memory here works not out of an actively
conscious attempt to reconstitute perfectly a string of words,
but rather out of a passion for what Peabody styles the 'flight
of song'.
This is not to say that oral cultures never ambition or
achieve verbatim repetition, particularly in ritualized recitation
learned by intensive drill (Sherzer 1980:2-3 of typescript,
and n. 3; 1981, n. 3). But even ritual is not typically alto-
gether verbatim, either in oral cultures or in chirographic
cultures retaining a heavy oral residue. 'Do this in memory
of me' (Luke 22:19), Jesus said at the Last Supper. Chris-
tians celebrate the Eucharist because of Jesus' command. But
the words of institution which he spoke before he gave this
command and to which the command refers, the words that
Christians repeat in their liturgy as Jesus' words ('This is
my b o d y . . . ; this is the cup of my blood...'), are not set
down in exactly the same way in any two passages of the
New Testament that report them. The command to remember
does not call for verbatim memory. Even in a text here, the
still highly oral Christian Church remembered in the original,
 Oral Remembering and Narrative Structures / 23

resonant, pretextual way, rich in content and meaning—a way

that can frighten unreflective literate and computer people
disposed to associate recall with verbatim retrieval, which is
to say with itemization rather that with truth.
With the new and radically analytic ways of thought that are
set up by writing and print and electronics, and that revamp
the field in which memory had moved, human intellectual and
verbal activity enters radically new phases, which actively
develop over the generations as writing and its sequels are
more and more interiorized. For interiorization is a lengthy
and complex process. We are only now beginning to learn
what it has really meant.
NOTE

This paper is an adaptation of a section of a forthcoming

book, Orality and Literacy, by Walter J. Ong, to be published
by Methuen (London and New York) in their New Accents
series. Copyright 1981, Walter J . Ong.
REFERENCES
Abrahams, Roger D. 1968. Introductory remarks to a rhetori-
cal theory of folklore. Journal of American Folklore 81.143-
158.
Abrahams, Roger D. 1972. The training of the man of words
in talking sweet. Language in Society 1.15-29.
Biebuyck, Daniel, and Kahombo C. Mateene, eds. 1969. The
Mwindo epic from the Banyanga. Berkeley: University of
California Press.
Brink, Cfharles] O[scar]. 1971. Horace on poetry: The
'Ars Poetica.' Cambridge, England: Cambridge University
Press.
Havelock, Eric A. 1963. Preface to Plato. Cambridge, Mass.:
Belknap Press of Harvard University Press.
Havelock, Eric A. 1979. The ancient art of oral poetry.
Philosophy and Rhetoric 12.187-202. [A review article treat-
ing Peabody (1975).]
Kelber, Werner H. 1979. Markus und die Miindliche Tradi-
tion. Linguistica Biblica 45.5-58.
Kelber, Werner H. 1980. Mark and oral tradition. Semeia
16.7-55.
Lord, Albert B. 1960. The singer of tales. Harvard Studies
in Comparative Literature 24. Cambridge, Mass.: Harvard
University Press.
Ong, Walter J. 1967. The presence of the word. New
Haven and London: Yale University Press.
Ong, Walter J. 1971. Rhetoric, romance, and technology.
Ithaca and London: Cornell University Press.
Ong, Walter J. 1977. Interfaces of the word. Ithaca and
London: Cornell University Press.
24 / Walter J . O n g , S.J.

Ong, Walter J. 1978. Literacy and orality in our times.

ADE Bulletin No. 58, September, 1-7.
Peabody, Berkley. 1975. The winged word: A study in the
technique of Ancient Greek oral composition as seen princi-
pally through Hesiod's works and days. Albany: State
University of New York Press.
Sherzer, Joel. 1980. Tellings, retellings, and tellings within
tellings: The structuring and organization of narrative in
Cuna Indian discourse. Paper presented at the Centro
Internazionale de Semiotica e Linguistica, Urbino, Italy,
July [to be published in the Proceedings].
Sherzer, Joel. 1981. The interplay of structure and function
in Kuna narrative, or, How to grab a snake in the Darien.
[This volume, 307-323.]
PERSUASIVE DISCOURSE
AND ORDINARY CONVERSATION,
WITH EXAMPLES FROM ADVERTISING

Robin Tolmach Lakoff

University of California, Berkeley

The very work that engaged him [in an advertising

agency] . . . wafted him into a sphere of dim platonic arche-
types, bearing a scarcely recognizable relationship to any-
thing in the living world. Here those strange entities, the
Thrifty Housewife, the Man of Discrimination, the Keen
Buyer and the Good Judge, for ever young, for ever hand-
some, for ever virtuous, economical and inquisitive, moved
to and fro upon their complicated orbits, comparing prices
and values, making tests of purity, asking indiscreet ques-
tions about each other's ailments, household expenses, bed-
springs, shaving cream, diet, laundry, work and boots,
perpetually spending to save and saving to spend, cutting
out coupons and collecting cartons, surprising husbands
with margarine and wives with patent washers and vacuum
cleaners, occupied from morning to night in washing, cook-
ing, dusting, filing, saving their children from germs, their
complexions from wind and weather, their teeth from decay
and their stomachs from indigestion, and yet adding so
many hours to the day by labour-saving appliances that
they had always leisure for visiting the talkies, sprawling
on the beach to picnic upon Potted Meats and Tinned Fruit,
and (when adorned by So-and-so's Silks, Blank's Gloves,
Dash's Footwear, Whatnot's Weatherproof Complexion Cream
and Thingummy's Beautifying Shampoos), even attending
Ranelagh, Cowes, and Grand Stand at Ascot, Monte Carlo
and the Queen's Drawing-Rooms.
—Dorothy L. Sayers, Murder Must Advertise

25
26 / Robin Tolmach Lakoff

In the field of discourse analysis, much attention has been

focused on certain forms of discourse, much less on others. In
particular, scholars have been concerned with ordinary conver-
sation, on the one hand, and written expository text, on the
other. While the treatment of these types as separate entities
has certainly taught us a great deal about the characteristics
of each of them individually, and something about the nature
of discourse in general, it leaves a great many questions un-
explored. For one thing, how do they relate to other types of
discourse--which may superficially or even more deeply resemble
one or the other? What are the universal characteristics of all
types of discourse, and what characteristics are specific to just
one or two? Can we devise a taxonomy of discourse types, a
means of unambiguously differentiating among them?
Classifying the basic forms of discourse in terms of their
differing and similar characteristics seems rather less glamorous
than writing a grammar, a system of rules, for one or all dis-
course types. But in fact, it is impossible to write a grammar
without knowing the basic units involved: grammar consists of
instructions for the combinations of these basic elements.
It would also be profitable to look at discourse not from its
surface form (as conversation, say, or literary text) but more
deeply, with interest in its deeper purpose. When we look at
a range of discourse types, we notice that they appear quite
different from one another. Clearly, this disparity is due to
differences in what each is intended to accomplish, so that a
successful performance of a type of discourse is one that ac-
complishes what the speaker, or the participants as a whole,
had set out to do, rather than merely one that conforms to
some particular surface configuration.
This paper is somewhat experimental in nature, as I want to
address some of these questions, and see how far we can get
toward at least preliminary answers. As a start, I want to con-
sider one possible distinction among discourse types in terms of
function or purpose: ordinary conversation, on the one hand,
and something we can call 'persuasive discourse' (PD), on the
other. I will come to the problem of the definition of the latter
shortly. I am not suggesting that this is the only distinction
one can make among discourse types, nor that it is necessarily
the major one. Certainly, we can divide up the spectrum in
many ways, all intersecting: oral/written, formal /informal,
spontaneous/nonspontaneous--just to list some possibilities. I
am phrasing the question as if we are to inspect and perhaps
justify a dichotomy, but we should keep in mind the possibility
that no dichotomy will emerge--that we cannot divide discourse
neatly into persuasive/nonpersuasive realms, that some types
may cut across this distinction and it may prove irrelevant for
others.
With all these caveats in mind, however, it still seems useful
to begin as I proposed, since persuasion is a function
 Persuasive Discourse and Ordinary Conversation / 27

attributable to at least some discourse types. I would suggest

at the start that ordinary conversation is not persuasive in the
sense of having persuasion as its major goal. That is not to
say that in ordinary conversation (OC) we do not persuade, or
try to persuade, other participants. But persuasion is not
what we enter into the conversational experience for. We do
not come away from an informal chat saying, 'Wow, that was a
great talk! I persuaded Harry that bats eat cats!' Rather, an
experience of OC is good if we come away feeling that a good
interaction has been had by all, that we all like each other and
wouldn't mind talking to each other again. Granted that getting
these ideas across is in a sense persuasive, it is not so in the
sense that getting someone to worry about ring-around-the-collar
is persuasive.
One important determinant of technically persuasive discourse
is nonreciprocity: discourse is defined as reciprocal only in
case both, or all, participants in it are able to do the same
things, and if similar contributions are always understood
similarly. A classroom lecture obviously is nonreciprocal: one
participant selects the topics, does, most of the speaking, and
determines the start and finish of the discourse. The power in
such discourse is held by the person holding the floor, at least
to the extent that that person makes most of the explicit deci-
sions as to the direction the discourse takes, its start and
finish. On the other hand, it can be argued that the audience
holds power in such a situation, for it can go or stay, be
attentive or not--and by these decisions negate the effect and
purpose of the other's speaking. I return later to the question
of what constitutes power in a discourse, but here it can be
noted that it is meaningful to raise this question only for non-
reciprocal discourse.
Discourse that is truly reciprocal i s , at the same time, neces-
sarily egalitarian, at least ostensibly. Ordinary conversation,
for example, is normally fully reciprocal: any participant has
the same conversational options as any others, and if one can
ask a question and expect an answer, so can the others; if one
can ask a particular type of question, or make a certain sort of
statement (say, a question as to the other's financial affairs; a
statement about the other's personal appearance), the other has
the same privilege in turn, and if one can refuse to answer, so
can the other. Violations of this principle do, of course, occur
in OC, but when they do, participants feel a rule has been vio-
lated, that the conversation is making them uncomfortable, while
nonreciprocity in a lecture is expected and reasonably comfor-
table.
An example of an intermediate, hence problematic, case is
psychotherapeutic discourse. In many of its forms, there is
the appearance of an egalitarian, reciprocal conversation, but
in terms of deeper intention, the reciprocity turns out to be
only superficial. The therapist can ask questions which the
client soon learns not to ask; and if the latter should attempt
28 / Robin Tolmach Lakoff

to ask such a question, the therapist, rather than give an

answer, will usually treat the question as a tacit invitation to
ask another question, or make an interpretation: 'I notice
you're curious about my personal life'. Further, many of the
marks of power that belong to the lecturer in the classroom also
belong to the therapist: the decision when to begin and end,
and--while the client ostensibly picks the topics of discourse--
the determination of what the client's contributions actually
mean. The client, however, typically holds the floor for the
major part of the discourse--an anomaly in terms of the relation
between floor-holding and power-holding in typical conversa-
tional settings, which makes it especially difficult to acquire
proficiency in therapeutic dialog.
With these assumptions in mind, one can attempt a definition
of persuasive discourse as a type of discourse that nonrecipro-
cally attempts to effect persuasion. Discourse, then, is to be
considered persuasive only in case it is nonreciprocal, and the
intent to persuade is recognized explicitly as such by at least
one party to the discourse. By 'persuasion' I mean the attempt
or intention of one participant to change the behavior, feelings,
intentions or viewpoint of another by communicative means. The
last is important. Communicative means may be linguistic or
npnlinguistic (say, gestures), but they are abstract and sym-
bolic. A gun held to the head may indeed induce a change in
someone's behavior, but it is not communicative in this sense.
Hence I do not consider a direct physical threat a type of per-
suasive discourse. Types such as advertising, propaganda,
political rhetoric and religious sermons clearly do fall into this
category. While lectures, psychotherapy, and literature might
belong here under some interpretations, they are problematic
and are dealt with later: they seem intermediate between PD
and OC. On the other side, while direct physical intervention
is clearly outside of our definition of PD, brainwashing is more
difficult to assign to either category. It is true that very often
there is no direct physical force involved. But in brainwashing,
as the term is ordinarily used, there almost necessarily are
physical interventions--whether isolation from other people and
familiar surroundings, privations of numerous kinds, physical
discomfort and torture--so that, perhaps, on the grounds that
brainwashing is rooted, however indirectly, in nonsymbolic
physical means of motivation, we ought not to consider brain-
washing among the types of persuasive discourse, although it
is a demonstrably effective means of persuasion.
As we attempt to make more precise our definition of what is
persuasive and what is not, we are confronted with a problem
that has, itself, propagandists overtones. Perhaps because we
live in a society in which egalitarianism is upheld as a paramount
virtue, we extol anything that has the appearance of equality,
distrust anything that does not. We are not apt to ask whether
there are certain kinds of activities and situations in which
equality is unnecessary or even impossible, and find it hard to
 Persuasive Discourse and Ordinary Conversation / 29

imagine using phrases like 'power imbalance', 'inequality 1 , or

'nonreciprocity' without negative connotations. Hence, to talk
about discourse as 'reciprocal' implies that it is somehow good,
or beneficent; to call something 'nonegalitarian' is to imply that
those who customarily utilize it are manipulative and hungry for
control. (This may, of course, be true, and certainly there
are situations where a position of power is misused or abused.
It is further true that using the surface appearance of egali-
tarian and reciprocal discourse for deeper nonreciprocal and
power-seeking purposes is illegitimate and deserves censure,
unless justification can be given for this deceptiveness. But
discourse that is overtly and explicitly nonegalitarian seems not
to present any danger, nor to deserve the opprobrium heaped
upon it so often.)
Additionally, for numerous reasons, there are certain cultural
preferences in discourse types. Some we are prone to admire
and respect; others are illegitimate, 'dirty', debauched. We
would prefer to keep our likes and dislikes in neat, overlapping
piles: what we like for one reason should be admirable on all
grounds, and vice versa. Hence, if we have been trained to
despise one type of discourse--say, commercials or political
propaganda--we would like to believe that it is 'persuasive' be-
cause we consider persuasive discourse, since it is nonrecipro-
cal, malign; and if a type of discourse i s , at least to its practi-
tioners, beneficent and pure--for example, psychotherapeutic
discourse--there is tremendous pressure to deny that it is
'persuasive', that there is anything nonreciprocal about it, or
that there is any sort of power imbalance involved, for believ-
ing these claims would seem, to its proponents, to vitiate the
claims for benignness for the discourse type in question.
Uncertainty and conflict arise, of course, when we simul-
taneously judge a type of behavior good or bad, depending on
which aspects of it we focus on; but we have to dispense with
the idea that the attribution of power imbalance and nonreci-
procity is name-calling. It is not; it is mere definition and
should be considered value-free, with the added assumption
that some discourse types must, to be effective, be non-
reciprocal and power-unbalanced, the only issue in this kind
of word being whether it is effective, not the value of the
method by which that effect is achieved. We must try not to
heap obloquy on the commercial, and praise on the therapeutic
discourse, because of their purposes, at least not while we are
trying to discover their properties and their positions within a
taxonomy, and eventually a grammar, of discourse.
With this problem out of the way, let us turn to the question
of definition. Some of the factors have already been alluded to
at greater or lesser length in this paper, but I summarize here
aU the relevant points as they appear to me, in classifying dis-
course types, determining how they differ and what aspects are
universal, and differentiating between persuasive and nonper-
suasive discourse.
30 / Robin Tolmach Lakoff

First, and perhaps most important, is reciprocity, about which

much has been said earlier in this paper. Connected with reci-
procity is bilaterality. A discourse may be reciprocal and bi-
lateral, like OC; nonreciprocal and unilateral, in that true par-
ticipation occurs on only one side, like a classroom lecture
(though I make amendments even to this statement shortly); or,
most complex, nonreciprocal but bilateral, like psychotherapeu-
tic discourse, where both parties most typically can make true
contributions to the conversations, but the contributions may be
of different surface forms, and certainly are open to different
interpretations. Turn-taking is a natural concomitant of reci-
procity, though (as in psychotherapeutic discourse which in-
volves turn-taking) the two can be separated; but a non-turn-
taking, or unilateral, discourse can never be reciprocal. 1
Discourse may be spontaneous or not, or rather, can be con-
ventionally spontaneous or not. Thus, OC is at least conven-
tionally spontaneous: we distrust apparent OC if we have
reason to suspect any of the participants is working from a
script, or has planned significantly in advance. But a work of
art, or a lecture, is not supposed to be spontaneous, and the
lecturer feels no embarrassment about referring to notes.
Hence, we find differences in the use of hesitation devices,
pragmatic particles, cohesion, and so forth. A spate of
y'knows distresses us far less in OC than in a lecture--and
an ordinary conversation style without hesitations and other
devices reflective of spontaneity would make most of us very
uncomfortable. Spontaneity is, of course, related to reciprocity
and bilaterality: it is almost impossible to plan the flow of your
conversation when another participant has as much right as you
to determine its direction.
Another characteristic of persuasive discourse is novelty.
Ordinary conversation thrives on ritual and custom: while the
topics of our conversation, and the precise way we talk about
them, differ from time to time, our overall style does not shift,
nor in general does the way in which a given society holds an
informal conversation change over time. The general mode of
conversation today is not, at least judging from novels and
other contemporary evidence, significantly different from the
way it was done 200 years ago. Openings and closings--the
most ritualized elements--have changed very little over time:
while colloquially we introduce new forms of these occasionally--
Hiya, ciao, and so on--we eventually return to the old standbys,
hello and goodbye. The rest of the conversation follows a style
that can best be described as unstylized: it has no set pattern,
and hence no new patterns can be substituted for the old. Per-
suasive discourse, in all forms, is different: a defining feature
of persuasive discourse is its quest for novelty. This is mani-
fested on the lexical level, in the form of slogans and neolo-
gisms; syntactically; semantically, in that new concepts are
continually being introduced and talked about; and pragmatically,
in the way in which PD addresses hearers, its register, its
 Persuasive Discourse and Ordinary Conversation / 31

directness or indirectness* and many other factors. What is

crucial here is that PD wears out; OC does not.
A feature common to most forms of PD is that there is an
audience, rather than an addressee. Actually, audiencehood
goes along with unilaterality: an audience is a hearer or group
of hearers who play only that role, and do not take the active
role of speaker. The role of an audience is much more passive
than that of an addressee. In some forms of discourse, we
find both, at least ostensibly: in dramatic performances in
any medium, we often find conversations--involving speaker
and addressee in ordinary conversation--taking place in the
hearing of an audience, which does not participate. But much
of the dialogue uttered by the participants in the drama itself
differs in striking ways from true ordinary spontaneous conver-
sation. Some of the difference, of course, has to do with the
fact that the dialogue is constructed rather than spontaneous,
so that many of the uses of spontaneity are absent. 2 But
other differences are directly due to the fact that the dialogue
is occurring, in such situations, for the benefit not of the im-
mediate participants, but of the audience, and therefore many
of the contributions found in OC designed to make the other
feel good, or inform the other about necessary facts, are ab-
sent, and other things are present which would ordinarily not
be found in OC because all participants are already aware of
them and the contribution is therefore redundant. But the
audience is not aware, and needs to be apprised of the infor-
mation, and so it is. An audience has a different role than
does an eavesdropper in true OC.
In addition to the major distinction of audience/addressee,
there are distinct types of audience, based on the role the
latter is to play in the discourse. The role of audience ranges
from totally nonparticipatory to a conscious and active involve-
ment. Indeed, an audience can exist even in case the speaker
is not aware of its presence--an eavesdropper, that is, whose
role is precisely to remain undiscovered and therefore to give
no clues whatsoever as to its presence. In ceremonial functions
--at a wedding, for instance--the audience is known to be pres-
ent, and indeed its presence as witness to the ritual is neces-
sary if the ritual is to be transacted successfully, but there its
role ends. It does not participate, it is not expected to under-
stand, or signal its understanding. Typically, the audience at
a wedding does not indicate by any means, verbal or otherwise,
its agreement or complicity. In fact, it does not matter to the
members of such an audience if the ceremony is conducted in a
language they do not understand (say, Hebrew at an Orthodox
Jewish wedding). Their role is simply to be present, not to
derive anything themselves from what they hear. On the other
hand, an audience at a classroom lecture has as part of its
function to understand and, perhaps, give approbation--in the
form of evaluations later, yes, but more importantly, immediately
in the form of nodding and other nonverbal backchannels. A
32 / Robin Tolmach Lakoff

good lecturer, however large the audience, devises ways of dis-

covering and utilizing this nonverbal response, and without it--
say, speaking to a television audience--someone accustomed to
lecturing in the presence of an audience is lost. Hence, we
often find television discourse being taped in the presence of
a live audience, to provide speakers with this all-important,
though nonverbal and unconscious, feedback.
A third level of audience response is seen in certain forms of
religious and political gatherings: the audience is expected to
participate with audible and explicit backchannels: 'amen/right
on!' or applause and cheers. The audience here is very im-
portant to the speaker: it signals by its response whether
persuasion is occurring in these maximally persuasive forms of
discourse.
Related to the foregoing, but perceptible at a more abstract
level of analysis, is the use of the Gricean conversational
maxims. The maxims themselves are problematic for conversa-
tional analysis in that they were formulated by Grice (1975) for
quite different purposes than providing an understanding of
how discourse functions. Actually, in OC Grice's maxims are
seldom encountered directly. Rather, they are understood via
rules of implicature, and any ordinary conversations that ad-
hered for any length of time to the maxims themselves would
certainly strike participants as strange, oddly impersonal, often
literally unintelligible. By contrast, a lecture is expected to
adhere pretty closely to the maxims, and when implicature is
utilized it is for special purposes--irony, for instance—which
the audience is expected to appreciate as special. Indeed,
Grice's Conversational Logic, as he set it out, is optimally
applicable to the lecture, not the ordinary conversation. When
we attempt to extend the notion to other forms of discourse,
we find still more complicated difficulties, as I show in greater
detail later.
The presence of a power relationship among participants was
noted earlier. Important in this regard is the question of how
power is determined. It has been suggested that conversational
power belongs to the one who controls topic and floor, but
there is an alternative view that power is in the hands of who-
ever has the choice of whether to continue to participate,
whether to be persuaded. The hearer* at a lecture may listen
attentively or may whisper and shuffle; or they might even
leave, individually or en masse, and then the lecture truly
ceases to exist. Psychotherapeutic clients may or may not con-
tinue to appear for their sessions. The audience for the com-
mercials may buy the sponsor's product or not. Still, while
this may in some general sense be power, we can distinguish
from it power within the discourse; the power to motivate the
discourse in a certain direction, to begin or terminate it ex-
plicitly. And this power rests with the floor-holder (or with
the therapist in the case of psychotherapeutic discourse). As
with the issue of persuasion generally, we tend to consider
 Persuasive Discourse and Ordinary Conversation / 33

power an evil, and its appearance in discourse a sign of the

corruptness of the discourse. But in fact, as long as it is
explicitly acknowledged that the imbalance exists, there is no
problem.
Finally, as a distinguishing feature among discourse types,
we can think about the means of persuasion. Again, this is an
area heavily laden with value judgment, which is not helpful for
our present purposes. We can make distinctions among the
means of persuasion, in those types of discourse that fall into
the range that we have called truly 'persuasive': propaganda,
advertising, and political rhetoric, and those that have strong
persuasive elements: lectures, psychotherapeutic discourse,
and literature. In fact, it can be argued that what distin-
guishes the first category from the second is precisely the
means--ostensible in any case--of persuasion. The first oper-
ates by appeal to the emotions, the second--largely, or at least
theoretically--by appeal to the intellect. We tend to assign a
desirable connotation to intellectual persuasion, since it appears
to treat us with respect, take our most crucial human values
into account, give us a real chance to weigh and judge, and so
on; while emotional appeals seem to circumvent our reason and
to appeal to our base nature, giving us no chance to make a
real decision. In particular, within the last several decades,
'subliminal' advertising designed to present a tachistoscopic
image to the mind so that conscious perception is circumvented,
leaving only the appeal to the unconscious, by definition moti-
vated only by appeals to emotion and instinct, has aroused
particular outrage and inspired a good deal of preventive legis-
lation, despite more recent evidence that it is less effective
than its proponents hope and others fear.
I discuss further on the use of Gricean maxims in persuasion.
It is important here to note that the Rules of Conversation are
perhaps deceptively applied, or nonapplied; but they are cer-
tainly invoked, because they constitute, for the audience, evi-
dence that we are being persuaded by reason, intellectual argu-
mentation. Hence the appearance of conformity to the Gricean
maxims is critical if we as newly sophisticated consumers are to
be subliminally seduced. The appearance of reason conceals the
appeal to emotions and justifies it for the buyer.
Indeed, all forms of persuasive discourse have gone through
changes over time--partly necessitated by the requirement of
novelty, but also in part by the increasing sophistication of the
consumer and the need to present products as reasonable ones
to purchase. The change in the realm of advertising is especi-
ally interesting. In older advertising (say, up till the 1930s,
roughly speaking), there was a heavy preponderance of print,
of words, and far fewer and less striking pictures. This was
in part due to the state of reproductive techniques: photog-
raphy was rather primitive, color reproduction ruinously expen-
sive, and graphic techniques relatively unsophisticated so that,
for the money, words were the most economical means of
34 / Robin Tolmach Lakoff

persuasion. But the words, if we examine them, were far more

directly an appeal to the emotions than is wording--or, perhaps,
even illustrations--in advertising today. Part of the change is
due, of course, to increasingly vigilant regulation by federal
agencies concerned with truth in advertising. But ironically--
and not surprisingly, given our bias toward seeing ourselves
as logical and rational--regulations have almost invariably been
framed in terms of the wording used in advertising, at most
encompassing explicit illustration: marbles in the soup, for in-
stance. You cannot, in an ad, say in words, 'Glotz Detergent
will make your marriage happy': the Fair Trade Commission
(FTC) will come down hard on you. But you can say in words,
'Glotz gets your husband's shirts their whitest', alongside a
picture of a young, vital couple glowing at each other, sur-
rounded by cheerful children, a dog, a white picket fence.
The irony is that, in fact, the second approach is far more ef-
fective than the first, since it circumvents intellectual judgment
('How can using a detergent make Harry more responsive to
me?') and goes directly to the realm of the unconscious, capable
only of desires, fears, and needs.
If much of this discussion has the ring of a psychology text-
book, many modern advertising techniques trace their genealogy
directly to the far-famed couch in Vienna. Indeed, much of
modern motivational psychology, the basis of advertising, de-
rives from Freud, directly or indirectly. For it was Freud
(1900/1953) who pointed out the basic distinction between the
processes of the conscious and the unconscious mind. The un-
conscious works by the laws of the primary process; the con-
scious, by the secondary process. Primary process thought is
preverbal--symbolic, nonsequential, and visual--while secondary
process thought, more directly 'rational', is auditory and
verbal. Hence, if one wants to persuade by circumventing the
processes of rational thought, it makes good sense to emphasize
abstract images--music and pictures, for instance, rather than
words in logical sequence. And if the FTC is concerned with
deception in advertising, it would do better to pay attention to
the nonverbal means of persuasion which can be much more de-
ceptive. But the problem i s , of course, that it is difficult for
the investigator to prove what the picture of the happy family,
or the vital adolescents guzzling Coke, is communicating, since
we think of communication in terms of logical symbols--words.
Edward L. Bernays (perhaps not coincidentally Freud's
nephew), the inventor of modern public relations and hence of
many of the modern forms of persuasion, spoke of 'the engineer-
ing of consent' (1952). It is this that upsets us as consumers
and intrigues us as investigators. If we are so suspicious of
persuasion and its techniques, how are we 'engineered' to give
consent? Bernays referred not only to the relatively harmless
influence of advertising, but also to the more baneful effects of
political rhetoric, including propaganda. The latter term itself
can be given some scrutiny.
 Persuasive Discourse and Ordinary Conversation / 35

Originating in a religious context, with positive connotations

(the 'propagating' of the faith), the term 'propaganda' eventu-
ally came to mean, viewed from the perspective of another re-
ligion, a form of improper influence or pressure. Hence 'propa-
ganda' today is exclusively a pejorative term (though Bernays
makes a plea for its rehabilitation). 3 One question is whether
the term really has true denotative meaning, as a special kind
of persuasion identifiable even when the argument is one with
which we are in sympathy; or whether 'propaganda' simply
means emotional persuasion, when the argument is not one we
approve of. We would agree that 'propaganda' is mainly appli-
cable to forms of persuasiveness that utilize emotion, usually of
a high intensity, often invoking fear and irrational desires.
But advertising typically does this, yet is not considered
'propaganda'. So we might want to add the proviso that propa-
ganda concerns changes in beliefs, rather than concrete buying
habits alone. Yet we might argue that much advertising is
propagandistic--not only the 'public service' advertising and
institutional advertising by power companies, but even adver-
tisements for soft drinks that suggest indirectly that youth
and a svelte body are minimal conditions for being allowed to
exist, or that create and reinforce all sorts of traditional sexual
and other stereotypes. This then suggests another facet of
propaganda: it is normally indirect. It is not present in the
explicit message, but somewhere in the presuppositions. When
we are serenaded to the effect that 'Coke adds life', the propa-
ganda is not in the admonition to buy Coke, but in the inference
that youth is desirable (and young people drink Coke).
This brings us back a bit roundaboutly to the relation be-
tween the Gricean maxims and discourse types. I have argued
that the Cooperative Principle (CP) is more directly applicable
to the classroom lecture than to any other type of discourse--
certainly more than to ordinary conversation. But even OC
makes reference to the maxims, if only via implicature. And in
OC, we understand that flouting of the maxims is due to our
desire to adhere to more socially (as opposed to intellectually)
relevant rules, rules of Rapport: when we have a choice be-
tween being offensive and being unclear, we invariably choose
the latter, and a majority of cases of OC implicature can be
seen to stem from this assumption.
But in persuasive discourse, the situation is rather different.
Whereas in OC and the lecture, our aim is to inform--at least,
it is so, other things being equal--in PD our aim i s , of course,
to persuade. The politician does not especially want a
knowledgeable electorate: he wants votes. The advertiser is
not interested in educating people about hygiene: he wants to
sell deodorant. It is not that the maxims are violated only in
case a more peremptory need intervenes, as in OC--they are
regularly infringed without explanation, cue, or apology.
Indeed, it might be argued that the PD need for novelty alone
is responsible for many instances of violations of the maxims,
36 / Robin Tolmach Lakoff

especially in advertising. For in OC, we have seen, novelty

is not especially valued. Part of the reason for this is that
familiarity aids intelligibility--that is, aids in the keeping of
the Cooperative Principle. What is new and requires interpre-
tation is in violation of the maxim of Manner (be clear). But
in PD it is this very violation that is striking, memorable--
efficacious.
If we understand this preference for novelty as an intrinsic
aim of PD, we can perhaps understand better why advertisers
cling to certain formulas despite--or rather, because of--the
contempt to which they are subjected by critics. For example,
many of us remember a slogan from the fifties: 'Winston tastes
good like a cigarette should'. The criticism heaped on the hap-
less preposition was staggering, yet the commercial continued to
appear in that form. (Lip service was later paid to linguistic
chastity by the addition, after the infamous slogan, of the re-
joinder, 'What do you want, good grammar or good taste?' which
was cold comfort for traditional grammarians.) We can under-
stand the company's clinging to the solecism if we understand it
as a Manner violation: focusing the hearer's (or reader's)
attention on style tended to obscure content, and thus to flout
Manner, if a bit indirectly. Indeed, anything neologistic will
have much the same effect, and will serve as good persuasion
for two reasons: first, because it violates Manner as just ex-
plained, and thus attracts the audience's attention; and second,
in the case of obvious neologism--as opposed to the simple un-
grammaticality of like for as- -it forces the audience to interpret
--as any violation of the Cooperative Principle does.
It is axiomatic among proponents of all forms of persuasive or
semipersuasive discourse (e.g. psychotherapeutic discourse and
the classroom lecture) that if the audience can be made to par-
ticipate at some level--that is, to function as addressee, not
wholly as audience--learning, or persuasion, will be much more
successful. Hence, if the audience is forced to interpret neo-
logism, or relate 'ungrammatical' forms to their textbook version,
they will probably remember better. And memory is the name
of the game in PD.
We find violation of the Cooperative Principle not only in Man-
ner, but in other maxims as well. In terms of neologisms and
other sorts of linguistic novelty--most easily considered as
Manner violations--we find:
(1) Lexical novelty (neologism):
stroft, a portmanteau of strong and soft; devilicious.
(2) Morphological/Syntactic novelty (in terms of category
shifts: like for as may belong here):
Gentles the smoke and makes it mild
Travels the smoke further
 Persuasive Discourse and Ordinary Conversation / 37

The soup that eats like a meal

Peanuttiest
(3) Syntactic innovation:
(Some of these can also be viewed as Quantity violations,
in that either not enough or too much information is given
for ordinary conversation understanding. These include
such odd usages as the following.)
(3a) Absence of subjects and often the absence of verbal
auxiliaries as well: **
Tastes good! And nutritious too!
(3b) Odd uses of the definite article, which is sometimes un-
expectedly inserted and is sometimes unaccountably ab-
sent:
Next time, I'll buy the Tylenol!
Baby stays dry! Diaper keeps moisture away from
baby's skin!
(4) Semantic anomaly (other than lexical anomaly):
(These include quantity violations, among other things.)
Cleans better than another leading oven cleaner.
Works better than a leading detergent.
In instance (4) we expect, because of Quantity (or perhaps
Relevance), a definite article in the item being compared with
the product being touted. Otherwise, the information is non-
informative, or useless: after all, what we need to know is
which is best. Here is a good example of persuasive discourse
adopting the surface trappings of informative (i.e. Cooperative
Principle obeying) discourse and thereby leading us to conclude
that it is informative (if it looks like a duck, and walks like a
duck . . . ) . Hence we understand it as informative, that is, in
keeping with the CP. Notice that when a maxim is infringed in
OC via implicature, we do not interpret the utterance as being
in keeping with the maxim by virtue of its surface appearance.
Rather, in these cases we are given signals by the infringing
speaker that the contribution is in violation of the CP, and we
are thus implicitly directed to put our interpretive skills to
work. In PD, on the other hand, the flouting of the maxim is
covert, and we are tricked into assuming that an act of infor-
mation is taking place in cases like this, where in fact it is not.

(5) Pragmatic novelty.

This includes anything aberrant about the discourse form
itself. In particular, since many commercials are framed
as mini-dramas, we see many unusual bits of dialogue
within these 30-second segments. One such type much
in vogue lately is the following.
38 / Robin Tolmach Lakoff

—You still use Good Seasons Italian?

—Not any more!
--No?
--I use new improved Good Seasons Italian!
Now in terms of ordinary conversation, the second speaker's
contribution violates Quantity, and would probably be treated
not as the joke it functions as in the commercial, but as a
rather stupid bit of obfuscation, if it were to occur in real
OC. This is generally true of the 'jokes' highly favored in
recent commercial genres, especially coffee commercials:
--Fill it to the rim!
--With Brim! [laughter]
Wife [with camera]: Give me a smile!
Husband [at breakfast table]: 5 Only if you give me
another cup of your coffee.
Humor, in ordinary conversation, can often be viewed as a
permissible Manner violation. But the humor here is of a rather
different order, especially as it seems both strained and puerile.
This may in part even be purposeful: reassuring the audience
that the folks in the commercial are no different from them,
they make awkward jokes too, and this just shows they are
good, spontaneous people. That this seems especially charac-
teristic of coffee commercials, I think, is because the ambiance
of these commercials is intended to suggest a sort of easy-going
informality where joking of this kind is in place. But beyond
this, these jokes fulfill another communicative purpose: if we
assume that, unlike OC, the major purpose in commercials is
to ensure that the sponsor's name is remembered, rather than
creating an easy and natural atmosphere of warmth (but, in
many types of commercials, the summum bonum is getting the
sponsor's name memorably associated with an easy and natural
atmosphere of warmth--that is, superficial rapport), these
'jokes' see to it that the sponsor's name gets mentioned in a
prominent position, usually at the very end of the commercial,
and in a loud voice with stress--as we stress the punchline of
a joke more than the end of a normal sentence.
Another pragmatic anomaly is seen in unusual patterns of
intonation. For example (a commercial for Sunrise coffee):
I like it I'll drink it,
with no pause between the clauses. In fact, the stress of this
utterance is what we would expect if we had a syntactic de-
pendent clause at the left (e.g. Since I like it so much . . . ) .
The intonation here is steady rather than falling, as would be
normal for two syntactically independent sentences. I think the
only way we can make sense of this is as an attention-getting
novelty.
 Persuasive Discourse and Ordinary Conversation / 39

It seems that in the past, most advertisers focused their

attention on lexical and syntactic novelty, while more recently,
pragmatic novelty seems to be favored. What we may be seeing
here is metanovelty: the audience is jaded with the older forms
of newness, and advertisers must press ever onward into the
more mysterious reaches of language to get a response from us
at all.
Some of the novelties I have been talking about can be justi-
fied on additional grounds, beyond their usefulness as floutings
of some part of the Cooperative Principle. For example, many
of these work for brevity, a real desideratum when the point
must be made in 30 seconds. But I think that brevity alone
seldom justifies any special usage. Rather, all the special fea-
tures of PD are overdetermined.
I have illustrated the flouting of the Cooperative Principle
with regard to advertising, in which it is particularly glaring
and has, indeed, been exalted to an aesthetic feature of the
medium. But it can be found just as easily in other forms of
PD. Its function in psychotherapeutic discourse is special, and
is not dealt with here. But certainly we are accustomed to it
in political rhetoric--especially in the form of hyperbolic viola-
tions of Quantity, and metaphorical floutings of Manner. It is
noteworthy that, of all the forms of violations of CP that are
found and tolerated in PD, Quality alone is missing. Quality
violations are what bring the FTC down on the sponsor, im-
peachment threats down on the politician. The others seem be-
yond reproach. We may want to examine again the assumption
that Quality is on the same level as the other maxims, when in
fact we feel that abrogation of this maxim has very different
psychological and social implications. Indeed, violation of the
other maxims can always be justified for either aesthetic or
social reasons, without further ado; but we have only a sub-
category of Quality violations that are tolerable-- and only
dubiously so: the 'fib' or 'white lie'. In fact, Quality viola-
tions are the only ones ordinary language has Ga separate word
for--and a word with bad connotations at that.
I have referred to the importance of Rapport rules and strate-
gies in OC, and their superseding of the Cooperative Principle
when they conflict with it. Since PD is nonreciprocal and has
an audience, rather than participating addressees, it is not
surprising that Rapport as such plays no role. We do find, in
some types of commercials, attempts to establish a one-to-one
relationship with the audience on television: eye contact and
twinkling, casual register, and the like. But in fact, most of
the conventions of OC that exist for Rapport purposes are ab-
sent from commercials, and from most forms of PD as well--
openings and closings, for example, are either totally missing
or heavily truncated (as in psychotherapeutic discourse).
All this discussion raises a very troubling issue in discourse
analysis of the kind I am attempting here: what is the role of
the CP in persuasive discourse, and especially in advertising?
10 / Robin Tolmach Lakoff

I have spoken as if the CP were involved in our understanding

of commercials as it is in OC or the lecture--and yet we saw it
was utilized very differently in even these two types of dis-
course. Certainly the CP gives us a much needed and illumi-
nating handle on the workings of commercials and our under-
standing of them. But at the same time, invoking the CP as
the basis of our understanding of a discourse type makes an
implicit claim: since the CP enjoins us to be as informative as
possible, to say that our discourse is understood in relation to
it is to imply that the true underlying purpose of that dis-
course is to inform; and that, if information is not exchanged
in some utterance in an optimal way, there.is a special reason
for it--politeness, for instance, in OC. This assumption works
splendidly for discourse that is explicitly intended above all to
inform—that i s , of course, the lecture. But even with OC. it
gets us into trouble. For the main purpose of OC is not to in-
form, or even to exchange information. This can occur, and
conversations frequently are loaded with useful information, but
most often the information serves only as a sort of carrier, en-
abling the real business of the interaction to get done--inter-
acting. Hence Rapport supersedes the Cooperative Principle,
since Rapport is the point in OC. Then we might argue that
OC is not really CP-based, but Rules of Rapport-based, with
the CP a mere auxiliary. This is unsatisfactory, though, in
that we invoke the mechanisms of the CP to account for our
understanding of the utterance--implicature is, logically if not
psychologically speaking, secondary to the operations of the
Cooperative Principle.
The problem with persuasive discourse, however, is more
complex still. We are not intended to understand PD through
the use of implicature; there is no apparent reason that might
justify infractions, yet infractions are common and, as we have
seen, quite different from the sort we get with OC. Yet in PD
there is an appeal to our knowledge of the workings of the
Cooperative Principle. It is our very awareness of its being
violated, in such unexpected and inexplicable ways, that creates
the memorability and effectiveness of all forms of PD, especially
advertising. It is not clear, then, whether we should say that
PD, like OC, is predicated on a base of the CP, but merely is
expected to be in violation of it (it is not really clear what this
would mean), or whether a completely different basis must be
proposed. But if the latter, how do we account for our recog-
nition of the CP through its flagrant violation in PD? Or our
insistence on interpreting PD as if it were in accord with the
CP, although it is not (which, as we have seen, makes it per-
suasive)? I am not sure how to resolve this issue, and present
it here as something that will have to be determined by future
research, if we are to devise a taxonomy of discourse that ac-
counts not only for surface form, but for deeper intentions and
the relations between the two.
 Persuasive Discourse and Ordinary Conversation / 41

In any event, we have made a beginning, I think, and some

interesting facts as well as problems have emerged. We have
seen that there are valid bases on which to distinguish between
ordinary conversation and other types of discourse, and be-
tween truly persuasive discourse and intermediate types. I
have discussed the meaning of 'the engineering of consent',
and argued that it has to do with the manipulation of our ex-
pectations about the form and function of OC, translated into
PD: both surface form and deeper intention, in terms of the
Cooperative Principle, can be turned to persuasive effect. I
have argued that we must take a number of factors into account
in making these determinations: reciprocity and bilaterality;
spontaneity and novelty; power; the function of the audience/
addressee; the means of persuasion; and finally, the use of the
Cooperative Principle and the Maxims of Conversation.
In short, many discourse types which superficially look simi-
lar, at a deeper level of analysis function quite differently; and
many types which look different turn out to have deeper simi-
larities. We cannot understand discourse until our classification
includes, and compares, form and function together, and we can
hope to have a satisfactory grammar of discourse only when we
have arrived at a valid taxonomy. This brief discussion of one
parameter--persuasive and nonpersuasive discourse—is pre-
sented as a beginning.
NOTES
The quotation from Dorothy L. Sayers which appears at the
beginning of this paper is from her book Murder Must Advertise
(New York: Harper and Row, Publishers, Inc.).
The ideas in this paper owe much to discussion with many
others. In particular, Linda Coleman's (in press) work on ad-
vertising, as well as other forms of persuasive discourse, has
been stimulating. Students in Linguistics 153, Pragmatics
(Spring 1981), have been invaluable in refining my thoughts,
and their forbearance in serving as sounding-boards is grate-
fully acknowledged. Deborah Tannen's suggestions and inspi-
ration are similarly much appreciated.
1. An odd apparent counterexample occurred at the end of
the Phil Donahue show, June 16, 1981. At this point Donahue
turned to the camera (not the studio audience) and said: 'We're
glad you were with us. You were terrific!' It is hard to
imagine how the TV audience could have demonstrated its
terrificness.
2. More discussion of these points can be found in Lakoff
(in press).
3. Curiously, the Oxford English Dictionary has a definition
of propaganda that is wholly without negative connotations--
quite different, I think, from its normal use: 'Any association,
systematic scheme, or concerned movement for the propagation
of a particular doctrine or practice'.
12 / Robin Tolmach Lakoff

4. An added benefit of this form of utterance is that it

mimics a casual register and thus suggest folksy informality.
Comparison of these segments with true OC casual register
makes it clear, however, that the truncations found in com-
mercials do not occur in OC. One more advantage, of course:
a micro-millisecond saved is a tidy sum of money earned, given
current rates on prime time television.
5. The use of your here is typical of many kinds of commer-
cials (preceding the name of the sponsor's product, or its
generic category, as here) and is characteristic of the genre,
and not OC. Normal here is 'another cup of coffee', perhaps
'that coffee'. Your in this environment (a quantity violation in
that it provides unnecessary information) perhaps is intended
to suggest the addressee's responsibility for the adequacy of
the product.
6. And only with Quality do we find it necessary to differenti-
ate between purposeful ('lying') and accidental or neutral ('mis-
information', etc.) violations. For more discussion of the lin-
guistic and communicative problems about lying, see Coleman and
Kay (1981).
REFERENCES

Bernays, Edward L. 1952. Public relations. Norman, Okla.:

University of Oklahoma Press.
Coleman, Linda, (in press) Semantic and prosodic manipulation
in advertising. In: Information processing research in ad-
versiting. Edited by Richard J. Harris. Hillsdale, N . J . :
Erlbaum.
Coleman, Linda, and Paul Kay. 1981. Prototype semantics.
Lg. 57.1:26-44.
Freud, Sigmund. 1900/1953. The interpretation of dreams.
Vols. 4 and 5 of the Standard Edition of the Complete Psycho-
logical Works of Sigmund Freud. London: Hogarth Press.
Grice, H. Paul. 1975. Logic and conversation. In: Syntax
and semantics 3: Speech acts. Edited by Peter Cole and
Jerry L. Morgan. New York: Academic Press.
Lakoff, Robin Tolmach. (in press) Some of my favorite
writers are literate: The mingling of oral and literate
strategies in written communication. In: Spoken and written
language: Exploring orality and literacy. Edited by Deborah
Tannen. Norwood, N . J . : Ablex.
 MONEY TREE, LASAGNA BUSH, SALT AND PEPPER
SOCIAL CONSTRUCTION OF TOPICAL COHESION
IN A CONVERSATION AMONG ITALIAN-AMERICANS

Frederick Erickson
Michigan State University

Introduction. This is an analysis of a dinner table conver-

sation in an Italian-American family. Two issues are of special
interest: (1) how topics and topical items are cohesively tied
together in strips of discourse across turns at speaking and
(2) how this cohesion is maintained within and across multiple
conversational floors, as individuals talk simultaneously without
apparent interference, interruption, or other damage to intelli-
gibility or appropriateness.
Both topical cohesion and floor management are seen as practi-
cal problems of social organization in conversation. They are
work which must get done by the collective action of individuals
who are partners in listening and speaking relationships that
are enacted in real time.
The paper begins by reviewing a few key notions: 'conver-
sational work', 'social construction and production resources',
'floor', and 'topic'. This is followed by an introduction to the
transcript, describing the nature of the conversation as a social
occasion. An interpretive analysis follows the transcript itself.
The analysis attempts to identify some of the social interactional
grounds of topical cohesion and floor maintenance, to shed light
on the organization of a conversation in which such apparently
disparate topical items as a money tree, lasagna bush, and salt
and pepper could cohere and make sense.
The notion of 'conversational work' can be thought of both
as effort exerted toward a set of ends, and as the ends to
which the effort is exerted. The activity of conversation--
speaking and listening behavior--is thus seen as effort exerted
for purposes beyond itself. The organization of conversation
is constituted not simply by its own activity, but by the larger
purposes to which the activity of conversation contributes.
44 / Frederick Erickson
The portion of the dinner conversation to be considered here
involved talk that occurred near the end of the main course of
the meal, just before dessert was to be served. The dinner
partners were finishing up the instrumental work of eating.
Their talk can be seen as an expressive accompaniment to that
work. The talk did not accomplish discrete instrumental func-
tions, as in the promises, requests, and other 'speech acts'
that have been considered by speech act theorists and by lin-
guistic pragmaticists. Nor did talk in this conversation accom-
plish discrete expressive purposes as ends in themselves, as
in highly stylized displays of verbal art, e.g. toasts, ritual
insults, jokes. Unlike the 'speech events' that have been
studied in much of the literature of ethnography of speaking
(see the discussion in Sherzer 1977), speech itself did not
constitute the activity being undertaken.
The overall functions of talk before dessert seem to have in-
volved what Mali now ski termed 'phatic communion'--the conduct
of sociability for its own sake (Malinowski 1923:315). Even
this communicative purpose was not an end entirely accomplished
in the talk itself, since the food and its shared consumption was
another communication channel by which phatic communion was
being accomplished. But one can't talk fluently with one's
mouth full, as had been the case in the phase of the dinner in
which the primary focus was on food consumption. At the
point at which the transcript begins, there was a shift in
channel dominance from the gustatory work of eating and pass-
ing dishes to the conversational work of talking and allocating
turns at speaking. Despite the shift in channel dominance, the
overall event remained 'having dinner'. As we will see more
closely, the contents of the dinner table before the dishes were
cleared for dessert--the residue of unconsumed food, the condi-
ments, and the utensils for eating--provided an important re-
source for topical content.
Conversation as social construction. The work done in con-
versation is socially accomplished construction. The term 'social'
is meant here in a particular way: that of Weber's definition
of 'social action' as action that is taken in account of the actions
of others (Weber 1922:30):
A social relationship may be said to exist when several
people reciprocally adjust their behavior to each other with
respect to the meaning which they give to it, and when
this reciprocal adjustment determines the form which it
takes.
'Construction' involves making use of constraints provided by
the actions of others as structure points around which one's
own activity can be shaped. The constraints can thus be seen
as 'production resources' rather than as limitations or as deter-
ministic pressures.
 Social Construction of Topical Cohesion in a Conversation / 45

There are at least three types of production resources that

conversationalists can make use of: 'immediately local' re-
sources, 'local resources once removed' from the immediate
scene, and 'nonlocal' resources. Since the latter two types
are defined here in relation to the first, it is appropriate to
consider the notion of local production resources before con-
sidering the others.
Local production resources are those available within the
immediate physical setting and within the immediate action of
conversation as it occurs in real time. The term 'local produc-
tion' was developed by conversational analysts in sociology to
identify the socially adaptive character of talk as it naturally
occurs. Speakers are not seen as simply producing strings of
syntax, but as responding at one moment to what the self or
others did in the moment before, or prefiguring what will
come in the moment next. The adjacency relationships of 'next'
(e.g. answer slot as a reply to the adjacently prior question
slot) and the character of speech as addressed to a hearer
(termed 'recipient design 1 ) have been elegantly described by
conversational analysts (see Sacks, Schegloff, and Jefferson
1974). Their emphasis on sequential adjacency highlights the
reciprocal dimension of social organization in conversation;
individual actions are seen as taking account of the actions
of others back and forth across strategic moments of real time.
Another aspect of local production has been emphasized more
by researchers in the tradition of context analysis, in which
nonverbal and verbal aspects of interaction are studied together
(see Birdwhistell 1970; Scheflen 1973; Kendon 1977). Here it
is the complementary dimension of social organization that has
received the most attention, e.g. looking at the ways in which
the listening behavior of the listeners and the speaking be-
havior of the speakers, cooccurring synchronously, complete
each other's actions and thus mutually influence each other.
While conversational analysts have emphasized more the re-
ciprocal and sequential aspects of interaction, and context
analysts have emphasized more the complementary and simul-
taneous aspects of interaction, both see the social organization
of interaction as radically local. (Indeed, the reciprocity and
complementarity I have been describing can be thought of as
horizontal and vertical relationships of adjacency in real time).
Conversation is seen as a highly coordinated partnership, as
in the relations of immediate influence and regulation between
partners in a ballroom dance. Conversational partners are
seen as enabling and completing one another's actions in real
time performance through speaking and listening activity that
is both reciprocal and complementary.
From this perspective, an essential aspect of local production
is the continual activity of the partners in telling each other
what is going on in real time--what time it i s , what activity it
is now. This telling is done explicitly and implicitly, verbally
and nonverbally, by a host of surface structural means that
46 / Frederick Erickson

Gumperz (1977) calls contextualization cues. These cues are a

subset of the devices that Bateson (1972) called 'metacommuni-
cative'.
This is to take a special view of the relations between text
and context, one that is not usual among linguists. In con-
versation, text and context can be seen as mutually constitu-
tive rather than as dichotomous. To paraphrase McDermott
(1976:33), 'people in interaction form environments for each
other.'
Necessary as these local resources are, they are not the only
ones drawn on by conversationalists. As Goffman (1976) has
pointed out, conversation is organized not only in terms of
locally adjacent next relationships, but in terms of connections
across larger chunks of discourse, connections across the whole
history of the conversation.
These connections can extend across even longer strips of
time and across space, beyond the conversation itself. Influ-
ences on the immediate conversation can come from outside it,
across days, weeks, even years, as in the recurring meetings
of a committee, in long-standing family disputes ('Oh God,
you're not going to bring that up again!'), or in a military
attack, in which various subgroups begin an assault simul-
taneously, the officers having synchronized their watches. In
responding to what others have done outside the immediate en-
counter, conversationalists take action that is social by Weber's
definition, even though there is only an indirect connection be-
tween their actions inside the encounter and the actions of
other persons outside the encounter.
There is no current term for these indirect sources of influ-
ences on conversation which are still specific to the immediate
situation at hand (in that the influences have to do with the
particular biographies and shared history of the conversational
partners). Since the relationships of immediate adjacency in
conversational action have been called 'local' production re-
sources, we can call the influences that are not fully local--
but still group-member specific--'local production resources
once removed'.
This resource type can in turn be distinguished from a third
type: those nonlocal production resources that derive from the
wider social structure. Among these nonlocal resources are
knowledge of cultural traditions shared within a given speech
community that define appropriateness in ways of speaking,
knowledge that at the individual level Hymes and others have
called 'communicative competence' (Hymes 1974). Another non-
local resource is 'linguistic competence', knowledge of the gram-
mar and sound system of a language. While these nonlocal re-
sources are not specific to the members of the interacting group,
they influence the shape of collective action in the group through
the medium of the individual members' particular knowledge and
performance skills.
 Social Construction of Topical Cohesion in a Conversation / 17

In summary, three sources of social influence on the conduct

of conversation can be distinguished: the nonlocal, the local
once removed, and the immediately local. These can be thought
of as production resources. All three types of production re-
sources are employed in naturally occurring conversation. All
three are often employed simultaneously and need to be con-
sidered together in a holistic analysis of conversation.
Conversational floors and topical cohesion. Topics and the
speakers who produce them must have a floor to be in. Con-
versational floors are thus a local production resource for the
construction of topical cohesion. The 'floor' is a sustained
focus of cognitive, verbal, and nonverbal attention and re-
sponse between speaker and audience.
In maintaining a floor, audience and speaker interact socially
in the reciprocal and complementary ways discussed earlier.
Talk is addressed by recipient design to the audience, and the
audience responds to the talk. Without the ratification of
audience response, a speaker has nowhere to go with a topic
once it has been introduced. In consequence, as Keenan and
Schieffelin (1976) have pointed out, the notion of topic in con-
versation entails a relationship between speaker and audience.
It is possible for conversations to be socially organized with
more than one floor being sustained simultaneously by multiple
speakers and multiple audiences. The conversation studied in
this analysis is a case in point. For members of the Italian-
American speech community from which the conversational part-
ners came, multiple floored conversations are appropriate and
frequent occurrences.
In this type of conversational arrangement, a fundamental
organizational task for conversationalists involves determining
where the floors are, how many of them there are now (see the
discussion in Shultz, Florio, and Erickson in press). That task
is even more fundamental than the task of allocating access to
the floor through a turn exchange economy, an issue that has
been discussed by the conversational analysts (Sacks, Schegloff,
and Jefferson 1974).
'Topics' themselves involve social construction within the con-
versational floor. Once allocated, a turn at speaking must have
some topical content and some duration in real time. The real-
time nature of performance places practical constraints on
appropriateness of topic. These can be thought of as con-
straints on intelligibility and of fluency, which can also be con-
sidered as production resources.
Since the topic occurs in a floor, and floors are jointly main-
tained by speaker and audience participation, a topic must be
intelligible to the audience if the audience is to be able to do
its part in collaboration with the speaker. Because the topic
is spoken in real time, topical content must be immediately in-
telligible in the moment, since the flow of action cannot be
suspended to give the audience time out for extensive reflection.
48 / Frederick Erickson

Moreover, since floor time is occupied by topical content, it is

strategically necessary for speakers to produce topical content
fluently--continuously across real time—in order to be able to
maintain their position of participation in the floor.
The etymology of the term 'topic' points to the issue of flu-
ency as well as to the issue of intelligibility. As Ong (1977:
147ff. , 166) points out, the Classical Greek term for topic is
a spatial metaphor (topos 'place'). The notion of topic was
that of a location in semantic space. Knowledge of certain of
these 'places' was widely shared conventionally within a speech
community. Among Classical and Renaissance rhetoricians, the
term for these widely recognized topics was 'commonplaces' (see
the discussion in Lechner 1962). The orator memorized many of
these commonplaces: an assertion, together with constituent
topical items such as metaphors, anecdotes, citations of evi-
dence, quotes from literature, or previous oratory. As a cog-
nitive resource, the commonplace functioned as a memory stor-
age device, since retrieval of the general topical category
('place') entailed a connected set of specific items of information.
As a social interactional resource, the commonplace functioned
as a device enabling fluent production and comprehension. The
easily remembered topical items could be continuously spoken in
real time by the orator and were readily understandable in real
time by members of the audience.
While ordinary conversation differs from oratory in a number
of ways (ordinary conversation being characterized by short
turns at speaking and a relative lack of preplanning in dis-
course), people in ordinary conversation face the same funda-
mental practical problems faced by orator and audience--the
problems of achieving and maintaining fluently reciprocal and
complementary coordination of speech production and compre-
hension in real time. It is not surprising, then, that people
in ordinary conversation draw on some of the same production
resources as those found in formal rhetoric. In fact, the im-
provisatory character of ordinary conversation requires the
ready availability of commonplace topics as formulaic themes
around which conversational partners can improvise situational
variations.
As resources, the sources of commonplace topics in ordinary
conversation can be thought of as immediately local, local
once removed, and nonlocal. Immediately local topical resources
are those experiences currently being shared among the con-
versational partners, for example, the fact that it is raining
now, the food is on the table now. Local resources once re-
moved are topics derived from experiences previously shared
among the conversational partners, for example, what we did
on our summer vacation, what happened in last week's meet-
ing. Nonlocal topical resources involve knowledge of semantic
domains that is widely shared within the speech community,
for example, talk about sports, current national events, cars,
the arts, religion. Topics and their constituent topical items
 Social Construction of Topical Cohesion in a Conversation / 49

are used as commonplaces at each of these levels. Conversa-

tional partners have experience in how to organize informal
conversation around the topics; for example, they already know
how to talk about the immediately local topic of what the weather
is now, or the nonlocal topic of what kind of car one might or
might not want to buy.
The conversation. The dinner conversation discussed here
occurred in an Italian-American family living in a working class
neighborhood in a suburb of Boston. The conversation was
videotaped as part of a study of the communicative competence
of Bobby, the youngest boy in the family. His communication
was also studied at school in the first grade classroom he at-
tended (for discussions of the larger study, see Bremme 1977;
Florio 1978, Shultz and Florio 1979; and Shultz, Florio, and
Erickson in press).
Bobby's interaction with other family members was videotaped
on two days, one week apart, in early September, from the
time he came home from school through the end of dinner.
Other family members besides Bobby included his mother and
father, three older brothers, and a sister. Bobby was the
youngest. The portion of the dinner conversation presented
here comes from the second of the two dinners that were video-
taped. The night of the second dinner, Bobby's father was not
home; he had gone out of town on an overnight trip. The
guest, an Italian-American woman research assistant, sat in the
father's usual place at one end of the table (see diagram at the
top of the transcript). The mother sat in her usual place, at
the opposite end of the table from the guest. Some of the other
children in the family had left the table at the point in the meal
at which the transcript begins. The two remaining children,
7-year-old Bobby and his next older sibling, an 8-1/2-year-old
sister, sat facing one another on the opposite axis of the dinner
table from that along which the mother and guest were seated.
As the conversation progressed, the arrangement of social
participation changed. At first, the conversation was organ-
ized around a single conversational floor, with all four partners
orienting to it as a quartet of speakers. Later, the conversa-
tional duet or team—mother-guest along one axis of the table,
and brother-sister along the other axis. Then after a few min-
utes, the conversational arrangement returned to a single con-
versational floor. Thus the spatial arrangement of the partners
around the table was made use of as an immediately local re-
source in the social construction of discourse in the conversa-
tion.
Local resources once removed and nonlocal resources were
also used in the conversation. Since a major source of local
resources once removed were topics from the dinner that was
videotaped the week before, some description of what happened
in that dinner is necessary here by way of introduction.
50 / Frederick Erickson

During the first videotaped dinner, the father and mother

had functioned together as a conversational team. In the
second videotaped dinner, the guest occupied the role of the
other adult in the scene, a position usually occupied by the
father. In the previously recorded dinner, the father had a
significant role of leadership in the conversation; many of the
new topics that came up were either introduced by the father,
or were ratified by him through his giving attention to the
speaker who introduced the topic.
The previously videotaped dinner conversation had included
considerable discussion of the home-grown vegetables that were
part of the meal and that had come from the family's large gar-
den in the back yard. Another major topic had been how ex-
pensive things were becoming nowadays. This was linked to
the other major topic, since gardening and canning was a way
the family saved money on food. Discussion of the topics
'food from the garden' and 'things that are expensive nowa-
days' involved the use of listing as a prominent resource for
discourse organization. In the listing sequences, various family
members would alternate in contributing a single item in the list.
This seemed to be a conversational arrangement that was especi-
ally easy for the youngest children in the family to participate
in. All they had to do was come up with a list item every now
and then, and say that item at the point in real time in which
it was appropriate for the next item slot in the list sequence to
be performed. Fun was had with this; sometimes the list item
matched the overall topic by a kind of ironic connection rather
like punning. Talk was used in other kinds of playful ways;
for example, the father said that things are so expensive now-
adays, they need a money tree in the back yard as well as a
vegetable garden.
During the second videotaped dinner, speakers drew upon
topical resources from the previous videotaped dinner. In its
content, the topic 'things to eat that come from the garden'
harked back to last week's conversation. In terms of process,
the listing routines and playful uses of talk recalled the pre-
vious week's topics and the roles of family members as humor-
ists in the previous conversation.
Before turning to the transcript, a few notes about tran-
scription conventions are in order. Generally, the transcrip-
tion follows the conventions of the conversational analysts.
Special emphasis is placed on rhythmic and intonational fea-
tures of speech prosody. Dramatic shifts in pitch are occa-
sionally indicated by placing letters above or below the ordinary
line of text. Volume stress is occasionally indicated by an
underline. It is usually indicated by placement of the stressed
syllable at the left margin (without underline). A single line
of text usually represents a single breath group or tone group
of speech. Thus, the first syllable of each new line of text
is usually a stressed syllable, and syllables at the right margin
 Social Construction of Topical Cohesion in a Conversation / 51

of the line are enclitic to the syllable at the left margin of the
next line, as in pickup notes in music:
(6) G: I
love your salad
Periods (dots) indicate silence, with two periods indicating
approximately one-half second of silence (a sentence-terminal
pause), four dots indicating approximately one second of si-
lence, and a single dot indicating one-fourth second of silence
(roughly equivalent to a comma).
Sustained loud volume is indicated by capitalized letters.
This is used for Bobby's speech. In this conversation, he uses
two distinct pitch and volume registers: one very loud and
high pitched, with steeply rising and falling intonation con-
tours, and another at more normal volume and pitch level.
Elongation of syllables is indicated by successive colons
(SPOO::N). Simultaneously overlapping talk by two or more
speakers is indicated by a bracket:
(12) S: did
all this come foutta our ga:den?
M: Don't put your fingers in there.
Alternation of speech between which there is no overlap but
also no gap ('latching') is indicated by a bracket whose flaps
point in opposite directions, as in the following latched laugh-
ing between the mother and the guest:
(15) M: mhmhmhl
(16) G: LMhmhmh/
didn't you grow that in your yard?
The latching symbol is also used occasionally in the speech of
a single individual to indicate the absence of a gap between
the word or syllable at the end of one line and the beginning
of the next. This is more usually indicated by the slant mark
at the right margin of a line of text, as in (15). When a
slanted line is not followed by further speech by an individual,
the line indicates an interruption in speech which leaves a
word or phrase incomplete, e.g. 'n the di/ for 'n the dishes.
The numbers in the text do not indicate a constant unit (a
'turn' at speaking) because, given the frequent overlap of
speech and the interpenetration of one another's sentence and
clause units between multiple speakers, the notion of turn as
a discrete sequential unit often does not apply at various
points in this conversation. Consequently, the numbers in the
text are to be thought of as reference points for the reader
rather than as turn units. When the conversation splits into
two simultaneously occurring duets, this is indicated by the
subscript letters a and b.
52 / Frederick Erickson

The postural positions and interpersonal distance of the con-

versationalists are indicated in the transcript by diagrams ac-
companying the text. The rectangle represents the dinner
table, the brackets indicate the lines of the shoulders and
upper torso of an individual, and the arrow indicates the di-
rection of gaze of that individual. If there is no arrow for a
given individual, that means the individual was apparently
looking down at his/her plate. Bobby, the son (indicated by
53 in the transcript) shifts back and forth from one chair to
another during the conversation (see the diagram at the head
of the transcript), and this is shown in the postural diagrams
that appear throughout the transcript.
As the transcript begins, the people remaining at the dinner
table were finishing eating the main course of the meal, which
was lasagna and a salad. After a pause in talk during which
people were chewing their food, the mother addressed the
guest with an invitation to another helping of food (see Figure
1, pages 54-58).
Discussion. Let us consider four sets of production resources
and their use in social construction of topical cohesion within
and across multiple conversational floors: (1) posture and gaze,
(2) listing routines, (3) commonplace sources of topical content,
and (4) rhythmic organization of speech prosody.

Posture and gaze. As sources of contextualization cues, pos-

ture and gaze are nonverbal communication media that play an
important role at the immediately local level of production in the
establishment and maintenance of conversational floors. Speak-
ers in this conversation oriented to one another in three ways;
on the nonverbal channel by postural position and by gaze di-
rection, and on the verbal channel by the addressing of speech
to particular individuals as audience. These three dimensions
of orientation shifted together throughout the conversation, i.e.
a change in floor management (speaker-audience relationship)
cooccurred with changes in postural and gaze orientation. This
can be seen in the transcript by noting the configurations of
posture and gaze that were sustained from points (1) through
(19), across points (20a,b) through (28a,b), and across points
(29) through (43). The second of these three chunks of dis-
course is the one in which the conversational participation
structure shifted from a single floor, in which all four partners
interacted as a conversational quartet, to a double floor, par-
ticipated in by two pairs of partners: mother-guest, son-
daughter. Notice in the postural and gaze diagrams that in
the whole section before (20a), gaze and postural orientation
was shared across the two pairs of partners; for example, at
points (4) and (5), the guest and son faced one another and
looked directly at one another. From (20a,b) through (28b),
however, this pattern changed. Across that whole strip of
activity, mutual gaze and postural focus were not shared across
 Social Construction of Topical Cohesion in a Conversation / 53

the pairs of conversational partners. While the double floor

arrangement of talk lasted, the mother and guest focused pos-
ture and gaze orientation only on each other. This was also
true for the son and daughter, with what seems to be a momen-
tary exception at (24a), where the son gazed in the direction
of the mother. But what the son was actually doing at (24a)
was to gaze at the salad bowl, which was placed on the table
to his immediate left. Moreover, the mother did not direct gaze
or postural focus to the son, but continued at that point to
focus on the guest. So the pattern of across-team avoidance
of gaze and posture sharing continued to be maintained through
point (28b).
Looking now at point (29), it is apparent that the pattern of
postural and gaze configuration changed as the conversational
arrangement changed back to a single conversational floor. The
mother and son oriented to one another at point (29). From
then on, mutual orientation was again exchanged across the
pairs of the partners who from (20a,b) through (28b) had never
exchanged mutual gaze or postural focus. In sum, throughout
the whole conversation configurations of mutually sustained pos-
ture and gaze orientation seem to have been functioning as con-
textualization cues; by these means as well as by the content
of their talk, the conversational partners were telling each other
where the floor is now, and who is in it.
Topical content and process: Commonplaces and lists. There
are three main sources of topical resources in the conversation:
things in the garden, things on the dinner table, and topical
items said just previously. The first topical resource draws
upon the shared past history of the conversational partners by
invoking topics from last week's videotaped dinner conversation.
This is a matter of local production once removed from the im-
mediate scene. The second topical resource is derived from the
geography of the dinner table. Since the table and its contents
are there to be seen in the here and now by all the participants,
the use of the table by the son as a resource in the generation
of list items is a matter of immediately local production. Retro-
spective semantic tying back to the topical item said just pre-
viously is also a matter of immediately local production. All
three sources of topical content (especially the first two sources)
can be thought of as commonplaces, in that (1) all three sources
draw on understanding and experience shared in common by the
participants in the conversation, and (2) as places in semantic
space they are readily identifiable. Moreover, the first two
sources of topical content are semantic domains whose constitu-
ent parts—subtopics and list items—are connected by semantic
ties that are extremely obvious. All three sources of topical
content function well as resources for the fluency of speaker
production and listener comprehension in real time.
The overall sequence of topics and topical items in the tran-
script is shown in Figure 2, which also identifies the individual
5t / Frederick Erickson

_T

jA 0)
r in Hi 0)
Social Construction of Topical Cohesion in a Conversation / !>5

T
7-
TET <£ LJ

•j-o

1
o <••

2 S
ill Is?
•g bp«
 (20a) FAWKS . . AN' THE
(20b) M: L we'd have lasagna every other night
(D
(21a) S points to own napkin (21a) S: NAPKINS AN' THE GLASSES . N' THE (21b) M and G focus on each a
and glass, then to pep- PEPPAH N' THE (21b) G: Yeah that's one a1 my other, through posture »
per. D has focused on S, fa:vrite things to eat and eye contact 2.
through posture and eye o*
contact. S had pointed 7?
(clockwise) to pepper only. (22a) DRINKB n' salt and
D: |_'n salt rn
(22a) D reminds him of salt on love it
other side of table, and
S repeats verbal formula: (22b) M: . have la
'salt n' peppah'. He
says this less loudly than
the words that precede and S: peppah . . 'N THE MILK . . 'N the
follow 'salt and pepper', D: m i:lk
apparently as a sort of
aside. He then returns to
a loud voice in saying "N
THE MILK'. Apparently
'DRINKS' meant soda,
since both D and S add
'MILK' as a separate item.
(23a) S points to cheese, then (23a) cheese 'N the/
to salad bowl (23b) M shifts in chair but
S:/ 'n the doesn't look at S
(23b) M: . . n' they should have a
steak tree . . . .Ml E-
[HeHeHeHefHa:
(24b) G:LThey/
that wouldn't be (as good as) a
moneytree he he..you could get y/l |
(25a) D: salad bowl . . . .
you said (25b) M: (i think I'd rather have a
money tree"!
(26a) D points to milk and (26a) D: drinks . . . you said
soda bottles (26b) G: | Yeah I'd rather have a/
Social Construction of Topical Cohesion in a Conversation / 57

: I
alS

So
eg

g .

o
"Si
V«
X
{ %•
a • Sw
•g . >
•3. n'
f III

01
.2
|2
c *
•o o t3
§-ga
(39) S: N, N, N, N, I WANNA MAKE/ in
(39) M does not move, speak, or oo
LET'S MAKE SOME CAKE look at S--D is 1drinking
milk. At 'cuz I D puts
down glass on table. It
hits with a rap. D looks at •n
bowl
never get the
I get the
Cuz I
S
I
to
bo:wT1 o"
sr
):
(40) D: LNei
LNeither do
I:: m
(40) D looks at S but S looks
at M
(41) S: YEA : : : ( " : H? WHEN I MADE A (41) As D says 'yea::h' S turns
LYea::h? to face D
(42) S: | CAKE YOU GOT THE
SPOO::N: . . 'N . . 'N/
(43) /EVERYTHING
Social Construction of Topical Cohesion in a Conversation / 59

Figure 2. Flow chart of topical items.

TOPICS IN LAST WEEK'S I Want

'DINNER CONVERSATION ' more food? -(Table)

I Things to eat tonight i

j that we grew in our I
I garden in the back | G: Love your salad (Table)
| yard.

How much things cost |

nowadays. i ^ S: Did salad ingredi-
ents come from • (Table)
We need a money tree I garden?
in the back yard
(said by father).

-(Table)
60 / Frederick Erickson

who introduced the topic or constituent topical item, and illus-

trates the shift between single and double conversational floors,
Keeping in mind both the flow chart in Figure 2 and the
transcript of the conversation, it is apparent that topical con-
tent is drawn from what has just been said (as indicated by
arrows pointing to the next previous topical item), from the ob-
jects on the dinner table (also indicated by arrows), and by
direct and indirect reference to topics of last week's conversa-
tion (indirect reference indicated by arrows with dotted lines,
direct reference indicated by arrows with straight lines).
At the beginning of the transcribed portion of the conversa-
tion, after there had been a long silence while people at the
table chewed their food, the mother offered the guest more
food.
The guest responded with compliments about the food:
(2) G: Oh no thanks
this is great
(6) G: I
love your salad
This marks a transition in the dinner between a food consump-
tion-dominated period of chewing work and a talk production-
dominated period of conversational work. Reciprocally, the
guest responded with 'talk items' to the mother's offer of 'food
items' (on what was characterized earlier as the 'food channel'
of phatic communion). From that point on in the meal, talk be-
came the foregrounded activity (and the talk channel became the
primary medium of phatic communion), with food consumption
relegated to the background.
At point (10), the son harked back indirectly to last week's
topic, 'things to eat that came out of our garden', as he
pointed to the guest's salad bowl and said:
(10) S: MA::, DID
ALLA THIS COME OUTTA OUR
GA.-DEN?
After thep mother replied 'Everything came out of the garden',
the son introduced another feature from the previous week's
conversation, the overall character of playfulness in dinner
table talk. At point (14), the son responded to the mother
by playfully reading her previous everything literally, pointing
to the aluminum pan of (commercially prepared and frozen)
lasagna, and asking counterfactually if that, too, had come
from the garden. The mother acknowledged this sally of wit
with a laugh which was joined in by the guest, who then con-
tinued the counterfactual play by saying to the son:
 Social Construction of Topical Cohesion in a Conversation / 61

(16) G: didn't you grow that in your

iTl
yard"?]
S: Lin the di7~|
Lla
G: sagna bushes back there?
While the guest introduced the imaginary plant into the con-
versation, she also was interrupting the son, who had begun
to generate another item in a counterfactual list: 'things on
the table that grew in the garden'. His next item was dishes.
As he started to say this ('di/'), he was pointing to the plate
in front of him, which was right next to the pan of lasagna on
the table (refer again to the diagram of the table at the head
of the transcript). The next list items, napkins and glasses,
were said while pointing to his own napkin and glass, located
next to his plate. Next to his glass was the pepper shaker
(see table diagram). His pointing to the pepper began what
from his point of view was a counterclockwise movement of
pointing around the table. Next in the counterclockwise arc
came the drinks (soda) and milk, followed by the grated cheese.
Past the cheese was the guest's plate and her individual salad
bowl, but plates and the guest's salad had already been men-
tioned earlier, as had the items at the son's own place setting.
Continuing on the counterclockwise arc, the next item found on
the table was the lasagna pan, but that, too, had already been
mentioned. The only item yet unmentioned was the salad bowl,
which was the last item the son mentioned in his list.
It seems clear that the son was using the geographic location
of objects on the dinner table as a means of generating a list.
Further evidence for this inference is found in the comment of
the sister at (22a) and in her brother's reaction to that com-
ment. Apparently, the brother's idiosyncratic list formula was
not yet understood fully by his sister—it did not yet perform
a commonplace role for both speaker and audience, as evidenced
by the sister's apparent mistake in comprehension at this point.
(21a) S: AN THE
NAPKINS AN' THE GLASSES . 'N THE
PEPPAH . . . . f
. 'N THE
(22a) DRINKJ3 . . n salt and (said more
softly)
D: L'n salt
S: Ipeppah . . 'N THE MILK
At (21a) the son had begun his counterclockwise pointing rou-
tine, going from his glass to the pepper shaker to the drinks
(see the table diagram). As he got past the pepper to the
drinks (which were in next position according to his geographic
listing formula), his sister corrected him by reminding him that
salt goes with the pepper. But this was an apparent mistake
62 / Frederick Erickson

on her part. Salt and pepper go together as a pair (with

pepper in second position) on the basis of a commonplace
verbal formula in American English. The two do not go to-
gether according to the commonplace table geography formula
the son appears to have been using. (According to the table
geography, salt would have come next after the cheese). The
sister here seems to have been trying to cooperate in her
brother's list item-generating, but not to have understood the
generative principle he was using. The less loud way in which
the brother said "n salt and peppah' seems to have acknowl-
edged only as an aside the sister's irrelevant correction. After
the aside, the brother returned to his own list formula, speak-
ing again in his loud voice register.
The use of the dinner table as an immediately local production
resource for item-generation was accompanied by use of an en-
tirely nonlocal production resource. This was the syntactic
structure of the listing formula itself, knowledge of which is
shared not only beyond this family with the Boston area Italian-
American speech community, but beyond that speech community
with the linguistic community of English speakers generally.
Syntactically, the son's listing formula is of the very simplest
sort: noun slot plus conjunction plus definite article, reiter-
ated: 'the X and the Y and the Z'. This syntactic string is
produced with the simplest possible speech rhythm, a recurring
anapaest: (S2 J> J~2 J )• T n * s speech rhythm, too, is a
nonlocal resource. Taken together, the simple syntax, simple
speech rhythm, and obvious table geography are sets of r e -
sources that make for an almost foolproof system for fluent pro-
duction of topical items in real time. The listing rhythm con-
tinues like an air hammer, continuously cutting through the con-
versation that is being carried on by the mother and guest on
the cooccurring second conversational floor. Let us turn now
to consider the organization of conversation between the mother
and guest, from (20b) through (28b).
The sources of topical items are two: the immediately local
resource of the content of prior turns (especially the son's
mention of lasagna), and the local once-removed resource of
the money tree topic from last week's conversation (see Figure
2).
In the mother-guest conversation, there is a listing connec-
tion as a cohesion device, but it is of a slightly more complex
sort than the list found in the son-daughter conversation. The
list items are three counterfactual plants: lasagna bush, steak
tree, and money tree. The lasagna bush topic was initiated by
the guest at (17) and responded to by the mother at (18) and
(20a). The steak tree was introduced, after a brief pause, by
the mother at (23b). This was a play upon both lasagna bush
(from the conversation) and money tree (from the father's funny
line in last week's conversation). Reference by the guest to
the money tree immediately followed at (24b):
 Social Construction of Topical Cohesion in a Conversation / 63

(23b) M: . . an' they should have a

steak tree M
hehehehefha:
(24b) G: LThey/
that wouldn't be (as good as) a
moneytree hehe . . you could get y/
(25b) M: I think I'd rather have a
money tree
Interspersed between the mentions of the three imaginary
plants are other comments. From (20b) through (22b), there
was a brief, elliptical discussion of lasagna and its pasta 'cou-
sin' , ravioli, in which the guest collaborated by saying that
she liked those foods, too:
(20b) M: If we actually had a lasagna bush in the back yard
we'd have lasagna every other night
(21b) G: Yeah that's one a' my
fa:vrite things to eat .
love it
(22b) M: have lal
M: 1 sagna . : . raviojjfl. .
G: m : : L mhm
(23b) M: . . a n ' they should have a
steak tree
The steak tree appeared in (23b) as the next item in the list
consisting of lasagna. ravioli, steak. Steak tree also tied back
semantically to the prior lasagna bush, and to the previous
week's money tree. From (23b) on, the interspersed comments
had to do with what one would prefer, a steak tree or a money
tree.
The whole strip involves counterfactual conditions. It was
tied together across turns by manipulation of syntactic and
lexical choices through the use of conditional tense, by the
modals would, should, and could, and later by the conditional
would rather. These syntactic construction patterns offer
alternative opportunities—full form, contraction, and deletion—
altering the length and rhythm of utterances. The full form is
found only three times:
(23b) M: an' they should have a steak tree
(24b) G: that wouldn't be as good as a money tree
you could get y/
More common is the contracted form, as in (20b) and (24b),
where we'd substitutes for we would, and I'd substitutes for
I would.
(20b) M: We'd have lasagna every other night.
(24b) M: I'd rather have a money tree.
64 / Frederick Erickson

Deletion of the modal also occurs, as in (22b), where the

deleted item seems to be we would:
G: ( I ) | love it
(22b) M: . . (we would) | have la
M: sagna . ravio: li
Notice here that the pronouns can also be deleted, as in the
guest's substitution at (22b) of love it for I love it.
The ultimate deletion seems to have occurred right at the end
of the strip, at (27b) and (28b).
(27b) M: /I'd rather have a big
tree. HA HA .
(28b) stea:ks. . I mean y'know I'll cook
CHA:COAL BROI:L or somethin'
By contrastive stress on stea:ks and by contrastive sustained
loudness on CHA:COAL BROIL the mother seems to have been
saying something like 'I wouldn't want steak all the time. But
I'd like it sometimes,, and if we had a money tree we could have
steak whenever we want it. And then we'd have it the best way
—charcoal broiled'.
It may be that after a pause she said stea:ks so elliptically
(deleting both noun and verb) because of the increased dead
air time in the other conversation (see full transcript). At
that point the son-daughter conversation had begun to run
down. The topic resources of the table had been exhausted,
and the syntactic listing formula had also been abandoned.
What was left were stressed one-syllable words—drinks, milk—
separated by pause (see (26a)-(28b)). Perhaps the mother's
utterance, stea:ks was echoing the food items in the other con-
versation .
There is considerable evidence that some attention was being
paid across the pairs of conversational partners. At the end of
the double conversational floor, as the mother said CHA-.COAL
BROIL after just having said stea:ks, the son looked across at
the mother (in the pause at (28b)). Talk between the son and
the daughter had stopped. Then the mother turned to the son,
looked directly at him, and at (29) said "baked potato'. This
is both a food item tie (baked potato is the second part of the
conventional verbal pair steak and baked potato) and a social
inclusion tie (the mother knows the son likes baked potato). At
this point the single conversational floor resumed.
Even better evidence for attention across pairs of conversa-
tional partners comes from close investigation of speech rhythm,
and it is to this final issue that we now turn.
Roles of rhythm in maintaining cohesion in topics and in
floors. The regular periodicity of speech and nonverbal
 Social Construction of Topical Cohesion in a Conversation / 65

behavior rhythm appears to enable conversational partners to

coordinate their action together within an overall ensemble.
Through rhythm, strategic next moments in real time can be
anticipated. This seems to be a necessary condition for inter-
action that is fully 'social', in the sense in which I have used
that term here (see also the discussions in Erickson 1981, Erick-
son and Shultz 1981, Scollon this volume, and Sudnow 1980).
In this conversation, the role of the reiterated anapaest
rhythm pattern in maintaining fluency in list-item production
has already been discussed. In that pattern the stressed sylla-
bles, which occur at a regularly periodic interval in relation to
each other, contain the next piece of new information—the next
noun in the list.
The tendency for a piece of new information to be found at
the point of the stressed tonal nucleus in a breath group occurs
not only in the relatively simple anapaest in the son's listing
formula. It occurs as well in the more complex patterns of the
speech of the mother and guest. In the doubly floored phase of
the transcribed text (20a,b)-(28a,b) in the adults' speech, the
stressed tonal nucleus often not only is a point of new informa-
tional content, but occurs at the end of a syntactic clause. In
real time performance, the stressed tonal nuclei occur at a regu-
lar rhythmic interval. In performance, these chunks of speech
are almost identical in length:
(20b) we'd have lasagna every other night
(21b) yeah that's one a'my farvrite things to eat
(23b) an' they should have a steak tree
When these clause units in the mother-guest conversation are
juxtaposed with anapaest units in the simultaneously occurring
son-daughter conversation, the two sets of conversational
rhythm pattern articulate together in an overall rhythmic shape,
or ensemble. Moreover, the two conversations are articulated
so that the stressed tonal nuclei in each conversation occur
either at a point of silence in the other conversation, or occur
simultaneously. In either case, this rhythmic integration across
the two pairs of conversational partners enables both conversa-
tions to proceed with neither one upsetting either the rhythmic
cohesion of the other or the points in the other conversation at
which important new information is being communicated.
In short, it seems that rhythmic fine tuning enables this
multiply floored conversation to proceed without having the
overlapping talk constitute an interruption. I am arguing that
this rhythmic integration is social in nature. Conversational
partners appear to be rhythmically field-sensitive. They appear
to be taking action on account of the rhythmic action of others,
not only within conversational pairs, but across pairs as well.
More generally, rhythm seems to be the fundamental social glue
by which cohesive discourse is maintained in conversation,
66 / Frederick Erickson

within and across turns and sets of turns, and within and be-
tween the speech of speakers in conversational floors.
These points can be illustrated by displaying the rhythmic
patterns of speech in musical notation. The passage in Figure
3 comes from points (18) through (23b) in the transcript.
Notice that in measure 2, as the son had gotten his list go-
ing, Mn the dishes nT the forks', he completed that unit in time
for the mother's 'we'd have lasagna every other night' not to
interfere with his speech or be interfered with by his speech.
In that instance, the two avoided interference by avoiding
overlap. In measure 3, even though the mother's and son's
speech overlapped, interference was avoided by the exact match-
ing of stressed syllables (night and napkin) and by the mother's
doubling of speed relative to that of the son right before the
next stressed syllable (see every other in relation to 'n the).
Another way of avoiding interference while preserving overall
rhythmic ensemble is rapid-fire alternation between speakers, of
onsets and offsets of speech. The Western musical term for this
alternation, found in medieval European vocal duets, is 'hocket'.
Hocketing can be seen in measures 6 and 7, where the most
finely tuned coordination across the two conversations is appar-
ent—from the son's drinks to the daughter's 'n salt, to the
guest's tove it to the mother's lasagna which was accompanied by
the son's 'n salt 'n and was followed by the guest's mm. This
is a dramatic example of individuals interacting closely in real
time and yet staying out of one another's way in time. Viewed
in the total rhythmic context at this point, the mother's deletion
of the modal we would before have lasagna can be seen as a
syntactic accommodation to the rhythmic organization of what the
other speakers were doing reciprocally and complementarily in
immediately adjacent and simultaneous moments of real time.
People do, indeed, seem here to have been functioning as en-
vironments for one another.
Conclusion. In the musical notation we can see local produc-
tion resources being used in a radically local way to maintain
cohesion at the level of utterance and topical item. But in the
same time we are also seeing nonlocal production resources be-
ing used, for what is being produced locally in the conversation
is an instance of a culturally patterned way of speaking that is
widely and nonlocally shared within the ethnic speech community:
the Italian-American way of conducting multiply floored conver-
sation in which simultaneously occurring and rapidly alternating
speech does not constitute interruption. The overall pattern of
multiple flooring is not specific to Italian-Americans. It has
been found among Americans of East European Jewish descent
(Tannen 1981) and Americans of Cape Verdean descent (Gomes
1979). However, the specific organizational features by which
multiply floored conversations are managed by speakers do ap-
pear to be speech community-specific.
Social Construction of Topical Cohesion in a Conversation / 67

00
c

I
IT4i
>
* c
r
CO j
£ i
UL
-
t jr
E
<*-1

"e

i .5
n

c —1
w

-S
[ 1-
—1

A. c S

1
i.
i r "•

T
-E 1 •S
>•
-«•. :

4
c
H

OS O Q
O
68 / Frederick Erickson

In addition, we have seen both immediately local resources for

topical content being used (the dinner table) and local resources
once removed being used (last week's conversation). It has
been argued that topical cohesion and the conversational floors
within which topics are manifested, are both products of social
construction. The resources for this production are seen as
being simultaneously local, local once removed, and nonlocal.
The analysis points to the need for theory which comprehends
these various levels of organization, from syntax and moment-to-
moment rhythm, through discourse, to social life in general and
to broadly shared cultural patterns. The analysis also points
to the need for empirical procedures that unite the ethnography
of speaking with sociolinguistic and linguistic microanalysis.
NOTE
The research reported here was supported in part by a grant
to the author from the Spencer Foundation and by a post-
doctoral fellowship to Jeffrey Shultz from NIMH. The author
is indebted to Deborah Tannen for editorial advice, to Doro
Franck, Brigitte Jordan, and Ron Simons for advice and criti-
cism, and to Susan Florio for collaboration in the original re-
search (she is the 'guest' in the transcript) and for comments
on this analysis.
REFERENCES

Bateson, Gregory. 1972. The message this is play. In:

Steps toward an ecology of mind. New York. Ballantine
Books.
Birdwhistell, Ray L. 1970. Kinesics and context: Essays on
body motion communication. Philadelphia: University of
Pennsylvania Press.
Bremme, Donald W. 1977. Accomplishing a classroom event:
A microethnography of first circle. Qualifying paper, Har-
vard Graduate School of Education.
Erickson, Frederick. 1981. Timing and context in everyday
discourse. In: Children's oral communication. Edited by
O. W. Dickson. New York: Academic Press.
Erickson, Frederick, and Jeffrey Shultz. 1977. When is a
context? Some issues and methods in the analysis of social
competence. Quarterly Newsletter of the Institute for Com-
parative Human Development 1(2). 5-10. Reprinted 1981 in:
Ethnography and language in educational settings. Norwood,
N.J.: Ablex Press.
Erickson, Frederick, and Jeffrey Shultz. 1981. The counselor
as gatekeeper: Social interaction in interviews. New York:
Academic Press.
Florio, Susan. 1978. Learning how to go to school: An eth-
nography of interaction in a kindergarten/first grade
 Social Construction of Topical Cohesion in a Conversation / 69

classroom. Doctoral dissertation. Harvard Graduate School

of Education.
Goffman, Erving. 1976. Replies and responses. Language in
Society 5.257-313.
Gomes, Louis Antone, J r . 1979. Social interaction and social
identity: A study of two kindergarten children. Ph.D.
dissertation. Harvard Graduate School of Education.
Gumperz, John J. 1977. Sodocultural knowledge in conversa-
tional inference. In: Georgetown University Round Table on
Languages and Linguistics 1977. Edited by Muriel Saville-
Troike. Washington, D.C.: Georgetown University Press.
191-212.
Hymes, Dell. 1974. Foundations in sociolinguistics: An ethno-
graphic approach. Philadelphia: University of Pennsylvania
Press.
Keenan, Elinor, and Bambi Schieffelin. 1976. Topic as a dis-
course notion. In: Subject and topic. Edited by C. Li.
New York: Academic Press.
Kendon, Adam. 1977. Studies in the behavior of social inter-
action. Bloomington: University of Indiana Press.
Lechner, Joan Marie, O.S.U. 1962. Renaissance concepts of
the commonplaces. New York: Pageant Press.
Malinowski, Bronislaw. 1923. The problem of meaning in primi-
tive languages. In: The meaning of meaning. By C. K.
Ogden and I. A. Richards. New York: Harcourt, Brace
and World. Supplement 1, 296-336.
McDermott, Ray P. 1976. Kids make sense: An ethnographic
account of the interactional management of success and failure
in one first grade classroom. Unpublished dissertation.
Stanford University.
Ong, Walter, S.J. 1977. Typographic rhapsody. In: Inter-
faces of the word: Studies in the evolution of consciousness
and culture. Ithaca and London: Cornell University Press.
Sacks, Harvey, Emmanuel Schegloff, and Gail Jefferson. 1974.
A simplest systematics for the organization of turn-taking for
conversation. Lg. 50.696-735.
Scheflen, Albert E. 1973. Communicational structure: Analysis
of a psychotherapy transaction. Bloomington: University of
Indiana Press.
Sherzer, Joel. 1977. The ethnography of speaking: A critical
appraisal. In: Georgetown University Round Table on Lan-
guages and Linguistics 1977. Edited by Muriel Saville-Troike.
Washington, D.C.: Georgetown University Press. 43-58.
Shultz, Jeffrey, and Susan Florio. 1979. Stop and freeze:
The negotiation of social and physical space in a kindergarten/
first grade classroom. Anthropology and Education Quarterly
10.3:166-181.
Shultz, Jeffrey, Susan Florio, and Frederick Erickson. (in
press) Where's the floor? Aspects of the cultural organiza-
tion of social relationships in communication at home and at
school. In: Ethnography and education: Children in and'
70 / Frederick Erickson

out of school. Edited by Perry Gilmore and Alan Glatthorn.

Washington, D.C.: Center for Applied Linguistics.
Sudnow, David. 1980. Talk's body: A meditation between two
keyboards. Harmondsworth, Middlesex, England: Penguin.
Tannen, Deborah. 1.981. New York Jewish conversational style.
International Journal of the Sociology of Language 30.133-149.
Weber, Max. 1922 (1978). Wirtschaft und Gesellschaft. Vol. 1,
pp. 1-14 (Tubingen, 1911). Translated as: The nature of
social activity. In: Weber: Selections in translation. Edited
by W. G. Runciman. Cambridge: Cambridge University Press.
7-32.
 DISCOURSE AS AN INTERACTIONAL ACHIEVEMENT
SOME USES OF 'UH HUH1 AND OTHER THINGS
THAT COME BETWEEN SENTENCES

Emanuel A. Schegloff
University of California, Los Angeles

1. From the standpoint of students of 'discourse', conver-

sation and other forms of talk in interaction are subvarieties
of discourse. What comes to be minimally criterial for discourse
is the presence, in some sort of coherent relationship, of a
spate of language use composed of more than one sentence (or
whatever other unit is treated as grammatically fundamental).
It is then common to be concerned (1) with the basis for the
apparent coherence between the several components of the dis-
course, (2) with the cognitive structure of the unit, and (3)
with the mechanisms by which it is analyzed or decoded on
reception. There may be an effort to discern quasi-syntactic
relationships between successive parts of the discourse--be-
tween successive sentences, for example. And discourse units,
such as paragraphs, may be found to be constituted by such
quasi-syntactic relationships. The actual enactment of the
discourse--for example, its telling--often seems to be treated
as the behavioral realization of a preplanned cognitive unit.
The prototype discourse for such an approach is the narra-
tive or lecture, which readily lends itself to treatment as the
product of a single speaker, whose cognitive apparatus under-
lies and shapes it.
For the student of talk in interaction, discourse (still mini-
mally defined as a spate of talk composed of more than one
sentence or other fundamental unit) is more usefully treated
as one type of production in conversation (or other speech-
exchange situation). Note that, although some sorts of ob-
jects for analysis, such as written stories, memoranda, and
legal documents, may appear suitable for analysis under the
former conception but not the latter, in many cases the same
objects of inquiry are seen differently from the two points of
view. The common discourse-analytic standpoint treats the

71
72 / Emanuel A. Schegloff

lecture, or sermon, or story told in an elicitation interview,

campfire setting, or around the table, as the product of a
single speaker and a single mind; the conversation-analytic
angle of inquiry does not let go of the fact that speech-
exchange systems are involved, in which more than one par-
ticipant is present and relevant to the talk, even when only
one does any talking.
Let me recount an old experience. Once I had trouble
understanding certain monologues in Shakespeare. I was
watching a series of rehearsals of The Winter's Tale by the
Canadian Shakespeare Company on public television, and had
gotten down my Complete Shakespeare and was following the
text. In the monologue in question, I could see how line 2
followed 1, 3 followed 2, and line 4 followed 3; but 1 just
could not figure out how line 5 followed line 4. And then I
saw in a series of rehearsals that, the authoritative text in
front of me to the contrary notwithstanding, line 5 did not
follow line 4; some action followed line 4, and line 5 followed
that action. And what was at issue in the rehearsals was
what that action should be and who should do it, for the sense
of line 5, and ensuing lines, would be affected in a major way
by it.
Anyone who has lectured to a class knows that the (often
silent) reactions of the audience--the wrinkling of brows at
some point in its course, a few smiles or chuckles or nods,
or their absence--can have marked consequences for the talk
which follows: whether, for example, the just preceding
point is reviewed, elaborated, put more simply, etc., or
whether the talk moves quickly on to the next point, and x per-
haps to a more subtle point than was previously planned.
If this is the case in such a situation of talk-in-interaction as
the lecture, then its relevance should be entertained as well
for experiments, elicitation interviews, and ordinary conver-
sation.
Clearly, different speech-exchange systems are involved in
lectures and ordinary conversation, with different turn-taking
practices providing quite differently structured opportunities
to talk or participate in other ways. That is one reason why
the reference to lecture situations describes wrinkled brows
and smiles and nods, rather than utterances or even 'uh huh's.
Clearly, as well, in several different types of speech-exchange
situations, there can be occasions in which participation is
constructed by a speaker in continuing response to inter-
actional contingencies and opportunities from moment to mo-
ment, and occasions in which a participant has a preformed
notion, and sometimes .a prespecified text, of what is to be
said, and plows ahead with it in substantial (though rarely
total) disregard for what is transpiring in the course of this
talking. But these two extremes are not equally likely to
occur in the various types of speech-exchange1 situations; the
prespecified text, adhered to 'no matter what , is much less
 Discourse as an Interactional Achievement / 73

common (and for good structural reasons) in ordinary conver-

sation than in sermons or lectures. Even the wholly prespeci-
fied talk, which most approximates the enactment of a cognitive
object, must be adapted in its delivery to its occasion, and will
certainly have been designed with attention to its recipients and
the situation of its delivery in the first place—both aspects of
interactional sensitivity. However, it should be clear at the
outset that in what follows I am most centrally concerned with
what I take to be both the primordial and the most common
setting and organization for the use of language—ordinary con-
versation. Although much of what I have to say is relevant to
other settings, the way in which orientation to co-participants
and interactional structure matter to discourse and its forma-
tion, will vary in different speech exchange systems with differ-
ent turn-taking systems.
Important analytic leverage can be gained if the examination
of any discourse is conducted in a manner guided by the
following.2 (1) The discourse should be treated as an achieve-
ment; that involves treating the discourse as something 'pro-
duced' over time, incrementally accomplished, rather than born
naturally whole out of the speaker's forehead, the delivery of
a cognitive plan. (2) The accomplishment or achievement is
an interactional one. Quite aside from whatever individual
cognitive or processing achievements might be involved (which
are not to be treated only as anterior to the interactional), the
production of a spate of talk by one speaker is something
which involves collaboration with the other parties present,
and that collaboration is interactive in character, and inter-
laced throughout the discourse, that i s , it is an ongoing ac-
complishment, rather than a pact signed at the beginning,
after which the discourse is produced entirely as a matter of
individual effort. (3) The character of this interactional ac-
complishment is at least in part shaped by the socdosequential
organization of participation in conversation, for example by
its turn-taking organization, which is not organized to be in-
different to the size of the turns parties take, but whose
underlying (though supercessable) organization is designed to
minimize turn size. It is this feature which requires us to see
'discourse' and 'discourse units' which have overcome this
bias as achievements and accomplishments. (4) Because the
actual outcome will have been achieved by the parties in real
time and as, at each point, a contingent accomplishment, the
mechanisms of the achievement and its effort are displayed,
or are analyzably hidden in or absent from, various bits of
behavior composing and accompanying that discourse, and
analyzable with it.
One class of such behavior which is implicated in the
achievement of discourses in conversation is the concern of
this paper. Instances of the class take the form of vocaliza-
tions such as 'uh huh', 'mm hmm', 'yeah', and others as well
74 / Emanuel A. Schegloff

as head-gestures such as nods. 3 These, as well as other,

bits of 1talk and behavior produced by other than the 'main
speaker are regularly discarded when discourses—the stories,
the arguments, etc.--are extracted from the tangle of detail
which composed their actual occurrence. 'The story' is puri-
fied of them in the course of its extraction, both by lay re-
counters and by professional analysts. It is this separation
of bits of talk, otherwise intercalated with each other and con-
tingent on one another, into two distinct classes, of which one
is the 'real talk' (the story, the 'what-was-being-said') and the
other conversational 'detritus' (apparently lacking semantic
content, and seemingly not contributing to the substance of
what the discourse ends up having said), which makes possi-
ble the notion of 'discourse' as a single speaker's, and a single
mind's, product. It is a consequence as well that the inter-
actional animus and dynamic of the spate of talk can disappear
into the cognitive structure and quasi-syntactic composition of
the discourse. What has been discarded may itself be picked
up by investigators--typically other investigators, even other
sorts of investigators--for separate treatment under such
rubrics as 'accompaniment signals' (Kendon 1967) or 'back-
channel' actions (Yngve 1970; cf. Duncan and Fiske 1977).
But, as I urge later, the fact that both parts of the occa-
sion--the teller's telling and the behavior of the recipients--
may be subjected to study does not restore the interactivity
lost when the former is extracted from the latter. For the
parts of the telling appear to follow one another, rather than
each following some responsive behavior by a recipient (or the
lack thereof); and what recipients produced after this or that
part of the telling has been removed from the environment of
that to which it was responsive. From 'discourse' and 'listener
behavior' so conceived and studied, it is unlikely that one will
be able to reassemble the actual structure of 'talking at length
in conversation'.
In what follows, I first elaborate a bit on the meaning of,
and the reasons for, treating the occurrence of discourse in
conversation as 'an achievement'. One mechanism for that
achievement has its focus at points at which recipients or
hearers could begin talking but content themselves with 'uh
huh' and the like instead, after which prior speakers continue.
I briefly discuss recent treatments of vocalizations such as 'uh
huh' and 'yeah', and then offer some alternatives.
2. Why should the existence of a 'discourse' (a multi-
sentence unit) in ordinary conversation be treated as an
achievement? Elsewhere (Sacks, Schegloff, and Jefferson
1974) it has been argued that speakers construct utterances
in turns at talk out of describable structured units, with
recognizable possible completions. In English, some lexical
items (e.g. 'hello', 'yes', 'who'), some phrasal units, some
clausal units, and sentences constitute such 'turn-constructional
 Discourse as an Interactional Achievement / 75

units'.. The end of any such unit is a possible completion of

a turn, and possible completions of turns are places at which
potential next speakers appropriately start next turns. If
this is the case, then an underlying structure of turn distri-
bution is in operation which organizes interactive enforcement,
or potential enforcement, of a minimization of turn size.
If such a system is in operation, then a constraint for
single-unit turns has at least two sources. First, there is an
organizational basis for current nonspeakers to monitor for the
possible completion of first units in a current turn as a place
to start next turns. And second, there is an orientation by
speakers starting a turn to the organizationally motivated orien-
tation of others to so start up, which can engender a design-
ing of the talk in a turn to be so organized as to get what
needs to be said said before the end of the first unit's com-
pletion. The second of these can contribute to making the
smooth operation of the first viable and routine. A great many
turns at talk in conversation thus end up being one unit long.
With all of this, it is obvious that some turns at talk end up
having more than one unit in them. Nor is this an anomaly,
or counterevidence. It does invite exploration of the possible
existence (and features) of methodical ways in which such
multi-unit turns are achieved. If there are such ways, then
their use may serve as additional evidence of the underlying
organization which these methods are used to supersede, at
the same time as they explicate the work of achieving the
supersession--the discourse. Although this is not the place
to undertake an extensive, let alone exhaustive, account of
such methodical devices, several may be mentioned to supply
a sense of the sorts of phenomena that are involved.
One class of methods by which multi-unit turns may be
achieved is that composed of devices initiated by the potential
speaker of the multi-unit turn.
(a) The potential discourse-speaker may indicate from the
beginning of the turn an interest in producing a more-than-
one-unit turn. For example, the speaker may begin with a
list-initiating marker, such as 'first of all',1* projecting thereby
that after the turn-unit in which the 'first' is done, more will
follow. Note that there may otherwise be no particular need
to pre-mark an item as first in a list (i.e. besides leaving it
to be so discovered over the course of extended talk, by vir-
tue of eventual subsequent items) other than the problem of
getting to produce subsequent items. Beginning a turn this
way recognizes the turn-taking contingency, and, by project-
ing a multi-unit turn, invites recipients to hold off talking
where they might otherwise start, so that the 'post-first-units'
may have room to be produced.
(b) Indeed, the turn-position, or turn-opportunity, in which
the beginning of a projected multi-unit turn could be pro-
duced, may instead be entirely devoted to a whole turn which
is focused on doing the projecting (as the list-initiating marker
76 / Emanuel A. Schegloff

does within a t u r n ) . Some years ago, Sacks (1974) described

one such operation under the rubric 'story prefaces', and more
recently, I have described under the term 'pre-pre' a similar
logic of use underlying one class of occurrences of utterances
of the form 'Can I ask you a question' (Schegloff 1980). In
both cases, a course of talk is projected which involves more
than one turn-constructional unit, and the talk begins with a
display of that projection. Note that it remains for recipients
to honor this projection, and to withhold talk at the points at
which it would otherwise be appropriate. Although initiated
by the intending extended-turn speaker, if an extended turn
results, it will have involved interactive accomplishment by
both speaker and recipients, the latter being recipients only
by abjuring their possible status as speakers. The list-
initiating marker, or story preface, or 'pre-pre' (i.e. 'Can I
ask you a question') are the overt markers of orientation to
the constraints making achievement of discourse problematic,
and of the effort directed to superseding them.
(c) Speakers may also employ methodical devices for achiev-
ing a multi-unit turn at positions other than the beginning of
the turn in question. There i s , for example, what can be
called the 'rush through'--a practice in which a speaker, ap-
proaching a possible completion of a turn-constructional unit,
speeds up the pace of the talk, withholds a dropping pitch
or the intake of breath, and phrases the talk to bridge what
would otherwise be the juncture at the end of a unit. In-
stead, the speaker 'rushes through' the juncture without in-
breath, reaches a point well into a next unit (e.g. next sen-
tence), and there stops for a bit, for an inbreath, etc.
(Schegloff 1973). Here the turn-extension device is initiated
near the otherwise-possible-end of the turn, rather than at
its beginning. Once again, interaction and collaboration are
involved, for recipients could start up despite the displayed
intention of current speaker to continue, and produce thereby
at least a 'floor fight'. Once again, the turn-extension de-
vice exhibits, on the speaker's part, an orientation to the
imminent possibility of another starting up as s/he approaches
the end of the turn-unit. Once again, if successful at getting
to produce a multi-unit turn or discourse, the talk displays
the special effort involved in achieving it.
Of course, not all multi-unit turns are the result of speaker-
initiated methods designed to achieve them. Some multi-unit
turns are the outcome of a different methodical production.
A speaker produces a one-unit turn, at whose possible comple-
tion no co-participant starts a next turn. Then one way the
talk may continue is by the prior speaker talking again, some-
times by starting a new turn unit. 5 On possible completion of
the now added second unit, a multi-unit turn has been pro-
duced; of course, the same cycle may occur on the next
possible completion as well. 6 In cases of this sort, the course
of action which issues in a multi-unit turn is 'initiated' by a
 Discourse as an Interactional Achievement / 77

recipient, and not by an (intending) speaker of (what ends

up as) the multi-unit turn, or discourse unit. Once again,
interactional achievement is involved, each participant orient-
ing to the other(s), and all oriented to the underlying turn-
taking organization, which is itself an interactionally driven
and constrained organization. Once again, signs of the col-
laborative work are marbled through the talk--in this case,
in the form of a frequent slight gap of silence at the possible
turn completion which can issue in prior speaker resuming and
extending the turn into a multi-unit one.
In the preceding, I have tried to point to several methodical
routes by which multi-unit turns or discourses can come to be.
Each concerns how a second turn-constructional unit can come
to be produced at the point at which the underlying turn-
taking organization otherwise provides for turn-transition.
But sometimes quite extended spates of talk are involved--
stories, chains of argument, long descriptions, etc. The
point about the joint, interactive achievement of discourse is
not limited to the beginning of spates of discourse--the initial
possible transition point at which the turn stays with the same
speaker. Recurrently through an extended spate of talk,
places where others could start up appear, and when others
do hot start up full utterances, there are commonly small be-
havioral tokens by which interactive management of the possi-
ble transition occasion is effected--bits of assessment or the
absence of them where they are relevant, tokens of interest,
nods, smiles, 'uh huh's, and withholding of these, gaze direc-
tion with or without mutual gaze, and the like. It is on one
class of these that I concentrate in what follows.
3. The modern literature in which bits of talk, vocalization
or related behavior are extracted from what becomes ongoing
talk by another, and are subjected to treatment in the aggre-
gate, begins with the linguist Fries (1952). Fries treated
1
to-
gether thee a following sorts of forms (1952:49): 'yes , 'unh
hunh', 'y n T , 'I see 1 , 'good', 'oh', and others of lesser fre-
quency. Others have dealt with body-behavioral versions of
this behavior, and have discussed the vocalized forms in the
course of doing so (Kendon 1967; Dittman and Llewellyn 1967,
1968). The. most common term now in use for such items,
'back-channel communication', was introduced by Yngve (1970),
and includes a much broader range of utterance types, in-
cluding much longer stretches of talk. The term 'back-
channel' has been adopted by Duncan and his associates (for
example, Duncan and Fiske 1977), together with the broad-
ened definition of the class. Duncan and Fiske (201-202) in-
clude not only expressions such as 'uh huh', 'yeah', and the
like, but also completions by a recipient of sentences begun
by another, requests for clarification, 'brief restatement' of 7
something just said by another, and 'head nods and shakes'.
78 / Emanuel A. Schegloff

Throughout this literature, two related characterizations

have been offered to deal with these bits of behavior. Ac-
cording to one, these bits of behavior are evidence of atten-
tion, interest, and/or understanding on the listener's part.
(Thus Fries 1952:49, '...signals of this continued attention...',
or Kendon 1967:44, '...appears to do no more than signal...
that he is attending and following what is being s a i d . . . ' ) . 8
A second use of such behavior proposed in this literature is
that it '...keeps the conversation going smoothly' (Dittman
and Llewellyn 1967:342), or '...appears to provide the auditor
with a means for participating actively in the conversation,
thus facilitating the general coordination of action by both
participants...' (Duncan and Fiske 1977:202-203).
I do not intend to comment extensively on this second
characterization beyond noting that once an organization of
conversation is established in which nonspeaker interpositions
are a recurrent part, their presence will be part of 'going
smoothly' or of active participation; but this does not tell us
why active participation is taken to involve this sort of be-
havior, or why the absence of such interpolations undercuts
the 'smoothness' of the conversation, if indeed it does (cf.
Schegloff 1968:1092-1093). However, it is the capacity of 'uh
huh' and cognate bits of behavior to betoken attention and
understanding which is the most common proposal about these
events taken in the aggregate, with each removed from its
context of occurrence; and it is to this sort of characteriza-
tion that the following points are addressed.
(1) The term often used in the literature to describe 'uh
huh' and similar productions is 'signal', and it is unclear what
the implications of this term are for the strength of what is
believed to be done by these bits of behavior. It is worth
noting, however, that 'uh huh', 'mm hmm', 'yeah', head nods,
and the like at best claim attention and/or understanding,
rather than showing it or evidencing it. The references to
'signals of continued attention' or 'signal. ..that he is attend-
ing and following' treat these as more than claims, but as cor-
rect claims, and this need not be the case; it i s , at any rate,
a contingent outcome, and not an intrinsic characteristic of
the behavior being described.
(2) It is unclear why any particular behavior--such as 'uh
huh' or a head nod--should be needed to address the issue of
attention, whether to claim it or to show it. Regularly, these
bits of behavior are produced when there are otherwise pres-
ent on a continuous basis sorts of behavior which are under-
stood as manifestations or exhibits of attention, such as con-
tinuing gaze direction at speaker. 9 Aside, then, from the
issue of whether 'uh huh' etc. evidence attention or claim it,
there is the issue of why attention is taken to be problematic
in the first place, in need of showing or claiming.
(3) If, for the moment, we treat the issue of attention as
having its relevance established, then it may be noted that
any instance of an indefinitely extendable set of utterances
 Discourse as an Interactional Achievement / 79

would either claim or show attention to, or understanding of,

an immediately preceding utterance by another. That i s , a
vast array of types of talk following an utterance by another
exhibit an orientation to it; accordingly, the claim that 'uh
huh' exhibits an orientation to,.or attention to, preceding talk
does not help discriminate 'uh huh' from any other talk, or
tell us what 'uh huh' in particular does or can do, and there-
fore why a participant might choose to produce it rather than
something else.
(d) If, however, we aim to understand how bits of behavior
such as 'uh huh' and the like may be taken as bearing on the
attention, interest, or understanding of their producers with
respect to the talk being produced by another, then we should
also note that 'uh huh*, 'yeah 1 , and the like are regularly
taken as betokening agreement as well. A search for the
mechanism by which interest, attention, or understanding are
exhibited by this behavior, should also deal with the apparent
exhibiting of agreement.
When 'uh huh's etc. are considered in the aggregate, then,
the characterization of the class as signaling attention, inter-
est, or understanding appears equivocal. Although it can be
argued that attention and understanding are generically rele-
vant in conversation, no ready account is at hand (when the
aggregate of cases is considered) for why these issues need
specially to be addressed, why they are addressed with these
tokens, why addressed at these particular points (if, indeed,
it is at particular points, on this account, that these tokens
are placed).
However, examination of particular occurrences of the sort
of behavior under discussion--of particular 'uh huh's, 'yeah's,
etc.--might yield answers to some of these questions. In
particular instances, for example, analysis may show that
attention was indeed problematic for the parties, and that an
'uh huh' or a nod was produced 'in response to' an extended
gaze by the speaker which appeared to solicit a sign of
attention/interest/understanding. Or, analysis may show that
certain usages by speakers regularly involve addressing the
issue of understanding in their immediate aftermath. Thus,
as described elsewhere (Sacks and Schegloff 1979), speakers
may use 'recognitional reference forms' (such as proper names)
to refer to persons they think recipients know; but if speakers
are not certain that recipients know the intended referent,
they may mark the reference form with an upward intonation,
soliciting some signal of recognition (a special kind of under-
standing); if no such display is forthcoming, further tries,
involving further clues to the identification of the referent,
are provided, with display of recognition again solicited. Re-
cipients may betoken such a recognition with 'uh huh' or
may add to this token (especially if recognition was delayed)
some demonstration of recognition, as in (1) and (2).
80 / Emanuel A. Schegloff

(1) A: Ya still in the real estate business, Lawrence

B: -*• Wah eT uh no my dear heart uh, ya know Max
Rickler h
-»• ( 0 . 5 )
B: with whom I've been 'ssociated since I've been out
here in Brentwood has had a series of urn (0.?)=
A: -*• tyeah
B: =bad experiences uhh h h h I guess he calls it a
nervous breakdown.
A: Yeah
(Sacks and Schegloff 1979:19)

(2) L: . . . w e l l I was t h e only one other t h a n than t h e uhm

tcnFords? Un
"*• Missiz Holmes Ford? You know uh
the t h e cellist?
[
W: -*• Oh y e s . She's s h e ' s the cellist.
L: Yes
(Sacks and Schegloff 1979:19)
With this b a c k g r o u n d , one can note t h a t even in t h e absence
of overt solicitation by upward intonation of some display of
recognition,
1
after recognitional
r
reference one commonly finds
'uh huh and the like, ° as in (3).
(3) Bee: hh This feller I have- (iv-) "felluh"; this ma:n.
(1T.2) t! 'hhh He h a : : ( s ) - uff-eh-who-who I have
fer Linguistics is real ly too much, "hh h=
l
Ava: •> Mm hm? Mm hm,
(TG, 198-201)
It is not that some substantial proportion of 'uh huh's etc. are
thus accounted for, but that an analytically coherent set of
cases can be assembled in this way from a series of analyses
of individual cases, the basis for the coherence of the class
being derived from the sequential environment in which those
particular tokens are produced. Although appeals to signalling
attention, interest, and/or understanding appear equivocal when
invoked on behalf of the aggregated occurrence of tokens such
as fuh huh', 'yeah', and the like removed from their particular
environments, such accounts may be viable and strong when
introduced for delimited and described cases in which the rele-
vance of these issues for the parties to the conversation at
that point in the talk can be shown. Appropriate sets of such
analyzed single cases may then be assembled to display re-
current practices, themes, structures, etc.
4. Is there nothing more general, then, that can be said
about such utterances as 'uh huh' and the like, when they
compose all of their producer's vocalization on that occasion
 Discourse as an Interactional Achievement / 81

of talking? Two observations seem to me to have sufficiently

general relevance to bear mention in this connection.
Perhaps the most common usage of fuh huh', etc. (in en-
vironments other than after yes/no questions) is to exhibit on
the part of its producer an understanding that an extended
unit of talk is underway by another, and that it is not yet,
or may not yet be (even ought not yet be), complete. It
takes the stance that the speaker of that extended unit should
continue talking, and in that continued talking should con-
tinue that extended unit. 'Uh huh1, etc. exhibit this under-
standing, and take this stance, precisely by passing an oppor-
tunity to produce a full turn at talk. When so used, utter-
ances such as 'uh huh' may properly be termed 'continuers'.
Note that the sorts of issues mentioned earlier as arising
with respect to the 'signalling attention and understanding'
accounts bear differently here.
(a) For talk-in-interaction whose turn-taking organization
makes possible-completion-of-one-speaker's-talk a place where
another can start up a next turn, it is structurally relevant
at such places for parties to display their understanding of
the current state of the talk. For example, as Sacks pointed
out years ago, participants sometimes begin a turn by produc-
ing an 'uhm' just after the possible completion of a prior turn,
then pausing, and then producing a turn, rather than just
delaying the start of their turn until they are 'ready'. They
may be understood to proceed in this fashion precisely in
order first to show their understanding of the current state
of the talk and their stance toward it (i.e. 'a prior turn is
over, it is an appropriate occasion for a next turn, I will
produce one'), in some independence of the actual production
of the turn they eventually produce. So also is it relevant
for parties to display their understanding, when appropriate,
that an extended turn is underway, and to show their inten-
tion to pass the opportunity to take a turn at talk that they
might otherwise initiate at that point.
(b) 'Uh huh's, etc. as continuers do not merely claim an
understanding without displaying anything of the understand-
ing they claim. The production of talk in a possible turn
position which is nothing other than 'uh huh' claims not only
'I understand the state of the talk', but embodies the under-
standing that extended talk by another is going on by declin-
ing to produce a fuller turn in that position. It does not
claim understanding in general, but displays a particular
understanding through production of an action fitted to that
understanding. x 1
(c) Except for the limited set of behavioral productions that
are used to do 'continuers', it is not the case that any in-
stance of an indefinitely extendable set of utterances would
achieve this outcome or do this job. Most other forms of talk
would be full turns in their own right, rather than ways of
passing the opportunity to produce such a turn, and would
82 / Emanuel A. Schegloff

fail precisely thereby to display understanding of, or respect

for, an extended unit still in progress.
The 'continuer' usage is most readily illustrated by data in
which clear marking of the end of the extended unit, or dis-
course, is provided, and until the occurrence of which the
'in-progress' character of the talk is clearly visible. Among
the ways in which such marking may be done are the several
ways of announcing, at the beginning of the unit, the sort of
thing that will be its possible end. For example, there are
story prefaces (cf. Sacks 1974) which may characterize the
sort of event the forthcoming story is about (for example,
'a funny thing happened...'), such that the unit will not be
possibly complete until such an event has been mentioned, and
may be over at the end of its mention. Or there are 'prelimi-
naries to preliminaries' (Schegloff 1980) in which an 'action-
type' is projected (like 'question' in 'Can I ask you a quesr
tion?') as that to which preliminaries are leading; the prelimi-
naries may then be developed as an extended discourse (e.g.
a description, a story, etc.) until such an action is done
(e.g. such a question is asked) as these preliminaries could
be leading up to. Several instances are given in (4) and (5).
(4) 1 B°: I've listen' to all the things that chu've said,
an' I agree with
2 you so much.
3 B°: Now,
4 B°: -> I wanna ask you something,
5 B°: I wrote a letter.
6 (pause)
7 A: Mh hm,
8 B°: T'the governor.
9 A: Mh hm::,
10 B°: -telling fim what I thought about i(hh)m!
11 (A): (Sh:::!)
12 B°: + Will I get an answer d'you think,
13 A: Ye:s,
(BC, Red: 190)
(5) 1 B: •*• Now listen, Mister Crandall, Let me ask you
this.
2 A cab. You're standing onna corner. I_
heardjuh
3 talking to a cab driver.
4 A: Uh::uh
5 B: Uh was it- uh was a cab driver, wasn' i'?
6 A: Yup,
7 • B: Now, yer standing onna corner,
8 A: Mm hm,
9 B: I live up here in Queens.
10 A: Mm hm,
11 B: Near Queens Boulevard,
 Discourse as an Interactional Achievement / 83

12 A: Mm hm,
13 B: I'm standing on the corner of Queens
Boulevard a::nd
14 uh: :m ( ) Street.
15 A: Right?
16 B: Uh, I- a cab comes along. An' I wave my
arm, "Okay,
17 I wancha I wancha." You know,
18 A: Mm hm,
19 B: Uh::m, I'm waving my arm now. Here in
my living room.
20 hhhh!
21 A: heh heh!
22 B: A:nd u h , he just goes right on by me.
23 A: Mm hm,
24 B: A::nd u h - t w o : : , t h r e e : , ( . ) about three
blocks,
25 beyond me, where- in the direction I'm going,
there
26 is a cab stand.
27 A: Mm hm,
28 B: Uh-there is a hospital, (0.?) uh, a block
(0.?) u p ,
29 and there is a subway station, right t h e r e .
30 A: Mm hm.
31 B: Uh now I could 've walked, the three or
four blocks,
32 to that cab stand,
33 A: Mm hm,

34 B: Bud I, had come out-of where I was,

right there
35 on the corner.
36 A: Right?
37 B: -»• Now is he not suppose' tuh stop fuh me?
38 A: If he is on d u t y ,
(BC, Red: 191-193)

Note that after the projection of a question upcoming, the re-

cipient of the extended talk confines himself almost entirely
(the alternatives are touched on below) to continuers--'uh h u h ' ,
'mm hmm', 'right', and the like, until a question is asked (of
the sort analyzably projected; not just any subsequent ques-
tion; not, therefore, the one at line 5 in (5). The extended
unit then being completed, and a determinate action being
called for by the question, the recipient of the discourse
addresses himself to the question. The same form of utter-
ance may be produced (for example, the 'yes' at line 13 in
(4)), but in this sequential environment it is a full t u r n ,
rather than passing one.
84 / Emanuel A. Schegloff

What will constitute the end of an extended spate of talk is

not always named or characterized as it is in the aforemen-
tioned forms; still it is regularly readily recognized by the
participants. Sometimes, however, misunderstandings occur,
and a continuer produced to display an understanding that an
extended unit is in progress and is not yet completed thereby
displays a misunderstanding, as in (6) (taken from the same
corpus of telephone calls to a radio talk show, as was the
source of (4) and (5).

(6) 1 B: This is in reference to a call, that was made

about a
2 month ago.
3 A: Yessir?
4 B: A woman called, uh sayin she uh signed a
contract for
5 huh son who is- who was a minuh.
6 A: Mm hm,
7 B: And she claims inna contract, there were
things given,
8 and then taken away, in small writing,
9 ((pause))
10 A: Mm hm
11 B: Uh, now meanwhile, about a month ehh no
about two weeks
12 before she made the call I read in, I read or
either
13 heard-uh I either read or hoid onna television,
where
14 the judge, hadda case like this.
15 A: Mhhm,
16 B: And he got disgusted an' he says 'I'- he's
sick of these
17 cases where they give things in big writing,
an' take
f
18 em, an' take Tem away in small writing.
19 A: Mhhm,
20 B: An1 'e claimed the contract void.
21 A: Mhhm,
22 B: Uh what I mean is it c'd help this woman
that called.
23 You know uh, that's the reason I called.
(BC, Gray, 74-75)
At line 21, A produces another in the series of continuers
that have helped propel B's telling; this one, it turns out, is
'mistaken', for the caller had apparently intended 'An' 'e
claimed the contract void* to be the end--perhaps hearable as
'a solution' for the woman to whose earlier call he r e f e r s . 1 2
It is worth noting that 'trouble' around the end boundary of
discourse units need not be understood as 'cognitive' or
 Discourse as an Interactional Achievement / 85

processing error; it can be the vehicle for thoroughly de-

signed interactional effects (cf. the discussion of reengage-
ment of turn-by-turn talk at emergence from a story in
Jefferson 1978).
These instances allow me to remark on several additional
points which may provide some sense of the interactional tex-
ture involved here.
1. Note that after a continuer, the speaker of the extended
unit may 'do the continuing' in various ways (and it should
be underscored that this locus of talk should be investigated
precisely for the work of 'doing continuing'). In (4), the
first continuers are followed by increments to the turn-unit
(sentence) already in progress; in (5), some continuers are
followed by increments to the prior sentence (for example,
lines 10-11); others are followed by starts of new sentences,
(for example, lines 12-13); still others are followed by what
could be counted as new sentences by virtue of their gram-
matical independence, or as increments to the prior by virtue
of their linkage by conjunction--by just such a token as marks
'continuation' (for example, lines 22-26). In this respect,
then, there is no major differentiation between sentences and
multi-sentence units or discourses; the same mechanism can
engender an elaborated version of the former or the latter.
2. Note that the bits of behavior produced by the recipient
of the extended talk vary. Two points may be advanced here.
First, even when little other than continuer usage is involved,
the tokens employed for it vary. I have referred to 'uh huh',
'yeah', etc. throughout this paper, and have not addressed
myself to the differences between these tokens. I note here
only that the availability of a range of tokens may matter less
for the difference of meaning or usage between them (if any)
than for the possibility thereby allowed of varying the compo-
sition of a series of them. Use in four or five consecutive
slots of the same token may then be used to hint incipient
disinterest, while varying the tokens across the series, what-
ever tokens are employed, may mark a baseline of interest.
Second, in some of the positions at which some sort of con-
tinuer is relevant (as may be shown, for example, by the
speaker withholding further talk until one is produced, as in
(4), lines 5-7, or (6), lines 8-11), the immediately preceding
talk may be such as to invite some sort of 'reaction' aside
from, instead of, or in addition to the continuer. And one
does find throughout extended units--especially stories--
markers of surprise ('Really?'), assessments ('oh my', 'wow',
'you're kidding', 'isn't that weird', 'wonderful', etc.) , and
the like. In the fragments I have cited, we may note the
laugh in (5) at line 21, and in (4) the laugh /assessment/
expletive at line 11. Note in the case of the latter that it
follows a selection of idiomatic phrasing that indicates 'scold-
ing' (and this has already been reported as directed to a high
political official), and its last word is delivered with a laugh
86 / Emanuel A. Schegloff

token as well. 1 3 In the case of (5), note that at lines 16-17,

the teller 'packages' the telling in a very dramatic format with
exaggerated self-quotation, which could have been designed
to engender a more forthcoming appreciation than this 'mm hmm'
provides. 11 * Note, then, that although B does continue talking
after the continuer, here she does not continue with the ex-
tended unit that was 'in progress', but shifts from a descrip-
tion of the events being told about to a description of the cur-
rent scene of the telling, using the recurrence of 'waving the
arm' as the bridge. The description appears designed to
underscore 'incongruity' and to elicit a response to it, but
even the first effort at this fails to get a response ('I'm wav-
ing my arm now.'); she then adds another (she could have
resumed the story) to underscore the incongruity even further
('Here in my living room'), to which she appends a laugh token
as well. This time she does get a response of the sort she
has apparently been after. (Note: one is tempted to write
'of a sort fitted to the character of her talk', but, of course,
it is precisely the assessment of the character of her talk
which is at issue in the sort of response A makes or withholds.
It may be suggested that the mechanism by which a series of
same continuer tokens displays incipient disinterest involves
the availability of tokens of surprise, special interest, assess-
ment e t c . , the nonproduction of which shows the recipient not
to be finding in the talk anything newsworthy, interesting, or
assessable. Varying the continuer tokens may mask the ab-
sence of other types of response token; using the same one
continuously may underscore i t . )
The general point I want to make here is that the operation
of continuers and of the other bits of behavior produced by
recipients in the course of, or rather in the enabling of, ex-
tended talk or discourse by another, is designed in a detailed
way to fit to the ongoing talk by the teller, and 'to fit' may
involve either 'cooperating' with what that talk seems designed
to get, or withholding; both of these are fitted to the details
of the locally preceding talk, and cannot be properly under-
stood or appreciated when disengaged from it. When disen-
gaged, there is no way of telling that the 'mm hmm' at line 18
in (5) is not only a continuer, but is possibly withholding a
laugh; and without that, one may not be in a position to under-
stand why the teller next abandons the story for a description
of her telling posture. In brief, disengaging the listener be-
havior from its local sequential context not only undercuts the
possibility of understanding what it is doing; it can remove an
important basis for understanding what is going on in the dis-
course itself.
The preceding discussion having ended with an account of
some of the interactional texture in particular data fragments,
it is in point to recall that the concern of this section is to
see what more general assertions can defensibly be put forth
 Discourse as an Interactional Achievement / 87

to characterize what tokens like 'uh huh' may be doing. One

I have suggested is the usage as 'continuer'.
The continuer usage rests on the observation that 'uh huh',
etc. passes the opportunity to do any sort of fuller turn at
all, on the grounds that an extended unit is already in pro-
gress. Note, however, that, were a fuller turn done, it
would be some particular type--it would be of some particular
form, and would be doing some particular action or actions.
In passing the opportunity to do a fuller turn, a participant
therefore is also passing the opportunity to do something in
particular- -the opportunity to do whatever might have rele-
vantly been done at that point. We just discussed a case in
which an 'mmhrn' was alternative to a laugh; but we clearly
cannot say that 'uh huh 1 , etc. is generally a way of with-
holding laughter, because there is no way of showing that
doing laughter is generally relevant, and if something cannot
be shown to be relevantly present, then it cannot be rele-
vantly absent, or withheld. Of course, laughter is not gener-
ally relevant; it was relevant in the case I have discussed be-
cause the other party did something to make it relevant, and
that is why one needs the local sequential environment--to see
what the other parties have done that makes some sorts of
next actions relevant, which r uh huh' may be displaying the
withholding of. The question is: are there any kinds of
actions which have some kind of 'general relevance' in conver-
sation, by which is meant that they are not made relevant by
the particulars of someone's immediately preceding talk or be-
havior? There is at least one candidate.
One kind of talk that appears to have quite a general po-
tential provenance is what has elsewhere (Schegloff, Jefferson,
and Sacks 1977) been termed 'other-initiated repair' or 'next-
turn repair initiation'. A variety of constructional formats are
used to do the job of initiating the remedying of some problem
of hearing or understanding the just prior talk of another--
several of the WH-question terms, such as 'who', 'what', etc.,
as well as 'huh', partial (and sometimes full) repeats of prior
turn, partial repeats plus one of the question words, and
others (pp. 367-369). It appears that there are no systematic
exclusion rules on the possible relevance of next-turn repair
initiation in any possible turn position. Although next-turn
repair initiation is generally withheld until after completion of
the turn in which the trouble-source occurred, it appears cor-
rect to say that such repair initiation is regularly potentially
relevant after completion of any unit of talk by another. 1 5
Its use exploits its positioning--next after the unit in which
the trouble-source occurred. If it is the case (Schegloff,
Jefferson, and Sacks 1977:363) that any talk can be a trouble-
source, then 'after any talk' can be a place for repair to be
initiated on it. Speakers can look to the moments after some
unit of talk to find whether repair on that talk is being initi-
ated; indeed, speakers who will be continuing can leave a
88 / Emanuel A. Schegloff

moment of nontalk for such repair to be initiated if the talk

just produced is to be treated by others as a trouble-source.
Then fuh huh 1 , nods, and the like, in passing the opportunity
to do a full turn at talk, can be seen to be passing an oppor-
tunity to initiate repair on the immediately preceding talk. 1 6
Note that, if tokens such as 'uh huh' operate to pass an
opportunity to initiate repair, the basis seems clear for the
ordinary inference that the talk into which they are interpo-
lated is being understood, and for the treatment in the liter-
ature that they signal understanding. It is not that there is
a direct semantic convention in which 'uh huh' equals a claim
or signal of understanding. It is rather that devices are
available for the repair of problems of understanding the prior
talk, and the passing up of those opportunities, which 'uh
huh' can do, is taken as betokening the absence of such
problems.
Further, the use of other-initiated repair as one way of
pre-indicating the imminent occurrence of disagreement
(Schegloff, Jefferson, and Sacks 1977:380) suggests why 'uh
huh's and the like can be taken as indications of agreement
with the speaker of an ongoing extended unit. For if dis-
agreement were brewing, then opportunities to initiate repair
would supply a ready vehicle for the display and potential de-
flection of that disagrement. Passing the opportunity to raise
problems of understanding may be taken as indicating the ab-
sence of such problems. It may also be taken as indicating
the absence of that which such problems might have por-
tended--disagreement—and thus be taken as indications of
agreement.
It must be noted, however, that there is a difference be-
tween this usage and the continuer usage. It was noted1
earlier that with regard to the 'current state of the talk ,
'uh huh' does more than claim an understanding, but embodies
it in particulars and acts on it. With respect to the under-
standing of, and agreement with, what a prior speaker has
said and done, 'uh huh' is merely a claim of understanding.
Such a claim may turn out to be incorrect; and passing one
opportunity to initiate repair is compatible with initiating re-
pair later. The status of 'uh huh' as an indication of under-
standing or agreement is equivocal in a way in which its status
as a continuer is not, as participants who have relied on it will
have discovered and regretted.
In this section, I have tried to formulate what appear to me
to be the only two general characterizations that can be sus-
tained when applied to singular, particular, situated instances
of vocalizations such as 'uh huh': a usage as continuer and a
usage to pass an opportunity to initiate repair. For the rest,
the treatment of them in the aggregate, separated from the
talk immediately preceding them, loses what they are doing.
Perhaps more germane to the official topic of this Georgetown
University Round Table, along with that is lost the character
 Discourse as an Interactional Achievement / 39

of the ongoing talk during which they have been produced.

Thereby our understanding of discourse is weakened. I close
with several observations on this theme.
5. Among the themes I have stressed most strongly is that,
at least in conversation, discourse must be treated as an
achievement. There is a real, recurrent contingency concern-
ing 'who should talk now'; the fact that someone continues is
an outcome coordinatedly achieved out of that contingency.
There is a real, recurrent contingency concerning what who-
ever-gets-to-talk should talk on; the fact that the same speaker
who talked before talks again and talks more of the same thing
is an outcome achieved out of this contingency (they could
have gone on to repair what preceded; they could have paren-
thesized into a comment about their talking; they could have
'touched-off into something entirely different, e t c . ) .
Once it has happened that 'a speaker continues' (for exam-
ple, 'a teller continues his story'), that appears entirely
'natural'; we lose sight of what were contingent alternatives;
they do not become 'ex-alternatives' or 'alternatives - not-
taken'; they simply disappear, and leave the achieved outcome
in the splendid isolation of seeming inescapability. For ana-
lysts, this is a great loss. Good analysis retains a sense of
the actual as an achievement from among possibilities; it re-
tains a lively sense of the contingency of real things. It is
worth an alert, therefore, that too easy a notion of 'discourse'
can lose us that.
If certain stable forms appear to emerge or recur in talk,
they should be understood as an orderliness wrested by the
participants from interactional contingency, rather than as
automatic products of standardized plans. Form, one might
say, is also the distillate of action and/in interaction, not only
its blueprint. If that is so, then the description of forms of
behavior, forms of discourse (such as stories) included, has
to include interaction among their constitutive domains, and
not just as the stage on which scripts written in the mind are
played out.
NOTES

My appreciation to the Netherlands Institute for Advanced

Study in the Social Sciences and Humanities (NIAS) for time
to reflect on some of the matters discussed here, while I was
a Fellow during 1978-1979, and to Anita Pomerantz and Michael
Lynch for useful discussion.
1. The behavioral vehicles for interaction between 'per-
former' and 'audience' may vary among various 'single speaker'
settings, but the fact of interaction is certainly not limited to
the academic lecture. Max Atkinson (private communication)
has been exploring it in political speeches in Great Britain.
90 / Emanuel A. Schegloff

2. See, for example, the paper by Marjorie Goodwin (1980).

These themes are relevant not only for discourse units, but
for 'sentences' as well. Cf. Charles Goodwin (1979).
3. Of course, not every occurrence of one of these vocali-
zations is an instance of the usage I am concerned with; not,
for example, occurrences which follow so-called 'yes/no ques-
tions' .
4. Once again, not all utterances of 'first' or 'first of all'
are list-initiating, although they do commonly project some form
of extended talk, if only by indicating that before an already
relevant action, something else is to be done, as in the follow-
ing segment:
Vic: I know who didit.
James: You know who didit,
t
Vic: Yeeah,
Vic: Ye:s.
James: Who wuzzit.
("O)
Vic: -*• First of a:: :11, un Michael came b y : , . . .
(US, 33)
5. The alternative is adding to the turn unit already pro-
duced, which can then be recompleted, as in the following:
Anne: Apparently Marcia went shopping fer all these things.
(l.TJT
Anne: Becuz uh: (0.5) Leah didn't seem t'kno:w, which
kid//d-
(Post-Party, I, 5)
6. On the possibilities discussed in this paragraph, cf.
Sacks et al. (1974:704, 709, 715).
7. Cf. note 16.
8. Kendon does describe another use of such interpolations
--a 'point granting' use.
9. In Fries' materials from telephone conversation, and in
Dittman and Llewellyn's experimental format (1967:348), the
parties are not visually mutually accessible, and this remark
is not in point.
10. As it happens, a number of Yngve's instances are of
this sort; cf. Yngve 1970:574.
11. Cf. Fragment 6, lines 20-23, and the discussion in
note 12.
12. Note that B's 'what I mean...' shows an orientation to
'having been misunderstood'. He does not go on to say he
means to help the woman and this was the reason for his call;
he uses a repair format to indicate that this is what he meant
before, which was not understood by A, as displayed by the
'mh hm' which indicates waiting for more to come. This bears
 Discourse as an Interactional Achievement / 91

on the remark earlier in the text that continuer tokens dis-

play an understanding of the current state of the talk, and
do not merely claim an understanding. It is the displaying of
what understanding their producer has which makes it possi-
ble for recipient of the continuer to find that understanding
flawed.
13. On the ways in which a laugh token can solicit a re-
sponse from a coparticipant, cf. Jefferson (1979).
14. If so, then the 'mm hm' may be used in lieu of, or to
display the withholding of, such a more forthcoming response,
a possibility further examined later. Note too that 'uh huh',
etc. can be delivered in an indefinitely extendable range of
ways; some 'uh huh's can mark surprise, appreciation, assess-
ment, etc.
15. Indeed, it can be relevant after a suspected talk unit
by another, as exchanges such as the following show:
(Silence)
A: Huh?
B: I didn't say anything.
(EAS: FN)
16. In this respect, 'uh huh', 'mm hmm', nods, and the
like are specifically alternatives to utterances such as 'huh?',
'what?', 'who?', and the like, rather than being comembers of
a category such as 'back-channel communications', as in Yngve
(1970) and Duncan and Fiske (1977). On the other hand, 'uh
huh', e t c . , in being alternatives to repair initiation, are in a
sense part of the organizational domain of repair.
In writing in the text of 'passing the opportunity to do a
full turn at talk', I appear to be joining the consensus re-
ported on, and joined by, Duncan and Fiske (1977:203) that
'back-channel actions, in themselves, do not constitute speak-
ing turns'. However, I do not believe that (a) this question
should be settled on conceptual or definitional grounds; (b)
the various components included in the term 'back-channel'
fare identically on this question; or (c) positions on the turn-
status of 'uh huh' are invariant to the occasion for the issue
being posed. I can here only suggest the basis for this
stance. Consider the fragments in (i) through (iii).
(i) D: But listen tuh how long-
[ ]
R: -*• In other words, you gotta string up the-
you gotta string up the colors, is that it?
(KC-4, 37)
(ii) R: Hey::, the place looks different.
F: Yea::hh.
92 / Emanuel A. Schegloff

K: Ya have to see all ou r new-

[ 1
D: -»• It does?
R: Oh yeah.
(KC-4,2; cf. Sacks et ah; p. 720)
(iii) 1 B: hhh And he's going to make his own paint-
ings,
2 A: mm hmm.
3 B: And- or I mean his own frames.
4 A: yeah.
(SBL: 1, 1, 12-11)

Note first that in both (i) and (ii), talk which requests clarifi-
cation (in (ii)) or repeats and solicits confirmation (in (i)),
which are two types of back-channel for Duncan and Fiske,
win out in floor fights, though, according to Duncan and Fiske,
it is a consequence of back-channels not being turns that in-
stances like these are not even counted by them as simul-
taneous turns. In my view, the issue of the turn-status of
some utterance should be approached empirically, i.e. do the
parties treat it as a turn; in (i) and (ii), clarification talk is
so treated. I believe much talk of this sort is treated by
participants as having full turn status. However, other sorts
of vocalization, such as 'uh huh', are not so treated, as Dun-
can and Fiske note, at least with respect to simultaneous talk
and its resolution.
When the issue is a different one, however, a different posi-
tion may be warranted. In (iii), for example, 'paintings' in
line 1 is an error, which is corrected at line 3 by its speaker.
This correction is undertaken after the recipient has had an
opportunity to do so, and has passed. With respect to the
organization of repair and its interactional import, it can
matter that B's self-correction follows a passed opportunity
for A to initiate repair. A silence by A in that position may
well have called attention to the presence of a repairable;
the 'mm hmr, in specifically not doing so, is doing something.
'Mm hm' is more than 'not a turn'; with respect to the repair
issue, it is very much like one.
Accordingly, it seems appropriate to me that the turn-status
of 'uh huh' etc. be assessed on a case-by-case basis, by
reference to the local sequential environment, and by refer-
ence to the sequential and interactional issues which animate
that environment.
REFERENCES
Dittman,. Allen T. , and Lynn G. Llewellyn. 1967. The pho-
nemic clause as a unit of speech decoding. Journal of Per-
sonality and Social Psychology 6.341-349.
 Discourse as an Interactional Achievement / 93

Dittman, Allen T . , and Lynn G. Llewellyn. 1968. Relation-

ship between vocalizations and head nods as listener re-
sponses. Journal of Personality and Social Psychology
9.79-84.
Duncan, Starkey, and Donald W. Fiske. 1977. Face-to-face
interaction: Research, methods, and theory. Hillsdale,
N . J . : Lawrence Erlbaum Associates.
Fries, Charles C. 1952. The structure of English. New
York: Harcourt, Brace.
Goodwin, Charles. 1979. The interactive construction of a
sentence in natural conversation. In: Everyday language:
Studies in ethnomethodology. Edited by George Psathas.
New York: Irvington Publishers.
Goodwin, Marjorie. 1980. Processes of mutual monitoring im-
plicated in the production of description sequences. Socio-
logical Inquiry 50.303-317.
Jefferson, Gail. 1978. Sequential aspects of storytelling in
conversation. In: Studies in the organization of conver-
sational interaction. Edited by Jim Schenkein. New York:
Academic Press.
Jefferson, Gail. 1979. A technique for inviting laughter and
its subsequent acceptance/declination. In: Everyday lan-
guage: Studies in ethnomethodology. Edited by George
Psathas. New York: Irvington Publishers.
Kendon, Adam. 1967. Some functions of gaze direction in
social interaction. Acta Psychologica 26.22-63.
Sacks, Harvey. 1974. An analysis of the course of a joke's
telling in conversation. In: Explorations in the ethnogra-
phy of speaking. Edited by R. Baumann and J. Sherzer.
Cambridge: Cambridge University Press.
Sacks, Harvey, and Emanuel A. Schegloff. 1979. Two prefer-
ences in the organization of reference to persons in con-
versation and their interaction. In: Everyday language:
Studies in ethnomethodology. Edited by George Psathas.
New York: Irvington Publishers.
Sacks, Harvey, Emanuel A. Schegloff, and Gail Jefferson. 1974.
A simplest systematics for the organization of turn-taking
for conversation. Lg. 50.696-735.
Schegloff, Emanuel A. 1968. Sequencing in conversational
openings. American Anthropologist 70.1075-1095.
Schegloff, Emanuel A. 1973. Recycled turn beginnings.
Public lecture at Linguistic Institute, University of Michigan,
Ann Arbor. Mimeo.
Schegloff, Emanuel A. 1980. Preliminaries to preliminaries:
'Can I ask you a question'. Sociological Inquiry 50.104-152.
Schegloff, Emanuel A., Gail Jefferson, and Harvey Sacks.
1977. The preference for self-correction in the organiza-
tion of repair in conversation. Lg. 53.361-382.
Yngve, Victor. 1970. On getting a word in edgewise. In:
Papers from the Sixth Regional Meeting, Chicago Linguistic
Society. Chicago: Chicago Linguistic Society. 567-577.
 THE PLACE OF INTONATION
IN THE DESCRIPTION OF INTERACTION
Malcolm Coulthard and David Brazil
University of Birmingham, U.K.

Paralinguistic phenomena in general and intonation in particu-

lar are areas of language patterning which have received com-
paratively little attention from linguists, who, for differing
reasons, have chosen to concentrate on segmental phonology,
morphology, syntax, and lexis. Although detailed descriptions
of intonation do exist and there is a fair measure of agreement
about the phonetic and phonological facts, at least of British
English, little work has been done on interactive significance
of intonation, though the recent work of Gumperz (1977 and
this volume) is an obvious and notable exception. Crystal
(1969) contents himself with a very detailed description of all
the phonological options without attempting to assign signifi-
cance to them. Halliday (1967) asserts that all 'English into-
nation contrasts are grammatical' and thus restricts their sig-
nificance to the language system, while Crystal (1975) argues
that 'the vast majority of tones in connected speech carry no
meaning', although a few do carry attitudinal options like 'ab-
sence of emotional involvement'.
Only O'Connor and Arnold set out to describe all intonation
choices as interactively meaningful, asserting that a major func-
tion of intonation is to express 'the speaker's attitude to the
situation in which he is placed' (1973:2). However, until
there is some set of agreed and mutually exclusive attitudinal
labels to match against the intonation choices, an attitudinal
description must be impossible: the experiment reported by
Crystal (1969:297ff.) shows the difficulties native speakers
have in matching attitudinal labels with intonation contours,
while the examples O'Connor and Arnold choose to present
undermine their claim to have managed to do so. For example,
they describe the significance of the rise-fall in relation to a
number of exemplificatory sentences. In (1), B is said to be
 The Place of Intonation in the Description of Interaction / 95

'quietly impressed, perhaps awed', whereas in (2), B is

thought to be expressing a 'challenging' or 'censorious' atti-
tude.
(1) A: Have you heard about Pat? B: "Yes!
(2) A: Why don't you like it? B: T d o .
In other examples, this same tone choice is said to convey that
the speaker is 'impressed, favourably or unfavourably--by
something not entirely expected', 'complacent, self-satisfied or
smug', or 'disclaiming responsibility, shrugging aside any in-
volvement or refusing to be embroiled'. It soon becomes evi-
dent that some, if not much, of the claimed attitudinal meaning
of the intonation contour i s , in fact, being derived from the
lexico-grammatical;;and contextual features of the examples
themselves.
Thus, although there is no dispute that speakers can vary
independently tempo, loudness, pitch, and voice quality, and
thereby alter aspects of the meaning of their utterances, any
systematic relationship between physical changes and semantic
ones has so far remained undiscovered. Indeed, Labov and
Fanshel imply that a search for systematic relationships may
be misguided when they suggest that the lack of clarity or
discreteness in the intonational signals is not 'an unfortunate
limitation of this channel, but an essential and important as-
pect of it' (1977:46). The result is that, in the absence of
any satisfying theory to account systematically for the inter-
actional meaning of intonation, those involved in the analysis
of interaction have, of necessity, taken only intermittent notice
of intonation choices, at those points where they felt they
could attach significance to them.
Perhaps the paradigm example of this approach to intonation
is the way in which Sinclair and Coulthard (1975) used the co-
occurrence of the prosodic features 'high falling intonation'
and a 'following silent stress' with now, well, OK, right, good,
to isolate those occasions when these lexical items were func-
tioning as 'frames', markers of boundary points in the ongoing
lesson.
More generally, most analysts have felt able as native speak-
ers to recognise, though not necessarily to describe, the into-
national features that mark certain declarative clauses as ques-
tioning in function and certain words as 'stressed', while
Jefferson's (1978) transcription system, which sets out to be
'one that will look to the eye how ft sounds to the ear' ( p . x i ) ,
marks also a 'continuing intonation', a 'stopping fall', and
three degrees of stress. Nevertheless, as none of the pub-
lished transcriptions have an accompanying tape and only
Labov and Fanshel provide fundamental frequency traces, it
is impossible to be sure what phonological features analysts
are focusing on, how consistently they are recognising and
marking them, how much agreement there is between analysts
96 / Malcolm Coulthard and David Brazil

on what constitutes a question-marking intonation or a particu-

lar degree of stress, and how far it is only the phonological
features they are responding to.
Thus, although no one else has explicitly stated it, it is
evident from the use made of intonational information that all
those involved in the analysis of verbal interaction would agree
with Labov and Fanshel (1977:46) that it is at the moment im-
possible 'to provide a context-free set of interpretations of
prosodic cues'.
Towards an interactionally motivated description of intonation.
For the past eight years, we have been developing a descrip-
tion which has as its final goal an account of the interactional
significance of paralinguistic cues. We do not claim, by any
means, to be able to handle the way in which all paralinguistic
features carry meaning (not, indeed, that they all have inter-
actional meaning), but we do feel we have a workable descrip-
tion of many pitch phenomena and sound principles for setting
up a description of other paralinguistic phenomena.
The first principle is that features which are acoustically on
a continuum must be analysed as realisations of a small number
of discrete units that 'form a closed set, defined by their
mutual oppositions' (Labov and Fanshel 1977:42). The second
principle is that there is no constant relationship between par-
ticular acoustic phenomena and particular analytic categories;
it is contrasts and not absolute values which are important.
These two principles are not, of course, novel and create no
problems theoretically or practically, as analysts of tone lan-
guages discovered long ago:
...tone languages have a major characteristic in common:
it is the relative height of their tonemes, not their actual
pitch which is pertinent to their linguistic analysis. ..the
important feature is the relative height of a syllable in re-
lation to preceding and following syllables. A toneme is
'high' only if it is higher than its neighbours in the sen-
tence, not if its frequency of vibrations is high. (Pike
1948:4)
A third principle is that there is no necessary one-to-one
relationship between paralinguistic cues and interactional sig-
nificances: on the one hand, as Bolinger's (1964) 'wave' and
'swell' metaphor suggests, a given pitch choice can at the very
least be simultaneously carrying both general information about
emotional state and a specific local meaning of the kind de-
scribed in detail further on in this paper; on the other hand,
certain interactionally significant signals--for instance, a re-
quest for back-channel support--may be carried by the co-
occurrence of a particular pitch choice and a particular kinesic
one, each of which singly has a different significance (Gosling
forthcoming).
 The Place of Intonation in the Description of Interaction / 97

Our final principle is to see intonation as primarily concerned

with adding specific interactional significance to lexico-gram-
matical items. Indeed, one of its major functions is to enable
the speaker to refine and at times redefine the meanings and
oppositions given by the language system. Thus we argue
that the intonational divisions speakers make in their utter-
ances are not grammatically motivated (though for explainable
reasons intonation unit boundaries frequently coincide with
major grammatical boundaries); rather they are motivated by
a need to add moment-by-moment situationally specific, into-
nationally conveyed meanings to particular words or groups of
words.
The description we are going to propose here is expressed
in terms of pitch choices, though this is almost certainly a
simplification. Intensity and durational features regularly co-
occur with the pitch choices, and it may well turn out that
the choices we describe as being realised by pitch phenomena
are being identified by hearers through associated intensity
and durational phenomena—we are only too aware of Lieber-
man's (1960) experiments on stress.
Our description sets out to account for the paradigmatic
options available at any point to a speaker and for the syn-
tagmatic structures he can build up. We have so far isolated
four systems of options--tone, prominence, key, and termi-
nation--all realised by pitch phenomena and all potentially
realisable in a single syllable; and we have four units of
structure--syllable, segment, tone unit, and pitch sequence--
of which the most important is the tone unit. The four into-
nation systems we have isolated all work within and attach
meaning to the tone unit; divisions in utterances are, as we
argued earlier in this paper, intonationally and not grammati-
cally motivated, and we agree with Laver (1970) that the tone
unit rather than the clause is 'the most likely unit of neuro-
linguistic pre-assembly'.
We see the tone unit as having the following structure:
(Proclitic segment) Tonic segment (Enclitic segment)
As this structure implies, tone units may consist simply of a
tonic segment, and many do; indeed, a considerable number
consist of no more than a tonic syllable, i.e. the syllable on
which there is a major pitch movement, as in (3).
(3) // GOOD / / , // YE£ / / , // ME / / , // JOHN //
Most tone units, of course, do consist of more than the mini-
mal tonic segment, and then the question of segmentation
arises. With the syllables following the tonic there is, in
fact, no analytic problem: even though the pitch movement
of the tone may be continued over succeeding syllables, for
98 / Malcolm Coulthard and David Brazil

reasons which we explain later, the tonic segment is considered

to end with the tonic syllable, 1 as shown in (4).

(4) Tonic segment: Enclitic segment:

/ / GOOD ness knows / /
/ / YES sir / /
/ / WET did / /
/ / JOHN ny's coming / /

However, while the final boundary of the tonic segment is

obviously unproblematic, recognising where the tonic segment
begins is a more difficult matter and depends on an under-
standing of the concept of prominent syllable.
It is not always easy, in the literature of phonology, to be
sure what significance is attached to such terms as 'stress',
'accent', 'salience', and 'prominence'. We hope we are not
gratuitously adding to the confusion by redefining two of these
terms, accent and prominence, to fit the conceptual framework
of our own description. By 'accent' we mean the attribute
which invariably distinguishes the marked from the unmarked
syllables in words like 'curtain, con'tain, re 'lc± tion, and dis-
tinguishes the lexical items from the others in a sentence like
'Tom is the 'best 'boy in the 'class. The expression 'word
accent', although tautologous from our point of view, may
serve usefully as a reminder that accent is an inherent
property of the word, which, being inherent, has no possible
contrastive significance. When we say Tom is^ the best boy in
the class we are not 'accenting' is; we are making it pitch
'prominent'. 2 'Prominence' is thus a property associated with
a word by virtue of its function as a constituent of a tone-
unit.
We are now in a position to define the scope of the tonic
segment. The presentation is simplified if we begin with some
examples in which prominence and word accent do cooccur.
(Prominent syllables are capitalised; the tonic ones are also
underlined.)

(5) / / he was GOing to GO_ / /

/ / that's a VERy TALL STORy / /
/ / i t was a WEDnesday Tl
The tonic segment begins with the first prominent syllable,
henceforth called the 'onset', and ends with the last prominent
syllable, the 'tonic'; in fact, these can be and often are one
and the same syllable. There are thus, by definition, no
prominent syllables in the proclitic and enclitic segments, as
shown in (6). If we expand the first example in (6) to (6a),
we can see that we now have four classes of syllable: un-
accented, he, was, -ing, to, a-; accented, -gain; prominent,
GO; and tonic, GO; it is interesting to speculate how far these
 The Place of Intonation in the Description of Interaction / 99

are, in fact, the four degrees of stress which Trager and

Smith (1951) proposed.
(6) Proclitic segment Tonic segment Enclitic segment
he was GOing to GO_
that's a VERy TALL STOR y
it was a WED nesday
(6a) / / h e was GOing to GO_ again //
Prominence, then, is a linguistic choice available to the
speaker independent of the grammatical structure of his utter-
ance and the accents of the constituent words' citation forms;
what then / / I S its significance //? Consider the question/
response pair in (7).
(7) Q: What card did you play?
R: // the QUEEN of HEARTS //
It is easy to see that in the response the word of is the only
word that could occupy the place between queen and hearts.
If we think of each word as representing a selection from a
set of words available at successive places, then at the place
filled by of there is a set of one. In this respect it can be
compared with the places filled by queen and hearts. If we
leave aside for the moment the slightly less straightforward
case of the, we can show the total range of possibilities as in
(8).
(8) ace
two hearts
(the) . of clubs
diamonds
queen spades
king
The speaker has a limited choice of 13 possibilities at the
first place and of four at the second, but this time the limi-
tation has nothing to do with the working of the language sys-
tem: there is no linguistic reason why the response should
not have been the prince of forks or the 17 of rubies, or any
of an enormous number of combinations. What imposes the limi-
tation is an extralinguistic factor, the conventional composition
of the pack of playing cards.
We use the term 'existential paradigm' for that set of possi-
bilities that a speaker can regard as actually available in a
given situation. This enables us to distinguish it from the
'general paradigm' which is inherent in the language system.
It is clear that at the place occupied by of, the two paradigms
coincide: there can be no possibility of selection in the
existential paradigm because there is none in the general
100 / Malcolm Coulthard and David Brazil

paradigm. We now want to argue that items are marked as

prominent to indicate that the speaker is selecting from a
range of oppositions in the existential paradigm. Thus we
can invent a context in which of can be situationally selective
—a correction of a foreigner's the queen in hearts would cer-
tainly be as shown in (9), and also a context in which queen
and hearts would not be selective and therefore nonprominent,
as in (10) and (11).
(9) // the queen OF hearts //
(10) Q: What heart did you play?
R: / / t h e QUEEN of hearts //
(11) Q: Which queen did you play?
R: // the queen of HEARTS //
In each of these examples the questioner sets up a context
which effectively removes the possibility of choice for one of
the items by indicating that he knows either the suit or the
denomination of the card. Thus the answerer's use of hearts
in (10) and queen in (11) is not the outcome of his making
any kind of selection, a fact which would probably result, in
many circumstances, in their being omitted altogether.
(12) Q: What heart did you play?
R: // the QUEEN //
(13) Q: What queen did you play?
R: // HEARTS //
Here again the existential paradigm is reduced to a set of
one by something additional to the language system. It is be-
cause shared understanding with respect to one of the variables
has already been acknowledged in the conversation that no
selection is involved. One may think, in this particular case,
of the wide range of options that comprise the general paradigm
at each of the two places being reduced by shared card-playing
conventions and then further reduced by shared experience of
the immediate conversational environment of the response.
The examples we have used so far suggest that the non-
prominent/prominent distinction is very similar to the textually
given/textually new distinction, but this is misleading; rather
we are concerned with the interactionally given. All inter-
action proceeds, and can only proceed, on the basis of the
existence of a great deal of common ground between the par-
ticipants: that is, what knowledge speakers (think they)
share about the world, about each other's experience, atti-
tudes, and emotions. Common ground is not restricted to
shared experience of a particular linguistic interaction up to
the moment of utterance; rather it is a product of the
 The Place of Intonation in the Description of Interaction / 101

interpenetrating biographies of the participants, of which com-

mon involvement in a particular ongoing interaction constitutes
only a part.
Thus one can create a situation in which items are contextu-
ally given, as in a game of cards when one person has, with-
out saying anything, put down the jack of hearts and a next
player verbalises, as in (14).
(14) // QUEEN of hearts //
Or one can create a situation in which items are available from
past experience, as in (15), when the addressee is known to
only drink coffee and the question is 'cup or mug1.
(15) // CUP of coffee //
It is perhaps more useful to see the situation not as one in
which a certain configuration of contextual features results in
the speaker choosing a particular prominence treatment for his
utterance, but rather as one in which his intonation choices
project a certain context of interaction. Thus in (15) it is
assumed and marked as assumed that coffee is not in doubt.
Key. In addition to making choices in the prominence sys-
tem, a speaker must also, for each tone unit, select relative
pitch or 'key' from a three-term system: high, mid, and low.
However, unlike Sweet (1906), we do not see mid key as the
norm for the speaker; rather we see key choices as made and
recognised with reference to the key of the immediately pre-
ceding tone unit. In other words, there are no absolute
values for high, mid, and low key, even for a particular
speaker; in fact, a given high key tone unit may well be
lower than an earlier mid key one. As we noted earlier, the
continually varying reference point is already well attested in
analyses of tone languages.
The key choice is realised on the first prominent syllable of
the tonic segment and adds a meaning that can be glossed at
the most general level as:
High key contrastive
Mid key additive
Low key equative
The way in which these intonational meanings combine with
lexico- grammatical ones is discussed in detail in Brazil et al.
(1980) but can be simply illustrated in the invented examples
in (16), where only key 3 is varied. In example (16), we see
key being used to indicate particular relationships between suc-
cessive tone units in a single utterance, but the same relation-
ships can occur between successive utterances. If we begin
with the polar options 'yes' and 'no', we quickly realise that
102 / Malcolm Coulthard and David Brazil

only when they cooccur with high key are they in opposition.
In other words, when wishing to convey 'y e s no * n o ' o r ' n o
not yes', a speaker must select high key.
(16) he GAMbled // and L 0 S T
Contrastive (contrary to expectations; i.e. there is an
interaction-bound opposition between the two)
he GAMbled // and LOST
Additive (he did both)
he GAMbled // and LogT

Equative (as you would expect; i.e. there is an

interaction-bound equivalence between them)
W0N T b e
(17a) well you ' HOME // ( (i) XM „ I WILL „
before SEVen ((ii) — // I WON'T //
In (i), ** the speaker chooses contrastive high key to mark the
choice of opposite polarity in his response; in (ii), the speaker
chooses to highlight an agreed polarity, and this apparently
unnecessary action is usually interpreted as emphatic, and
then in a particular context as 'surprised', 'delighted', 'an-
noyed', and so on. Much more usual than (ii) is (iii) in (17b).
WON'T h*» ,
(17b) well you HOME //( (iii) NO //p I WON'T //
before SEVEN l*(iv) YE£ / / p i WILL//
Item (iv) in (17b) sounds odd because the speaker is heard as
simultaneously agreeing and contradicting, or perhaps rather
agreeing with something that has not been said; the normal
interpretation would be that he had misheard. The contradic-
tion is, in fact, only made evident by the repeated auxiliary,
which carries the polarity, because interestingly, 'yes' is the
unmarked term of the pair and as a result, if the speaker does
not repeat the auxiliary he can choose either 'yes' or 'no', an
option which at times causes confusion even for native speak-
ers, as in (17c).
WONlT b e
(17c) well you HOME // ( (v) NO// (I agree I
< won't)
before SEVen// ( (vi) YES// (I agree with
your assess-
ment)
104 / Malcolm Coulthard and David Brazil

The previous day, Mrs. Thatcher had made a speech including

comments on immigration and the newscast was supposed to be
saying that in addition to other things she was going to do,
Mrs. Thatcher would make a statement on immigration that
would be 'considered', i.e. reasonable and well presented.
(19a) / / a conSIDered STATEment on immiGRAtion //
The high key choice in (19) for 'considered' marked this
statement as contrastive and the obvious contrast was with the
previous day's statement, which must therefore be seen as not
'considered' or even as 'ill-considered'.
Our examples of high key contrastivity have so far implied
that the contrast is a binary one between polar opposites, but
this is not necessarily so. In example (20), 'wife' could in
some contexts be heard as in contrast with the only other
possibility, 'daughter', and therefore as a flattering introduc-
tion (i.e. doesn't she look young?).

(20) / / M E E T el
IZabeth // Johns
But given the right context, 'wife' could be heard as in con-
trast to a whole series of other relations one might, in the
context, have assumed Elizabeth to be: his secretary, sister,
sister-in-law, friend, mistress... Thus high key marks for
the listener that an item is to be heard as in contrast but
leaves him to fill out the existential paradigm.
Low key marks an item as equative, as contextually synony-
mous; thus when the option is co-selected with 'yes' or a
repetition, the utterance does little more than acknowledge
receipt of the information, as in (21) and (22).
(21) D: Whereabouts in your chest?
P: On the heart side.
D: /

(22) A: What's the time?

B: Ten o'clock.
A: //Ten
°'CLOCK "
If a speaker reformulates in low key, he is indicating that he
does not feel he is adding any new information, but is simply
verbalising an agreement that the two versions are situationally
equivalent in meaning.
(23) A: What's the time?
B: Ten o'clock.
A:
^BEDtime "
 The Place of Intonation in the Description of Interaction / 105

(24) HE'S DEAD // and BURied //

The choice of mid key marks the matter of the tone unit as
additionally informing, and thus (24a) is slightly odd.
(24a) //HE'S DEAD // and BURied //
So is (25), from a newscast reporting how a Palestinian terror-
ist organisation had tried to invade Israel by balloon, but had
met disaster when the balloon
(25) //CRASHED // and BURNED / / .
This listener, at least, expected a low key for 'burned', indi-
cating 'as you would have expected'.
Pitch concord. It has long been accepted that some polar
questions seem to expect or even predict a particular answer
like (26i), while others like (26ii) appear to allow for
either.
(26i) You'll come, won't you?
(26ii) Will you come?
We want to suggest that, in fact, all utterances set up expec-
tations at a very general level about what will follow. In order
to demonstrate this, we need to discuss 'termination', a second
three-term pitch choice made this time at the tonic syllable.
When we look at transcribed texts, we discover a remarkable
tendency for concord between the 'termination' choice of the
final tone unit of one utterance and the 'initial key' choice of
the next; in other words, it appears that with his termination
choice a speaker predicts or asks for a particular key choice
and therefore, by implication, a particular meaning from the
next speaker. This is easiest to exemplify with questions.
In example (26i), the speaker is looking for agreement, i.e.
a mid key 'yes', and his utterance is likely, therefore, to
end with mid termination, as in (26a), to constrain the re-
quired response (remember that key and termination can be
realised in the same syllable).
(26a) A: // you'll C 0 M E // WON'T you //
B: YES // (I agree I will)
Choice of high termination for 'won't you' needs some ingenuity
to contextualize; the conflict between the lexico-grammatical
markers of a search for agreement and the intonational indi-
cation that there is a 'yes/no' choice makes it sound like either
a threat or a plea.
106 / Malcolm Coulthard and David Brazil

WON'T
A
(26b) // you'll COME // you //
Example (26ii), by contrast, naturally takes a high termination,
looking for a 'yes/no' contrastive answer, as in (26c), although
the persuasiveness of (26d) can be explained simply as the in-
tonation choice converting an apparently open request into one
looking for agreement.

(26c) A: // Will you

(26d) A: //WILL vou

COME // B: YES_ //
We can see this pitch concord working in examples (27) and
(28), both from the same doctor/patient interview.
skin
(27) D: // its ^ 5 1 // isn't i t / / P: MM//
(28) D: VERy HL r i t a t i n S you say /f p. VERy irritating^

The initial key choices in the answers have the meanings we

have already discussed, and in both we can see the first
speaker asking for or constraining a response of a particular
kind by his final termination choice. Thus, in (27), the doctor
ends with mid termination because he wants the patient to
agree with his observation, while in (28), he wants the patient
to exploit the contrastive 'yes not no' meaning of high key to
confirm what he has said. Had the doctor stopped at 'skin',
in example (27), his question would have had a very different
force, and he would again have been heard as asking for con-
firmation of a fact in doubt; but both the key and the lexical
realisation of the rest of the utterance show that what is re-
quired is agreement with a presumed shared opinion.
The pressure towards pitch concord can, of course, be dis-
regarded; the patient could have responded to the doctor's
mid key 'isn't it' with a high key 'yes' or 'mm', but telling the
doctor he was right would, in these circumstances, sound like
noncompliant behaviour, suggesting perhaps annoyance at an
unnecessary question. In example (29), the patient solves his
dilemma by selecting the predicted agreeing mid key but lexical-
ising the correctness just to be sure.

(29) D: // F I V E t i U e r ROAD// ISn't it//

P: // THAT'S corRECT// YES //
All the examples we have discussed so far have been of
pitch concord between questions and answers, but this phe-
nomenon of pitch concord now enables us to explain a paradox
in classroom discourse. On the one hand, the third, follow-up
 The Place of Intonation in the Description of Interaction / 107

item in an exchange is, as defined by Sinclair and Coulthard

(1975), optional; on the other, it is so important that 'if it
does not occur we feel confident in saying that the teacher has
deliberately withheld it for some strategic purpose' (1975:51).
One explanation of the paradox lies in the peculiar nature of
much classroom questioning--the teacher is not seeking infor-
mation in the accepted sense, as he already knows the answer,
whereas it is essential for the pupils to know whether their
answer is the one the teacher was looking for; hence there is
a situational necessity for the follow-up. There is, however,
a more satisfactory explanation. When we look at examples
like (30), we discover that very often the pupils are in a very
real sense requesting a high key, evaluative follow-up by end-
ing their response with high termination. Only when they are
confident do they end with mid termination requesting the
teacher's agreement with what they have said.

(30) T: / / W H Y W O u l d y O U W a n t t o b e STRONG//
MUSCLES
P : // to MAKE //
T . n to MAKE MUSCLES /; Y E S ; /

While high and mid termination place concord constraints on

what follows, low termination does not; it marks, in fact, the
point at which prospective constraints stop and thus occurs
frequently at the boundaries of exchanges, as in (31a,b).
(31a) D: Where abouts in your chest?
P: On the heart side
D: II

(31b) D: And how long have you had those for?

P: Well I had them a--week last Wednesday.
D: // a WEEK last

It is not unusual in certain types of interaction for an answer

to end with low termination. Example (32) is unremarkable.

(32) A: // h a V e y° U G O T the
TIME//
B: / / i t s THREE

In choosing low termination, the second speaker does not pre-

clude the first from making a follow-up move but he certainly
does not constrain him to do so as he could have done by
choice of high termination. If the first speaker chooses to
continue in the same exchange and produce a follow-up, one
option is a low key 'thanks', which one might expect if the ex-
change occurred between strangers in the street in Britain, in
which case the item would serve simultaneously to acknowledge
108 / Malcolm Coulthard and David Brazil

receipt of the information and to terminate the encounter.

(In the United States, one would expect a mid or even high
termination 'thanks', allowing for or constraining, respectively,
the 'you're welcome', 'sure', 'OK' which invariably follows.)
If the exchange had occurred during a longish interaction,
the acknowledging function could equally well have been real-
ised by an 'mm', a repetition, or an equative reformulation.
(32a)
THREE o'CLOCK
to GO "
Form and function. We can now use these observations on
the significance of pitch concord to explain one of the major
problems in discourse analysis: why some items which are
declarative or moodless in form are taken to be questioning in
function. Following example (32), we discussed the possibili-
ties for the follow-up; options we did not discuss were those
in which the speaker ends in mid or high termination, rather
than low. The exchange could have ended as in (32b), and
the message would have been 'I take "three o'clock" as equiva
lent in meaning in this context to "time to go" (indicated by
choice of low key), and I assume you will agree' (mid termi-
nation predicting mid key 'yes I agree').
(32b) A: Have you got the time?
B: It's three o'clock.
A:
"TIME to SO //
Another alternative would be (32c), and this time the speaker
is heard as both adding the information that he considers
'three o'clock' to be 'time to go' and asking for positive con-
firmation in the form of a 'yes/no' response.

(32c) // TIME to — //
We can see the difference that termination choice makes in
these two extracts from a doctor /patient interview: in (33),
the repetition with low termination is heard as exchange final;
in (34), the repeated item with high termination is heard as
eliciting.
(33) D: How long have you had these for?
P: Well I had them a week last Wednesday
D: / / a WEEK last , ^ n e s d a y / / /
D: // HOW many atTACKS have you HAD //
 The Place of Intonation in the Description of Interaction / 109

(34) D: What were you doing at the time?

P: Coming home in the car. I felt a tight pain in
the middle of the chest.
D: //TIGHT p a i n / /
P: / / Y O U KNOW // like a - // DULL ACHE //

There are two significant points about these observations:

first, although the items with mid or high termination are
initiating and in some sense questioning, the pitch movement
on the tonic is falling--not rising, as is often claimed in the
intonation manuals; in other words, it is definitely termination
and not tone choice which carries the eliciting function;
second, we are now able to identify the function of these items
through the phonological criteria which realise them and do not
need to draw on assumptions about speaker's and hearer's
knowledge or A-events and B-events, as suggested by Labov
(1972).
As philosophers have frequently pointed out, the two major
assumptions underlying orders are that the speaker has the
right to tell the listener to do X and that the listener i s , in
the most general sense, willing or agreeable to doing X. From
what has been said here about termination choices, key con-
cord, and the meanings of choices in the key system, one
would expect orders to end with a mid termination choice,
looking for a mid key agreeing //YES//, //SUREly//, //CER-
tainly//. It is thus quite fascinating to discover that most
classroom instructions, even those in a series and to the
whole class, when no acknowledgment is possible or expected,
also end with mid termination, symbolically predicting the ab-
sent agreement.

(35) FOLD your ^ ^ ;/ LOOK at the ^ ^ /; LOOK

at t h e
CEILing //
at the FLQQR „ LOOK at the pOQR//

It is also instructive, if not worrying, to realise that when

parents and teachers get cross because their instructions are
being ignored, they typically switch to high termination which
paradoxically allows for the high key contrastive refusal.

(36) P: // PUT it P-2M // C: // — //

The pitch sequence. There have been several unsuccessful
attempts to isolate a phonological unit above the tone unit and
defined as a sequence of particular tones; it is possible, how-
ever, to see tone units linked together by pitch phenomena.
110 / Malcolm Coulthard and David Brazil

We noted earlier that the particular significance of low termi-

nation is that it does not place any constraints on a succeed-
ing utterance, and we find it useful to regard all the tone
units occurring between two successive low terminations as a
phonological unit which we have called the 'pitch sequence 1 .
Pitch sequences are often closely associated with topic--
speakers appear to use a drop to low termination to signal
their apprehension that a particular mini-topic is ended. The
next pitch sequence may begin in mid key, in which case the
key choice indicates that what follows is additively related, or
topically linked, with what has just ended. In (37), the
doctor ends one part of the examination and begins another
linked one.

(37) D: //It's ^ ^ s M n / / ISn't it //

P: // MM //
D: // SCALy // LET'S have a L O O R / / / OPen your
mouth WIDE //
On other occasions, the next pitch sequence begins in high
key and the contrastive meaning serves to mark the beginning
of a completely new topic. In fact, if we now generalise, we
discover that the frames which Sinclair and Coulthard isolated
on item-specific intonation criteria are actually high key, pitch
sequence initial items following low termination, pitch sequence
final ones.
(38) T: So we get energy from petrol and we get--energy
from food
// TWO k i n d s o f m&rgy III M M t h e n „

Indeed, once one recognises them, the pitch phenomena appear

to be much more important than the lexical items in marking
boundaries; a reexamination of some of the classroom data
shows that at certain points, where on topical grounds one
felt a need for a boundary but had accepted that as no frame
occurred the teacher had not marked and probably had not in-
tended one, there are pitch marked boundaries.
(39) T : Good girl, energy, yes, you can have a team
point; that's a very good word
w e , r e USmg// E N e r & y // w e t r e
PAR

USing// / / E N e r g y / / / when a ±™tI GOES

into the
GAJ^ge / / ' "
In other words, the low termination/high key, pitch sequence
boundary, here occurring between 'energy', and 'when a car',
appears to carry the transaction boundary signal.
 The Place of Intonation in the Description of Interaction / 111

In these few pages we have tried to present principles for

describing paralinguistic phenomena, and an analysis of how
certain pitch phenomena mean. We hope to have convinced you
of the validity of both.
f r
// but ^ ^ II ° US_ / / t h e REST i s S I l e n c e / / / THANK
you / /

NOTES

1. Apart from its function in determining tonic segment

boundaries, the significance of tonic pitch movement is not
discussed further in this paper as there are marked differences
here between British and American English.
2. A full discussion of the fundamental frequency charac-
teristics of prominent syllables can be found in Brazil (1978),
and a briefer but more accessible discussion in Brazil, Coult-
hard, and Johns (1980).
3. In all subsequent examples, // marks the mid line; items
that are high or low key are printed above or below this nota-
tional line.
4. All examples are assumed to have a falling tone.
REFERENCES

Bolinger, D. 1964. Around the edge of language: Intonation.

Harvard Educational Review 34.282-293.
Brazil, D. C. 1978. An investigation of discourse intonation.
Final report to SSRC on research project HR3316/1.
Brazil, D. C. , R. M. Coulthard, and C. M. Johns. 1980.
Discourse intonation and language teaching. London: Long-
man.
Crystal, D. 1969. Prosodic systems and intonation in English.
London: Cambridge University Press.
Crystal, D. 1975. The English tone of voice. London:
Edward Arnold.
Garfinkel, H. 1967. Studies in ethnomethodology. Englewood
Cliffs, N . J . : Prentice-Hall.
Gosling, J. (forthcoming) Kinesics in discourse. University
of Birmingham: English Language Research.
Gumperz, J. 1977. Sociocultural knowledge in conversational
inference. In: Georgetown University Round Table on Lan-
guages and Linguistics 1977. Edited by Muriel Saville-Troike.
Washington, D . C . : Georgetown University Press.
Halliday, M. A. K. 1967. Intonation and grammar in British
English. The Hague: Mouton.
Jefferson, G. 1978. Transcript notation. In: Studies in
the organisation of conversational interaction. Edited by
J. Schenkein. New York: Academic Press, xi-xvi.
112 / Malcolm Coulthard and David Brazil

Labov, W. 1972. Rules for ritual insults. In: Studies in

social interaction. Edited by D. Sudnow. New York: Free
Press. 120-169.
Labov, W., and D. Fanshel. 1977. Therapeutic discourse.
New York: Academic Press.
Laver, J. 1970. The production of speech. In: New hori-
zons in linguistics. Edited by J. Lyons. Harmondsworth,
England: Penguin.
Lieberman, P. 1960. Some acoustic correlates of word stress
in American English. Journal of the Acoustical Society of
America 32.451-454.
O'Connor, J. D., and G. F. Arnold. 1973. Intonation of
colloquial English. 2nd edition. London: Longman.
Pike, K. L. 1948. Tone languages. Ann Arbor: University
of Michigan Press.
Sinclair, J. McH. , and R. M. Coulthard. 1975. Towards an
analysis of discourse. London: Oxford University Press.
Sweet, H. 1906. A primer of phonetics. Oxford: Clarendon
Press.
 TOPIC AS THE UNIT OF ANALYSIS
IN A CRIMINAL LAW CASE
Roger W. Shuy
Georgetown University and
Center for Applied Linguistics

Whenever a research project involves a large amount of data,

an obvious problem develops: 'How can we get a handle on the
data in order to find its salient aspects?' Vygotsky (1962)
warned that the most critical aspect of data analysis is finding
the appropriate unit of measurement. For years, language re-
search has suffered from the existence of its analytical rou-
tines. Because of the development of mean length of utterance,
T-units, word frequency, and other commonly used measures
of language, such routines were applied to masses of data,
whether they needed it or not. We also often misused the
more traditional analytical approaches of our field. But, if
the body of data under investigation consists of a half-hour
of tape-recorded conversation between suspected criminals,
and if the purpose of the analysis is to help determine whether
or not criminal activity has taken place, even a traditional
phonological analysis may not be the best approach to take.
If we follow Vygotsky's advice to find the appropriate unit of
analysis, we need to begin with the reason for the analysis
rather than the inventory of known analytical routines. We
ask ourselves, under such circumstances, 'What is the best
unit of analysis I can find in order to address the problem I
am attempting to solve?'
In this paper I describe how topic analysis was selected as
a crucial unit of analysis for the evidence presented by the
defense in the case of the State vs. Arthur Jones in the Fall
of 1979. * But first it is necessary to describe this case
briefly.
The State vs. Jones case. Arthur Jones is a wealthy busi-
nessman whose life has been plagued with court litigation.

113
114 / Roger W. Shuy

Jones married a woman named Gwendolyn, with whom he lived

for several years before they began to have marital difficulties.
During the ensuing divorce proceedings, which were bitter and
extended, a man who had worked for the Jones-owned business
went to the FBI with the story that Jones had asked him to
find someone to murder Gwendolyn along with the judge in
their divorce trial. This employee, Roy Foster, a salesman,
was told that he should engage in some linguistic fieldwork.
The FBI attached a Nagra tape recorder to his body and told
him to surreptitiously record Jones on tape actually soliciting
these murders. The resulting tape, of course, could then be
used as evidence against Jones.
Roy Foster then arranged a meeting with Arthur Jones in
the city parking lot. Foster attempted his linguistic fieldwork
with the goal of eliciting from Jones the agreement to solicit
these murders. The conversation lasted about 20 minutes.
Two days later, a second meeting between Foster and Jones
was also arranged. This 10-minute conversation was also re-
corded with the surreptitious body tape recorder, but with a
new twist. The FBI also video taped this event from a camera
hidden in a van in the same parking lot.
The District Attorney was satisfied, from these tape record-
ings, that Jones had actually solicited the murder of his wife
and the judge. An indictment was made and Jones stood trial
in a local state court in the following autumn. The trial ended
in a hung jury (8 against and 4 in favor) and a re-trial was
scheduled for the following winter.
Topic analysis. Elsewhere (Shuy 1981) I have described the
agonies one goes through when confronted with an opportunity
to participate in a court case. After listening to the tapes and
reading the state's transcripts, I finally hit on a way to ad-
dress the language issues involved. The basic linguistic ques-
tions were clear: exactly what did Jones agree to do? What
were his intentions? What were Foster's intentions? In order
to address these questions it was clearly necessary to see them
in relationship to the specific conversational topics in which
they occurred.
In such analysis of topic, I followed the general outlines of
topic-comment analysis as described by Chafe (1972) and Kates
(1980). That is, the structure is not defined by the gram-
matical relations of the terms, nor by the semantic structure.
Kates observes (1980):
In general, something is treated as a topic, whether it is
linguistically expressed or not, when it is taken as an in-
tentional object or structure (invariant) of some type. A
comment refers to some way in which that object can or
should or will or does, etc. , appear or manifest itself.
 Topic as the Unit of Analysis in a Criminal Law Case / 115

By mapping the topics of an interaction, therefore, one can

obtain a macro picture of one aspect of the structure of the
conversation which highlights the cognitive thrust of its direc-
tion.
Such a mapping is particularly important for the purposes of
a court case such as State vs. Jones for two reasons. For
analysis purposes, it divides the conversation into meaningful
units of analysis in which propositions can be seen in direct
relationship to their responses, as we shall see. For the pre-
sentation to the jury, topic analysis provides a clear guide to
the conversation at a macro level, to enable the untrained lay-
men of the jury to see holistically and not be bogged down
with memory lapses or details. Furthermore, as Greenfield
(1980) has observed, ' . . . i f we can establish directedness and
termination in the presence of the goal, we have established
a particular intention'. The very existence of the topics
establishes this kind of directness, and clear evidence of
resolution establishes termination. A criminal court case is
little more than the establishment of intentions and the evi-
dence of having carried them out. The issue in the Jones
case, it became clear, was one of determining what was in-
tended and what was accomplished. Intentionality is a slippery
thing, verifiable perhaps only in the mind of the person who
intends. Traditional court procedure is to ask defendants to
tell what their intentions were, but this i s , at best, self-
report data in which it is not reasonable to expect anything
but bias. When the actual conversation is recorded, however,
there is more to work with. My question became, 'How can
the structure of taped conversation help the jury infer the in-
tentions of the speakers? 1
Although my topic analysis of the Jones tapes was begun
initially to serve my own needs in determining the content of
the conversation, it soon became apparent that this same topic
mapping, if displayed properly, could prove equally beneficial
to the jury's understanding of what was actually going on.
Identifying topics. One question remained to be solved:
how could I know for sure when one topic ended and another
began? Several methods suggested themselves. First, there
is the clear and uncontestable change of subject focus. The
two separate conversations presented as evidence in this case
included talk about many things, including:
Doing Gwendolyn and the judge
Fred
Arrangements for future meetings
Some unidentified objects in their possession
The trunk of the car
Jones' sunglasses
A local activist group
The original plan
116 / Roger W. Shuy

Favors of friends
Their health
Jones' divorce
A car part that was broken
These topics were clearly separable from each other on the
basis of logic and content. The responses of the other
speaker, the one who.did not introduce these topics, however,
were not as easy to categorize. I chose to use the term,
topic 'response', to characterize the speech (or absence of
speech) once one of the foregoing topics was introduced. The
more commonly used term, 'comment' (Kates 1980; Keenan and
Schieffelin 1976) was not as descriptive for two reasons.
First, 'comment' did not effectively differentiate 'response'
from 'resolution', a distinction which has tremendous signifi-
cance in this case. That is, not every response was a resolu-
tion, and it is in the matter of resolution of topic that a court
case rests. Second, the term 'comment' is not as forceful,
familiar, or clear to a jury as is 'response'.
In addition to a content definition of topic, there are also
structural evidences for topic introduction or change. These
include intonation changes (Sacks, Schegloff, and Jefferson
1974), pauses, and either individualized or general topic-
marking devices at the onset of a speaker change (Keenan
and Schieffelin 1976), such as:
Hey, I got something here. + new topic.
Uh... + new topic.
Hold it. Wait a second. + new topic.
Now. + new topic.
Just one problem. + new topic.
Well, + new topic.
Uh, What else? + new topic.
This combination of internal cohesion of the subject matter
and prosodic and topic-marker phrases enabled me to be
reasonably certain of the topic units of these conversations.
Topics in the State vs. Jones case. At this point it would
be well to show what the topics of the conversations looked
like when displayed to the jury in the form of a chart. In
the first 20-minute tape there were 22 topics, as Figure 1
indicates.
By displaying these topics by speaker, we can gain insights
into which speaker was dominant in topic introduction and
which speaker recycled which topics. Color was used on the
chart presented to the jury (through expert witness testimony)
to enable them to visualize the differences between the topic
types (transitions, proposals, Fred, and details) and to mark
the topic recycling. In addition, key words from the actual
transcript were also written underneath the topic boxes to
 Figure 1. Topic introduction and recycling, Jones case, first tape.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 51
•g.
Jones P |Tr D P o*

lTrl P
in
Foster Tr P Tr F Tr Tr P P F P Tr F P |Tr Tr |D
Tr| = Iransitions (in brown) C
jp] = Proposal (in red) 3

|FJ = Fred (in green)

>
B] = Details (in blue)

Figure 2. Topic introduction and recycling, Jones case, first tape, topics 8-12. 3
0)
n
Topics 10 11 12
Jones Prop
Original plan a>
Foster Prop Fred Prop Tr n
u
(A
Price on Gwendolyn Alibi Do her first? Divorce
Talk to shooter
118 / Roger W. Shuy

remind the jurors of that portion of the conversations which

they had actually heard on the tape. Figure 2 provides a
close-up picture of a portion of that chart (topics 8-12) for
illustrative purposes.
In the 10-minute April 20 tape, there were 13 topics, as
Figure 3 indicates.
The second conversation. These charts, as I have indi-
cated, were actually used as part of my expert witness testi-
mony in this case. The defense attorney asked me questions
about the significance of the charts. Perhaps the best way to
illustrate this procedure is to quote directly from the proceed-
ings of my direct examination by the defense attorney during
the trial.
Attorney: Let me ask you this, Dr. Shuy: What--
generally, what is the purpose of a linguist in the
analysis of breaking down the tape recording of a con-
versation and the topics?
Shuy: One major purpose is to show the balance of
the introduction to topics, or the imbalance in this case,
of the introduction of topics. In normal conversation you
have a balance of topics introduced. Speakers are obliged
in conversation to introduce approximately equal numbers
of topics. One way to determine the nature of conver-
sation, the structure of conversation, is to determine not
only who introduced what topics, but what the equilibrium
is--what is the balance of topics introduced by each
speaker.
Attorney: And is there a balance of introduction of
topics in this tape?
Shuy: No, it's not balanced.
Then after questioning me about what topic recycling meant,
the attorney asked:
Attorney: Is there any observation you can make as
an expert in linguistics to the imbalance of the number
of topics that were introduced by Mr. Foster (18) as
opposed to the four that were introduced by Mr. Jones?
Shuy: Yes. One of the characteristics of a conver-
sation is that once a topic is introduced in a conversa-
tion, it does not tend to be reintroduced or recycled
over and over again if it is resolved.. .one doesn't keep
bringing up the same topic over and over again if the
listener does not respond to it.
This line of questioning went on for about one court day,
during which time I branched out from the macro picture
which was set by the topic introduction charts to the various
segments of each topic which were germane to the case. I
Topic as the Unit of Analysis in a Criminal Law Case / 119

Leave-
taking
three
days
Det.
Wait

Fred

Fred
Tell
• 5 *
ox u bo
te5 £g
c
3

lue
C
X) <w
&
•S
"C

•a
4-*
73
0)
SH
Q HI
II II
CO
Trunk

g ox
Fred

+->

Q
bo

u
o
2
chang
Could
plans

g oX &
C
co
U £
o
X 3
bj.

O •3S
What
NOW

o
120 / Roger W. Shuy

pointed out that a pattern could be seen by noting who intro-

duced which topics, as in Figure 4.
Figure t. Summary of topic types by speaker,
Jones case, first tape.
Foster Jones
Transition 8 1
Proposal 6 2
Fred 3 0
Details 1 1
18 1

This simple breakdown for the jury made it clear that Jones
was not in control of this conversation and that Foster was
dominant. The defense position, of course, was that if Jones
was the instigator of a plot to kill his wife and judge, isn't it
odd that he never brought up the subject?
Transition topics. From the transition topic control of Foster
it also is clear that he, not Jones, took responsibility for pro-
viding the conversational glue, the transitional, less substan-
tive topics which provided spacing for his recycling the pro-
posal topics which were his conversational agenda. One initi-
ates conversational transition to facilitate progress on a larger
agenda by offering small breaks from it or to uphold the social
obligations of conversation. The first way is more cognitive
and planned; the second is more social and spontaneous. The
single transition topic which Jones introduced in the first con-
versation came in the last third of the conversation, likely out
of a realization that he had not been holding up his end of the
conversation. Foster's transition topics, however, were more
planfully timed to move his agenda along. Jones had responded
to his substantive proposal topics in one of four ways: with
total silence; with token noises such as uh, well, or coughs;
by changing the subject entirely; or by responding to the
least significant part of a proposal topic. Foster reacts to
Jones' minimal responsiveness by moving the topic to less
threatening ones, trying to make Jones feel more comfortable
and perhaps even utter a few words. His theory of transi-
tion topic introduction seems to parallel that of a salesman who
is trying to get a customer to respond to his pitch. If the
customer does not bite on one product, the salesman tries
another line to elicit a response. When I called this phe-
nomenon to the attention of the attorney, he reacted enthusi-
astically, for Roy Foster's occupation was, in reality, that of
a salesman.

Details topic. In the same conversation, Jones introduced

one details topic (19), interestingly enough bounded on each
 Topic as the Unit of Analysis in a Criminal Law Case / 121

side by Foster transition topics (18 and 20). This details

topic, harmlessly enough, involved whether or not it was all
right for his secretary to take the message if Foster should
call him back.
Fred topic. The topic of a person named Fred, Foster's
boss at the company which Jones owns, was introduced only
by Foster. It appears that Foster had to miss work occasion-
ally in order to do personal work for Jones, and Foster intro-
duced this topic as a means of asking Jones to supply an alibi
for him to Fred. The topic is introduced three times (4, 10,
and 15). On the surface, this topic did not appear to be
favorable to Jones' innocence but, from my position, it had to
be noted for what it was. As it turns out, however, unknown
to me until after my testimony, Jones had testified that he had
hired Foster earlier to keep an eye on his wife's activities be-
fore and during their divorce hearings. Jones claimed that
this topic referred to that set of events.
Proposal topics. Jones did introduce proposal topics twice.
Rather than convicting him, however, these instances actually
argued for his defense. The proposal which he introduced
first as topic 9 and recycled as topic 22 was actually an alter-
native to Foster's proposal topic of 'doing the judge and doing
Gwendolyn'. The substance of Jones' alternative proposal was
that they should 'go back to the original plan'. To this day
I do not know what the original plan was. As an expert wit-
ness who could not know the facts external to the data pre-
sented as evidence, this information was not available to me.
If the prosecution were to discover what the original plan was,
they would have to get it from Jones himself, not from my
analysis.
The second tape. The second tape yielded a similar pattern.
Jones introduced 5 of the 13 topics, all nonsubstantive. He
introduced the greetings, mentioned the objects they held in
their hands, one transition topic, and two details topics.
Foster, on the other hand, introduced three substantive topics
about the presumed killer he had hired, two topics about their
activity at the trunk of their car, one reference to the objects
in their hands, one fleeting reference to Fred, and their
leave-taking.
Responses to topics. Once the topics have been clearly
identified and presented, it becomes possible to examine the
really crucial aspects of a court case of this type, the re-
sponses. One should keep in mind that the jury had no ex-
perience in listening to a mass of conversation largely about
killing people, and separating who said what to whom about
what topics. It is not accidental that over 99 percent of all
court cases involving tape recorded evidence result in
122 / Roger W. Shuy

convictions (Fishman 1973). There is said to be predisposition

on the part of jurors that if persons are taped they are most
likely to be guilty. There is also the contamination factor to
be dealt with. That i s , if two people are talking and one of
them talks about murder, it is likely to be overgeneralized
that both of them are talking about murder.
After a topic is introduced, it is expected of the conversa-
tional partner to respond. There are several alternative
possibilities of response available to such a partner: to add
something to the topic during his/her turn; to deny the
premise or facts of the topic; to ignore the topic with silence
or irrelevance; to respond to some, but not all, aspects of the
topic; to defer or change the topic to something else; or to
request clarification or amplification of the topic.
Participant responsiveness in a conversation can be an indi-
cation of intentionality at a level much deeper than the semantic
meaning of the words themselves. A participant tends to be
responsive when interested in or knowledgeable about the topic
being discussed. Once the conversation has been segmented
into topics, one can observe an interesting pattern in Jones'
responsiveness. If we discount transition topics on the
grounds that they call out only social responsiveness, not
cognitive responsiveness, we are left, in the first tape, with
only Fred's topics: proposals and details. The details topics
are all resolved when introduced. They are recycled only for
additional details. Jones' responses to the topic of Fred are
more animated and fuller than are his responses to the pro-
posal topics. Jones appears to be interested in Foster's ab-
sence from work more than in any other topic introduced by
Foster. This may be from relief that he is finally on a topic
about which the stakes are lower, or it may be because he
understands the topic better than the proposal topic.
Figure 5 is a representation of the topic resolutions of the
first conversation which I prepared for the jury during my
testimony in this case. Notice that the point of this chart is
to isolate for the jury the responses to the more substantive
topics of this tape: proposals and Fred. Although the Fred
topic turns out to be not terribly substantive, it still serves
a very useful purpose: that of contrasting with Foster's pro-
posals by showing what a clear resolution can look like. Jones
finally commits himself to helping Foster with his alibi to Fred
in topic 15, the third time it is introduced. In sharp contrast
to this resolution is the lack of commitment by Jones to Foster's
repeated recycling of his proposals about killing people. Jones
avoids resolution by using the strategies of incomplete re-
sponses (false starts and incomplete utterances); unclear re-
sponses (well... ,uh..., e t c . ) ; requests for clarification; off-
topic comments; presumed hearing failure; and responses to a
minor aspect of the topic not germane to its substance. In
short, there was no commitment, no resolution. In topic 17,
 Figure 5. Topic resolution, Jones case, first tape.

Foster's topics:

Fred/alibi 10 15 Resolved:
Jones
commits r
3"
<D
Jones: No response Requests 'I will C
clarification this time1 3

Do judge,
Gwendolyn Unresolved:
Jones never 3
commits
SL
Jones: Response Response
incomplete or incomplete, «'
unclear unclear.
'You can't Avoids 3*
keep on be- answering. 0)
ing absent' Requests
clarification. n
Answers minor
parts of
questions.
Jones' topic:
Original Resolved:
plan Foster
commits
n
Foster: Requests Suggests 'Done' V)
clarification hypothetical CD
conditions.
Changes
subject.
124 / Roger W. Shuy

however, there are some anxious moments.' Following the

state's transcript, in reference to the supposed hired killer,
Jones responds:
Jones: Well, he's not going to go wandering in there
if there's anybody else there. He's gonna—He'll know
what he's doing better than that. Do the judge, and
then his wife, and that would be it.,
On the surface these words appear to negate my assertion that
in topic 17 Jones does not resolve the topic. The reason it
does not negate my assertion is that the state's transcript of
the tape, provided by the FBI, is in error. Foster's state-
ment preceding those Jones remarks is as follows:
Foster: I mean I'll go along with whatever you say but
uh, what...are you gonna do if the son-of-a-bitch wants
to do... You know, They.. .They're awful close together.
Uh if he grabs that judge up and puts him in his car...
There's gonna be a hell of a stink but not near as much
as if he left that son-of-a-bitch bloody and bleeding in
his driveway...
Jones: Well, he's not going to go wandering in there
if there's anybody there. He's gonna--He'd know what
he's going to do better than that. He'd do the judge,
and then his wife and that would be it.
Careful listening to this tape revealed one deleted He'd in
Jones' speech transcript, converting a conditional into an im-
perative. The media cited the incorrect transcript for weeks
before and during the trial. The context of Foster's question
is clearly hypothetical. Jones' answer is equally hypothetical.
He goes on immediately after this, in fact, saying:
Jones: Or if he...or he might catch the judge coming in.
Since Jones' response is to a hypothetical case, not a spe-
cific one, it cannot be considered a resolution to Foster's pro-
posal to hire someone to kill Gwendolyn and the judge. For
this reason it is possible to say that there was no resolution
to this topic on this tape.
In contrast, Jones' proposal topic, 'go back to the original
plan', is resolved by Foster in topic 22 of the first conver-
sation. After responding by requesting clarification, com-
plaining', suggesting hypothetical conditions, and changing
the subject in topics 9, 17, and earlier in 22, Foster finally
commits to Jones' proposal with the word 'Done' at the end of
the conversation.
What this leaves us with, then, is topic resolution for
Foster's topic of Fred and for Jones' topic of the original plan.
 Topic as the Unit of Analysis in a Criminal Law Case / 125

What remains unresolved is Foster's topic of doing the judge

and Gwendolyn.
Other aspects of topic analysis. It is almost always the case
that an analysis which attempts to segment one aspect of the
data out of convenience or necessity runs risks of artificiality.
Here it is clear that the separation of topic from response is
difficult if not impossible. The process is more cyclical than
linear, and the reductionist fallacy is as present here as it is
in the current vogues in educational practice. The part does
not isolate from the whole; it only explains it. Nevertheless,
as a procedure, the isolation of topic served a very useful
end both for the analysis and for the presentation to (actually
the teaching of) the jury. Other aspects of the analysis used
in the defense of Arthur Jones could be presented separately,
but they really can be seen more clearly as part of the topic-
response patterns found in these conversations. Pauses, for
example, were shown to be significant aspects of Jones' re-
sponses to Foster's topic initiations. Interruptions played a
crucial role in the jury's understanding of both topics and
responses. Lax tokens were seen to be useful indications of
uncertain responses, and the analysis of uh-huh, all right,
and OK as place holders also proved to be very important
indications to linguistic laymen that not every response which
seemed to be positive actually was.
Conclusions. Several points of interest can be deduced from
this experience of utilizing topic as a unit of analysis in a
criminal law case. Perhaps the most crucial is that the prob-
lem being addressed dictated the selection of the unit of
analysis. The problem was to determine what the structure
of the conversations could tell us about the meaning of the
event. It was up to the jury, of course, to decide for them-
selves exactly what had taken place from what was said on the
tapes. It was not appropriate for me to tell them what the
intentions of the speakers were or what their words actually
meant. The court agreed that this was the sole province of
the jury to determine. My role, quite differently, was to help
the jury understand the structure of the conversation as a
clue to the possible intentions of the speakers and to help
them distinguish exactly what was said by whom. This is
partly linguistic analysis and partly teaching, which was
largely accomplished through visual displays in the form of
charts of the sort noted here.
Topic analysis helped the jury understand who controlled
the conversation and who produced minimal and often evasive
responses to those topics. It helped them understand the
separate agendas of Foster and Jones by noting their topic
recycling patterns. It helped them obtain a holistic or macro
picture of the entire conversation to use as a reference point
into which micro elements of the conversation could be
126 / Roger W. Shuy

appropriately placed. It helped them separate the substantive

aspects of the conversation from the less substantive ones.
Response analysis helped the jury determine exactly what
was avoided, deferred, or rejected. It was necessary for the
jury to understand the difference between the social require-
ments of conversation which require some sort of response,
even if evasive, and the cognitive requirements which require
positive or negative resolution. Conversational partners are
pressured'by both types of requirements, but the social re-
quirements offer constraints based on context and intentions
which are not easily recognized by the linguistic layman.
Response analysis sorted out for the jury the topics Jones
was willing to elaborate on or resolve from those which he was
unwilling to advance or commit to.
Topic and response were not the only linguistic analyses
used in the Jones case, but they turned out to be extremely
useful in helping to meet the major goals of my expert witness
testimony: to determine the structure of the conversations.
NOTE

1. For obvious reasons, the names of all participants in this

case have been changed.
REFERENCES

Chafe, Wallace. 1972. Discourse structure and human knowl-

edge. In: Language comprehension and the acquisition of
knowledge. Edited by Roy O. Freedle and John B. Carroll.
Washington, B.C.: V. H. Winston.
Fishman, Clifford. 1973. Wiretapping and eavesdropping.
Greenfield, Patricia Marks. 1980. Towards an operational and
logical analysis of intentionality. In: The social foundations
of language and cognition: Essays in honor of J. S. Briiner.
Edited by David Olson. New York: W. W. Norton.
Kates, Carol A. 1980. Pragmatics and semantics: An
empiricist theory. Ithaca, N.Y.: Cornell University Press.
Keenan, Elinor Ochs, and Bambi Schieffelin. 1976. Topic as
a discourse notion: A study of topic in the conversations
of children and adults. In: Subject and topic. Edited by
Charles Li. New York: Academic Press.
Sacks, Harvey, Emanuel Schegloff, and Gail Jefferson. 1974.
A simplest systematics for the organization of turn taking
for conversation. Lg. 50.
Shuy, Roger W. 1981. Linguistics in the courtroom. Paper
presented at Conference on Linguistics and the Humanities,
Arlington, Texas. Mimeo.
Vygotsky, Lev S. 1962. Thought and language. Cambridge,
Mass.: MIT Press.
 BUILDING STORIES:
THE EMERGENCE OF INFORMATION STRUCTURES
FROM CONVERSATION
Catherine E. Snow and Beverly A. Cold field
Harvard University

One way of thinking about the nature of the development in

children's language ability between the ages of 2 and 3-1/2
years is as a reorganization of the relationship between syn-
tactic and discourse skills in the service of transmitting infor-
mation. From this point of view, the following significant de-
velopments are characteristic of the growth in language be-
tween 2 and 3-1/2 years of age.

1. Semantic relationships that had been encoded by simple

intraspeaker apposition come to be encoded syntactically.
Thus, the young 2-year-old typically says things like (1) and
(2), or engages in conversations like (3), in which the se-
mantic relationships are derivable from context and world knowl-
edge, not made explicit in the utterances.

(1) sweater
chair
(2) mommy
cookie
eat
(3) Mother: Do you want a banana?
Child: Lunch.
The 3-1/2-year-old, on the other hand, typically says things
like (4), (5), and ( 6 ) , encoding the semantic relationships
syntactically within one utterance.

(4) sweater on the chair

127
128 / Catherine E. Snow and Beverly A. Goldfield

(5) Mommy's eating a cookie.

(6) I want a banana for lunch.
2. Information structures that at 2 years are expressed in
a sketchy, incomplete way, are by 4 years of age more com-
plete, more decontextualized, and less egocentric presentations
of the information.
3. Information structures that had been largely idiosyn-
cratic, dependent upon the child's own experience related to
the topic, become conventional. By 4 years, children can en-
code information in structures conventional enough to be under-
stood even if the information is new to the listener.
In this paper, we hope to demonstrate that the three indices
of language development we have sketched are related, and that
they emerge in a very concrete way from the child's history of
talking about specific things, again and again, with a knowl-
edgeable adult; the data base analyzed to support these claims
is a series of conversations between a mother and her child
about the pictures in one book. The conversations occurred
on 13 separate occasions, in the course of 11 months, starting
when the child was 2;5.18 (that is, 2 years, five months, 18
days). We presume that the types of conversation and the ef-
fects of recurrent discussion of a single topic on children's
language growth presented here are possible in a wide variety
of contexts, certainly not limited to book-reading; nonetheless,
for the purposes of clarity, analysis and discussion in this
paper are limited to situations of talking about events in pic-
tures .
The knowledgeable adult can be hypothesized to fulfill a num-
ber of functions during the conversations with the child that
produce the developments sketched earlier. These functions
can be identified as the product of two dichotomies (see Table
1), the first relevant to the information structure and the
second relevant to the time at which the adult's effect is felt.
The first dichotomy is the distinction between (a) the cate-
gories of information that need to be included for an adequate
representation of some set of similar events and (b) the spe-
cific informative content relevant to each specific event. The
second dichotomy is the distinction between (a) synchronic and
(b) diachronic effects of the adult partner in the conversations.
These two sets of distinctions are discussed in greater detail
in the next two sections. Table 1 presents the ways in which
they interact to provide four categories of functions which the
knowledgeable adult can fulfill in conversations with children.
The structure of information. A good description, a good
story, or a good argument has a certain structure--a skeleton
of categories or slots around which the details of any particu-
lar description, narrative, or argument are built. The notion
of an abstract structure which underlies the organization of
 Building Stories / 129

any narrative is the basis for proposals such as story gram-

mars (see, for example, Mandler and Johnson 1977, Rumel-
hart 1975) and, without committing ourselves to any particular
claim about the nature of such a structure, an assumption of
its existence is crucial to our understanding of the develop-
ments in children's language skills sketched earlier.
Table 1. Hypothesized functions of the knowledgeable adult
in the child's presentation of information structures.
Synchronic Diachronic
Categories of Adult uses questions to Adult models the cate-
information: elicit informative con- gories that are rele-
tent, so that an ade- vant for understand-
quate information ing a particular pic-
structure is presented. ture. Child acquires
Child has content, but these for later use.
does not know the
relevant categories.
Informative Adult answers the Adult models the in-
content: child's questions, pro- formative content
viding the informative relevant to a particu-
content relevant to a lar picture, and in
particular picture. subsequent discus-
Child knows the cate- sions, the child pro-
gories, not the content. vides that informa-
tive content.
This skeleton of the information structure is referred to in this
paper as 'categories of information'. The assumption is that
categories of information are quite general, relatively abstract,
and that some small set of them is adequate for dealing with a
wide variety of specific pictures or events.
The categories of information contrast with the 'informative
content', which is specific to a particular picture or event.
The informative content is the flesh hung on the skeleton--the
concrete understanding, meanings, lexical items, perhaps even
syntactic structures needed to talk about a particular picture
or event.
With reference to Table 1, it is proposed that a knowledge^
able adult can elicit and model the categories of information and
the informative content independently of one another, and in
separate utterances. Questions, for example, provide evidence
about the categories of information appropriate to the picture
under discussion, not about informative content, whereas re-
sponses to questions normally provide informative content after
the category of information has been established. (Responses
to questions may also provide evidence that a chosen category
of information is inappropriate, e.g. 'Why is he doing that?'
'Just because. No reason.')
130 / Catherine E. Snow and Beverly A. Gold field

Synchronic and diachronic effects. In the course of talking

to a child about a picture in a book--especially if the picture
is complex, action filled, and intrinsically narrative, as was the
case for the pictures analyzed here--the knowledgeable adult
can provide structure and add informative content in such a
way that a good, complete, and conventional information struc-
ture emerges from the conversation. The skillful adult elicits
from the child all the relevant informative content, but weaves
that (typically incomplete or unstructured) content into a con-
versation whose sum total constitutes a good information struc-
ture. This process is referred to here as the synchronic func-
tion of the adult. It was abundantly in evidence in the con-
versations analyzed, but is not discussed in detail here. The
synchronic function has been described, for a variety of situ-
ations, in other studies. Most closely related to this one, for
instance, McNamee (1979) documented the way in which a
kindergarten teacher elicited knowledge about a story from a
child, knowledge which the child possessed but could not dis-
play without the interactive support of the knowledgeable adult.
Wertsch (1979) has described a similar adult role in support-
ing children's problem-solving in a puzzle task; the adult
analyzed and structured the task so that the child's knowledge
and skills could be deployed to solve it. With younger chil-
dren, Scollon (1979), Snow (1977, 1978), Shugar (1978), and
others have described how adults structure conversations so
that children are effective conversational partners as well as
good providers of information. Michaels (1981) has described
how a match between child and adult in the way they organize
their information facilitates the interactive construction of good
information structuring. The synchronic function of the knowl-
edgeable adult is very important, both for parent-child and for
teacher-student interaction, but this paper concentrates on the
as yet unreported and unanalyzed diachronic function.
By diachronic function, we mean the effect of the adult's
elicitation and modelling on the child's performance at a later
time; in other words, that the child's presentation of the in-
formation structure looks like the adult's presentation of a few
weeks or months earlier. The diachronic function can be
differentiated in terms of whether the child is learning the
adult's presentation of categories of information, or of infor-
mative content, or both (see Table 1). It is for the diachronic
function of the knowledgeable adult that we present evidence
in this paper.
Growth of knowledge as a basis for language development.
In the foregoing discussion, we have repeatedly introduced the
notions of 'knowledge' and 'knowledgeable adult'. This empha-
sis is intentional. We would like to argue that the development
in the child's ability to present complete, complex, convention-
alized information structures with reference to any particular
picture (or object, event, situation, or relationship) is the
 Building Stories / 131

product of accretion of knowledge about that picture (object,

event, situation, or relationship).
This i s , clearly, an extreme statement, which probably has
to be moderated. Some development in cognitive capacity inde-
pendent of specific knowledge is surely occurring during the
period studied, and the growth in complexity, completeness,
and elaboration of information structures achieved at later
discussions is no doubt partly attributable to that develop-
ment. Nonetheless, we feel a strong case can be made that
the more important process is the simple accretion of knowledge
concerning: (a) what questions need to be answered about a
given picture, (b) what the central items and events in the
picture a r e , and (c) how one talks about the central items and
events. It is useful at least to t r y to push as far as it will
go our explanation that this development is the result of the
accretion of knowledge, before reverting to more general,
vaguer, and less testable explanations in terms of the child's
developing cognitive capacity.

The data base. The findings to be analyzed here are based

on 13 tapes of conversation between a mother and her firstborn
child, Nathaniel. The tapes are a subset of recordings made
at regular intervals during Nathaniel's third and fourth years.
Those selected for analysis here include all occasions on which
Nathaniel and his mother were reading Richard Scarry's Story-
book Dictionary (London: Hamlyn 1967). Table 2 gives more
information on the data base.

Table 2. Nathaniel's age and information about the pictures

discussed at each of the three series of book-
reading sessions.

Series 1 2 3
Nathaniel's age 2;5.18-2;6.2 2;6.19-2;6.22 3;4.8-3;4.21
Number of different
pictures of which
discussions are in-
cluded in the
analysis 41 27 45

Number of picture
discussions
analyzed 96 59 102

For ease of presentation, we analyze the 13 occasions as

three series: Series 1 includes t h e first four recordings, all
of which occurred on separate days within a period of two
weeks; Series 2 includes the next four, which occurred within
a period of four days starting about two weeks about Series 1;
and Series 3 includes five recordings within two weeks,
132 / Catherine E. Snow and Beverly A. Gold field

starting nine months after Series 2. Within each series, only

those pictures have been analyzed which were discussed at
least twice within that series; this selection was made because
we felt that the pictures which were discussed several times
within a period of a few days best reflected Nathaniel's de-
velopment of integrated and complete information structures.
Pictures which were discussed only once or only at six-month
intervals were talked about in much less sophisticated ways
than those discussed intensively over a shorter period of time;
often the discussions of the more rarely talked about pictures
focused only on the vocabulary items needed to identify the
major characters or objects in the picture.
During the period of time under consideration, all of Nathan-
iel's readings of this book with his mother (or, occasionally,
father) were recorded, so the conversations analyzed here
represent his total exposure to discussions with a knowledge-
able adult about this book for these 11 months. He had read
the book with his parents on several occasions before taping
started.
Because we are interested in those pictures for which some
elaborated knowledge structure is built up, we have analyzed
only those conversations about pictures discussed more than
once within a session. In fact, there was considerable rou-
tinization in Nathaniel's structuring of the book-reading
activity, at several levels (see Snow in press, Snow and Gold-
field 1980, for other analyses revealing the levels of routini-
zation), including the level of picture selection. Of the more
than 800 pictures available for discussion in the book, more
than 500 were never discussed in any of the 13 readings, and
352 were discussed more than once.
The character of the pictures is crucial both to the nature
of the conversations held and to the kind of analysis done (and
probably to the popularity of the Richard Scarry books with
Nathaniel and other children as well). The pictures, each
meant to illustrate one of the entries in the dictionary, are
highly narrative, presenting both actual and impending action.
The text is in most cases only supplementary to the narrative
presented in the picture (as well as being without notable
literary merit). In fact, only at Session 3 was the text read
aloud during the conversations between Nathaniel and his
mother, and then often the conversation proceeded without
reference to the information that had been read.
Analysis of the recurrent conversations. As we have out-
lined, the conversations were analyzed to reveal two functions:
laying out the relevant categories of information, and laying
out the relevant informative content. The following coding
scheme emerged from our attempts to make sense of the data
(note that, though some of these categories may be reminiscent
of story grammar categories, our categories were all generated
 Building Stories / 133

by the data themselves, not imposed by us in accordance with

any theoretical presuppositions).
1. Item labels. The item label category was indicated by
questions like 'What's that?' and 'Who's that?' The informa-
tive content within this category included labels for unique
referents, e.g. 'That's Dingo', or 'That's Dingo's car', as well
as class labels, e.g. 'That's a truck' or 'Those are all trees'.
2. Item elaborations. This category was introduced by
questions like1 'What kind of an airplane?', 'What's that part of
a car called? 'How many pigs?', or 'What color is it?' The in-
formative content scored under this category included labels
for superordinates or subclasses, for parts, and for specifica-
tions of type, number, and color.
3. Event. The category was typically introduced by 'What's
happening?' Content consisted of simple event descriptions,
including the main arguments of the relevant verb, e.g. 'Dingo
hit the apple cart' or 'The apples got knocked all over.'
4. Event elaborations. Content within this category included
provision of event labels (e.g. 'There was an accident'), or of
elaboration of an already introduced event by providing infor-
mation on location, position, consequence, or by explaining it.
5. Motive/cause. This category was introduced by 'Why' or
'How come', and content consisted of explanations for events
in terms of physical causality or psychological motives.
6. Evaluation/reaction. Content in this category consisted
of responses to the characters or the events by the readers,
e.g. 'Dingo's sure a bad driver' or 'That was silly, Dingo.'
7. Relation to the real world. The content in this category
consisted of explicating a relationship between a pictured item
or event and a real world item or event known to the child,
e.g. 'That's just like your fire engine' or 'The pigs are taking
a bath. You did that this morning!'
With reference to any of these seven categories, two tasks
had to be accomplished if the category were to be successfully
incorporated in the conversation: (a) the category of infor-
mation had to be introduced and (b) the informative content
had to be introduced. Two basic questions we asked of the
data were: (a) who undertook these tasks for the various pic-
tures discussed? and (b) did the responsibility for these tasks
shift from the adult in the earlier discussions to the child in
later discussions of particular pictures?
Results and discussion

Development of information structures. The first question of

importance in assessing the development of Nathaniel's knowl-
edge structures for these pictures is: how do the categories
of information discussed change from the first to the last ses-
sions? Is there any evidence that the information structure
134 / Catherine E. Snow and Beverly A. Coldfield

built up around particular pictures grows more complex, com-

plete, or sophisticated?
Frequencies with which the various categories of information
were introduced at each series of sessions are presented in
Table 3. It is clear that there is enormous change over time,
with most of the discussion concentrated on items, item elabor-
ations, events, and event elaborations during Series 1 and 2,
and a striking increase in motive/cause at Series 3.
Table 3. Frequencies of the various categories of information
at each of the three series.
1 2 3
Item label 133 74 72
Item elaboration 51 27 0
Event 71 38 136/78*
Event elaboration 67 6 37
Motive/cause 6 5 77
Evaluation /reaction 2 3 19
Real world 21 8 14
*At the third session, the event was presented 58 times in the
text which was read at the beginning of the discussion, where-
as the event was presented 78 times by Nathaniel or his mother
in spontaneous speech. Thus, the category event was pre-
sented 136 times, though only 78 of those were equivalent to
the mode of presentation of Series 1 and 2.
Of course, the introduction of new categories for discussion
does not cause reduction in discussion of the old categories--
one must mention the characters and objects involved in order
to talk about the event, and one must mention the event in
order to discuss its cause or consequences. The more sophisti-
cated categories are introduced to elaborate on the information
structure created in and remembered from the earlier discus-
sions, not to replace the earlier categories.
Shifting of responsibility. Given the development in the in-
formation structures achieved by Nathaniel and his mother, the
question arises: which of them is responsible? Is the intro-
duction of the information category 'event' during the early
series and the category motive/cause during the third series
due to Nathaniel's mother asking the relevant questions and
providing answers? Or do those changes reflect Nathaniel's
changing behavior? Evidence to answer this question comes
from calculating the proportion of introductions of a particular
category or of provisions of informative content within a cate-
gory that were due to Nathaniel (see Tables 4 and 5).
In the first series of conversations, Nathaniel took primary
responsibility for introducing the category item label (see
Table 4) and shared about equally with his mother the tasks
of introducing the categories item elaboration and event.
 Building Stories / 135

Table 4. Percentages of use of a given information category

introduced by Nathaniel or his mother.
1 2 3
M N M N M N
Item label 16.5 83.5 6.8 93.2 65.3 34.7
Item elabor-
ation 52.9 47.1 29.6 70.4 0.0 0.0
Event 50.7 49.3 28.9 71.0 58.9 41.0
Event elabor-
ation 61.2 38.8 66.7 33.3 33.3 66.7
Motive/cause 83.3 16.7 80.0 20.0 18.2 81.8
Evaluation/
reaction 100.0 0.0 66.7 33.3 84.2 15.8
Real world 57.1 42.8 37.5 62.5 100.0 0.0

His mother was responsible for all or most of the introductions

of evaluation /reaction, motive/cause, event elaboration, and
real world references. In the course of the 11 months, Nathan-
iel gradually assumed more responsibility for introducing all
these categories; especially striking is the high frequency of
his introducing event at Series 2 and motive /cause at Series 3.
The responsibilities for providing informative content were
assigned somewhat differently (see Table 5). During the first
series, Nathaniel provided more than half the content only for
item labels, and close to half only for event and real world
reference.
Table 5. Percentages of provision of informative content in
a given category by Nathaniel or his mother.
1 2 3
M N M N M N
Item label 42.8 57.1 27.0 73.0 27.1 72.9
Item elabor-
ation 68.6 31.4 44.7 55.3 -- __
Event 53.5 46.5 55.3 44.7 32.1 67.9
Event elabor-
ation 67.2 32.8 83.3 16.7 21.2 78.8
Motive/cause 83.3 16.7 100.0 30.0 86.8 13.2
Evaluation/
reaction 100.0 0.0 33.3 66.7 84.2 15.8
Real world 55.0 45.0 37.5 62.5 71.4 28.6
By the third series, Nathaniel was providing most of the infor-
mation about items and events, and his mother supplied most of
the information with reference to motive/cause. These highly
event-related categories, unlike the more peripheral evaluation/
reaction and real world references, show a pattern of comple-
mentarity between category and content during the third series,
136 / Catherine E. Snow and Beverly A. Goldfield

reflecting the by now well developed and smooth question and

answer organization of the dialogue.
In summary, then, at the earliest sessions Nathaniel most
often asked 'What's that?', often provided item labels himself,
and collaborated with his mother in describing and elaborating
events. He rarely asked or answered questions about motive/
cause. During the second series, he both introduced events
and provided informative content about events more than pre-
viously. By the third series, he shared the responsibility for
asking questions about items and events, and provided most of
the informative content about items and events. He did almost
all the asking, but very little answering, about motive/cause.
Those results reflect a real shift in responsibility from
mother to child, and the child's growing understanding of how
to talk about the pictures discussed. The nature of this shift
may be clearer if we take some examples of the dialogue. Con-
sider, for instance, the first recorded discussion of the picture
of an accident, held when Nathaniel was 2;5.18.

Nathaniel: Mother:
car
car
that's a car
car de de go up a apples
Yeah, Dingo crashed into the
apples.
apples, crash
crash
going backwards
Is Dingo going backwards?
Here, Nathaniel identified two items (apples and car) and two
events (go and going backwards), while his mother initiated
neither category nor content. The second discussion, at 2;6.0,
was shorter.
Nathaniel: ' Mother:
dat
What's that?
a bable
Spilled the apples, didn't he?
Dingo ran into the apples.
Had an accident.
Here, Nathaniel introduced category and content for item label
(dat and a bable), while his mother supplied content for event
and various event elaborations. Two days later, the following
discussion took place.
 Building Stories / 137

Nathaniel: Mother:
dis is dis
Who's that?
Who's dat?
Who's that?
Who's dat? i
That's Dingo.
Dingo
Dingo Dingo is a terrible driver,
er de de de crashing...
er de
Dingo's always having crashes,
yeah.
Here, Nathaniel introduces the category item label, eliciting the
content Dingo from his mother, then provides a miminal descrip-
tion of the central event crashing, while his mother gives a
fuller description of the event and a reaction to it.
Ten months later, the conversations begin almost ritually
with the new form 'read dis one'. As mentioned earlier,
Nathaniel's mother sometimes did read the text upon request,
although she often (as in the first conversation below, at
3; 4.9) avoided doing so.
Nathaniel: Mother:
Read dis one
What's happening there?
Dingo crashed in
What did he crash?
De apples in in dis
apple cart
Dingo had a n . . .
accident
My goodness!
accident
Here, Nathaniel's mother introduced the event and two event
elaborations, but Nathaniel provided all the informative content.
The next day, the same sequence occurs, but then Nathaniel's
mother provided a motive for the event.
Nathaniel: Mother:
Read dis one
accident. Dingo had an acci-
dent. My goodness. What'd
Dingo do?
Bump the apple underneath
the apple cart, turnin'
138 / Catherine E. Snow and Beverly A. Goldfield

Nathaniel: Mother:
yep
slow
? think 'cause he likes to
bump into apple carts. He
turned so he could bump into
the apple carts. Oh Dingo!
On the final recorded reading, at 3;4.21, Nathaniel described
the event, asked why, then provided a cause after his mother
had provided a motive.
Nathaniel: Mother:
Read dis one
accident. Dingo had an acci-
dent. My goodness. What
happened?
What happened? It gonna
get into the whole apple
cart. His car turned all
by itself.
yep because. ..Dingo wasn't
being very careful I guess.
Why?
Well, you know Dingo, don't
you?
He he wasn't watch he was
going?
Wasn't watching was going
What de what mm what
Dingo Dingo waved?
Did he wave?
yeah
Maybe.
waved Dingo
Another example of the degree to which Nathaniel's interests
had shifted in the later sessions to the category motive /cause
comes from the following dialogue, about the picture labelled
'Back'. The picture shows Mr. Fixit's truck, the back of
which has just been smashed by Dingo. Mr. Fixit is emerging
from a shop with a stove on his back. He is presumably in-
tending to put the stove in the back of his truck, which is
now, however, squashed beyond utility. The conversation
(held at 3; 4.8) proceeded as follows:
 Building Stories / 139

Nathaniel: Mother:
What he doin?
What's he carrying on his back?
Has a oven
Why nowhere to where to
put the oven?
Why?
Why?
What did Dingo do to Mr.
Fixit's lorry?
Smash
Right.
He's gonna fix it with the
oven?
This discussion builds on several previous ones, as shown by
Nathaniel's ability to presuppose a joint understanding of the
problem in his third utterance, 'Why is there nowhere to put
the oven?' (Nathaniel referred to stoves as ovens during this
period.) The most interesting aspect of this discussion is
what it reveals about Nathaniel's strategy at this age for under-
standing these pictures: take two salient events in the pic-
ture, and try to integrate them into a cause and effect struc-
ture. In this particular case,, that strategy yields a proposition
which is somewhat implausible by adult standards. Just as mor-
phological over generalizations reveal a child's knowledge of mor-
phological rules, this violation of reasonable real world relation-
ships reveals Nathaniel's control of the obligatory information
structure for these narrative pictures.
Similar overgeneralizations of the notion that the pictures
should be understood as events occurred during the second
series. The first example shows Nathaniel imposing the cate-
gory 'event' on a picture which presented no event, but only
an object to label (2; 6.19).
Nathaniel: Mother:
Who's this
cake
da bake da cake
There's the cake, in the bakery
window.
The second example shows Nathaniel interpreting what is
actually a pictured motive structure as an event. The picture
being discussed shows three beggars gazing wistfully at a cake
(2;6.22).
140 / Catherine E. Snow and Beverly A. Goldfield

Nathaniel: Mother:
Who's the cake
That's the cake,
eh they're pushing the
cake
They're pushing it?
pushing it
They're wishing they could eat
it.
These attempts to impose event structures on pictures to which
they are not appropriate reveal Nathaniel's implicit rules for
understanding and for talking about these pictures at 2-1/2,
just as the later search for motives and causes reveals that he
considered a causal structure to be crucial to a complete under-
standing of the pictures at 3; 4.
Conclusion. It has been the purpose of this paper to demon-
strate that what children say when describing pictured events
is a function of (a) their general knowledge about the categories
of information necessary for structuring such discussions and
(b) the picture-specific knowledge of the necessary informative
content. Furthermore, we have shown how both these levels of
knowledge emerge diachronically as well as synchronically from
interactions with knowledgeable adults. Children learn from
adults first, what questions to ask and second, how to answer
those questions. The learning is facilitated by the kind of
interactive situation analyzed here, in which precisely identical
contexts for discussion recur over time (see Snow and Gold-
field 1980, for further discussion of context specificity in lan-
guage acquisition).
We conclude from the data presented here that Nathaniel was
learning from the conversations with his mother how to talk
about the pictures in the Richard Scarry Storybook Dictionary.
Presumably, that rather restricted learning had more general
effects in three areas: (1) the language forms that Nathaniel
acquired in the context of these discussions were eventually
available to him for talking about a wide variety of things; (2)
the notions of item, event, motive, and cause as organizing
structures for talking about these pictures were also seen to
be relevant for discussing other kinds of phenomena, for exam-
ple, real-world events; and (3) Nathaniel was learning how to
think about both pictured and nonpictured events, internalizing
the interactively produced information structures in such a way
that they organized knowledge for him without any further de-
pendence on interaction.
 Building Stories / 141

REFEFENCES
Mandler, Jean, and Nancy Johnson. 1977. Remembrance of
things parsed: Story structure and recall. Cognitive Psy-
chology 9.111-115.
McNamee, Gillian D. 1979. The social interaction origins of
narrative skills. The Quarterly Newsletter of the Laboratory
of Comparative Human Cognition 1.4:63-68.
Michaels, Sarah. 1981. Sharing time revisited. Paper pre-
sented at the Ethnography in Education Research Forum *
University of Pennsylvania, March.
Rumelhart, David. 1975. Notes on a schema for stories. In:
Representation and understanding: Studies in cognitive
science. Edited by D. G. Bobrow and A. Collins. New
York: Academic Press.
Scollon, Ron. 1979. A real early stage: An unzippered con-
densation of a dissertation on child language. In: Develop-
mental pragmatics. Edited by E. Ochs and B. Schieffelin.
New York: Academic Press.
Shugar, Grace. 1978. Text analysis as an approach to the
study of early linguistic operations. In: The development
of communication. Edited by N. Waterson and C. Snow.
Chichester: Wiley.
Snow, Catherine E. 1977. The development of conversation
between mothers and babies. Journal of Child Language
4.1-22.
Snow, Catherine E. 1978. The conversational context of
language learning. In: Recent advances in the psychology
of language: Language development and mother-child inter-
action. Edited by R. N. Campbell and P. Smith. London:
Plenum Press.
Snow, Catherine E. (in press) Saying it again: The role
of expanded and deferred imitations in language acquisitions.
In: Children's language, Vol. 4. Edited by K. E. Nelson.
New York: Gardner Press.
Snow, Catherine E., and B. A. Goldfield. 1980. Turn the
page please: Situation-specific language learning. Unpub-
lished MS.
Wertsch, James. 1979. From social interaction to higher psy-
chological processes: A clarification and application of
Vygotsky's theory. Human Development 22.1-22.
 COMPETENCE FOR IMPLICIT TEXT ANALYSIS:
LITERARY STYLE DISCRIMINATION
IN FIVE-YEAR-OLDS

Georgia M. Green
University of Illinois

Why on earth would anyone imagine that 5-year-olds could

tell one literary style from another? I certainly would not have
if someone had not asked me if they could.
A couple of years ago, I remarked to someone that when my
daughter Robin was 2f, she claimed to recognize illustrations
she had never seen before, saying that we already had books
we had just gotten. In fact, what we already had were books
illustrated by the same artist, but the illustrations were of
course not identical: what she recognized was the illustrator's
artistic style. My interlocutor, who was a well-known expert in
language acquisition, asked if Robin also thought she recognized
stories she had never heard before when they were by the same
author as ones she had heard, if she recognized verbal style as
easily as artistic style. My response was that not only did I
not know if she had been able to do that at 2\\ I had no idea
if she could do it at the time, when she was 4i.
By the time I had figured out how to find out, I realized that
it would be about as easy to investigate the abilities of a more
representative population of children as it would be to explore
the abilities of one child, and there would be a lot to be gained,
for the question is of more than passing interest, and my inter-
est in it was more than idle curiosity: if children at the age
when reading instruction typically begins are sensitive to stylis-
tic properties of texts, then this has far-ranging implications
for their text-processing abilities, and these in turn have im-
plications for diverse aspects of the practice of reading instruc-
tion. I am going to return to these eventually.
So, the question was: given that one child, and I presumed
many others, interpreted the similarities in the illustrations of

142
 Literary Style Discrimination in Five-Year-Olds / 143

artists like Lionel Kalish, as shown in (1) and (2), as 'identi-

ties' of a sort at age 2J, could children recognize similarities in
verbal (or literary) style as indicating identity of authorship?
I arranged to carry out a small-scale experiment, with the aid
of a research assistant, Margaret Laff, in the kindergarten class
of a day-care center in a midwestern university community of
95,000. The participants were five girls and eight boys, rang-
ing in age from 5;0 to 6;1 years. These children had not begun
formal reading instruction, although two of them could read un-
familiar texts with some facility.
At our request, the regular classroom teacher read 10 books
to the class at times normal for such an activity and in the way
she normally would read to the children, showing the illustra-
tions and answering questions. It took 14 days for the books,
2 by each of 5 authors, to be read once. The 10 books, read
in the order in which they are listed, are indicated in (3).
(3) Exposure books:
1. Dr. Seuss. The Lorax. New York: Random House,
1971.
2. Margaret Wise Brown. Wait Tin the Moon is Full.
New York: Harper and Row, 1948.
3. Bill Peet. The Ant and the Elephant. Boston:
Houghton Mifflin, 1972.
4. Virginia Kahl. The Habits of Rabbits. New York:
Charles Scribner's Sons, 1957.
5. Beatrix Potter. The Tale of Mr. Jeremy Fisher.
New York: Warne, 1906.
6. Dr. Seuss. Happy Birthday to You. New York:
Random House, 1959.
7. Margaret Wise Brown. The Runaway Bunny. New
York: Harper and Row, 1942.
8. Bill Peet. Big Bad Bruce. Boston: Houghton
Mifflin, 1977.
9. Virginia Kahl. The Baron's Booty. New York:
Charles Scribner's Sons, 1963.
10. Beatrix Potter. The Tale of Peter Rabbit. New
York: Warne, 1902.
Shortly after the last book was read to the group, we pre-
pared the group for the task of indicating their identification
of new stories with an activity where they noted whether they
recognized which book an illustration was from. Five-page
booklets were distributed to the children. On each page of the
booklets five pictures had been photocopied in black and white.
Each picture represented a major character from a book by a
different one of the five authors mentioned in (3). In every
case the character came from one of the books read to the chil-
dren in class, and with only one exception, the character's
name occurred in the title of the book. The same five pictures
144 / Georgia M. Green
 Literary Style Discrimination in Five-Year-Olds / 115

appeared on each page but they were arranged in different

orders. For each page, the children were asked to put a
crayon mark on 'the picture that looks like it was drawn by
the person who drew the pictures in (title) and (title)'; the
two titles by each author were cited in turn. This was an un-
usual task for the children and a few seemed puzzled by it.
Though most seemed to know the correct answers, some may
have been distracted by wondering why we would ask something
so obvious. We also observed in at least one case that a child
would point to the correct answer, but for some reason could
not be persuaded to mark it. The children got from 2-5 cor-
rect, as indicated in (4); 9 got 3 or more correct.

(4) Illustrations task results:

Number correct 2 3 4 5 Total
Number of children 4 5 1 3 13
Then, five tape recordings of other stories by the same
authors were played individually to each child. The stories on
the tapes are indicated in (5).

(5) Test books:

1. Dr. Seuss. I Had Trouble in Getting to Solla Sollew.
New York: Random House, 1965.
2. Beatrix Potter. The Tale of Two Bad Mice. New
York: Warne, 1904.
3. Bill Peet. EH. Boston: Houghton Mifflin, 1978.
4. Margaret Wise Brown. Fox Eyes. New York:
Pantheon Books, 1951.
5. Margaret Wise Brown. The Little Fur Family.
New York: Harper and Row, 1946.

Each child heard the tapes in a different order. Some chil-

dren heard one story by each author; some heard 2 stories by
one author and one by each of 3 others. Thus, not all the chil-
dren heard all the authors. This was intended to serve as a
check on guessing strategies. Unfortunately, one-third of the
children in the second condition did not complete the task, so
we did not draw any conclusions about guessing strategies.
Before each story, the children were told that at the end of
the story they would be asked to think about which of the books
read by the teacher the new story most reminded them of. The
children were also told that when the story was over, they
would be asked to make a mark on a picture in a booklet similar
or identical to one used in the illustration identification task. x
Not all booklets were identical: the children who heard two
stories by the same author had five 4-item pages, while those
who heard one story by each author had five 5-»item pages.
146 / Georgia M. Green

When each story was over, the interviewer read these in-
structions to the child:
If you think this story was written by Beatrix Potter, who
wrote the stories about Peter Rabbit and Jeremy Fisher, put
a mark on the picture of Peter Rabbit.
If you think the story you just heard was written by
Virginia Kahl, who wrote the stories about Gunhilde and the
rabbits, put a mark on the picture of Gunhilde.
If you think that the story was written by Margaret Wise
Brown, who.wrote the stories about the runaway bunny and
the raccoon who wanted to go out at night, put a mark on
the little raccoon's picture.
If you think the story was written by Dr. Seuss, who wrote
the stories about the Lorax and the Birthday Bird, put a
mark on the picture of the Lorax.
(5-item group only) If you think the story was written by
Bill Peet, who wrote the stories about Big Bad Bruce and
the ant and the elephant, put a mark on the picture of the
bear.
After the child had marked a choice, the interviewer asked the
child three questions: (1) Have you ever heard this story be-
fore? (2) How did you know it was that one? (3) Tell me
something about the story that made you know who wrote it.
We did not expect to get much in the way of revealing or
even true answers to such questions (5-year-olds have been
observed to have no qualms about making up answers to such
questions out of whole cloth), but we were prepared to con-
sider anything indicating awareness of any stylistic property
to be significant.
Responses fell into one of three categories. Many were either
'off the wall' or simply uninformative. For example, in re-
sponse to the second question, 'How did you know it was that
one?', we got such responses as:
Well my dad told me.
I just knew. I was just thinking in my head. I remembered
in my mind who it was always written by.
Some of these children had correctly matched the author. Some
had matched it incorrectly. A good number of responses, how-
ever, seemed to indicate at least a vague awareness of style.
For instance, in response to the same question, 'How did you
know it was that one?', children who had correctly identified
the authorship of the story said things like:
 Literary Style Discrimination in Five-Year-Olds / 147

Because . . . uh . . . because they were talking the same.

Um, because of how they were talking.
Well, it sounds like she's the one (pause) that was talking.
It really sounds like the Lorax girl. See, in little parts of
it it sounded like she was talking. And she was talking in
the Lorax, I think, because she sounds the same as the
Lorax girl.
And a few comments showed that at least one child was con-
scious of certain determinants of style. For example, respond-
ing to the same question, this child said:
Because I heard the story of Big Bad Bruce and they said
something about the s . . . nort, and they said it too.
Most of the children, predictably, did not have the concen-
tration to perform the entire task at a single sitting (about 55
minutes), and did one or two stories at a time. Three or four
children did have the concentration to do this, however (two of
these were readers), and several were so intrigued with the
task of guessing the authorship that they interrupted the tape
to tell us the author (usually correctly) and preferred, con-
trary to our expectations, to go on to the next tape, rather
than hear the end of the story.
This part of the experiment was not conducted under the
best of circumstances. The tapes were unfortunately exces-
sively 'noisy 1 , and the listening accommodations were not parti-
cularly comfortable--usually the floor of a small room that was
not in use.
Children were allowed to discontinue the experiment at any
time if they did not wish to go on. Three children did not
complete the task. One listened to 4 out of 5 stories, one to
3 out of 5, one to 2 out of 5.
Thus the experiment was performed under a number of condi-
tions that could be expected to bias the results against the hy-
pothesis that children can identify stylistic traits of texts well
enough to match the authorship of novel texts to texts they
have already heard. (1) Children were exposed to only two
exemplars by each author prior to the testing. (2) Children
were exposed to each exemplar only once. (3) Exposure
stretched over 14 days. The two books by each author were
read for the most part 7 days apart. (4) The testing task was
lengthy. (5) The testing was conducted under uncomfortable
and distracting conditions. Nonetheless, when they listened to
tapes of a third work by each of the 5 authors, 6 of the 13
children who participated in the interviews were able to identify
correctly the authorship of 3 or more of the 5 stories, as indi-
cated in (6).
148 / Georgia M. Green

(6) Style test results

Number
of 4
Children

1 2 3
Number Correct

The probability of randomly choosing the correct item out of

5 is 0.2. The probability of doing this 3 or more times in 5
trials is around 0.06. This means that 6 children performed at
a level of accuracy highly unlikely to be attributable to chance.
The other 4 who completed the interviews performed with far
below chance accuracy. In other words, a large percentage of
the children performed in such a fashion as to imply that their
comprehension of stories was not limited to vague outlines of
plot and characterization, but extended to appreciation of the
subtler rhetorical and linguistic aspects of style. Apparently,
the other part of the group either (a) misunderstood the task,
(b) did not attend to the discriminants of style, or (c) fixed
upon arbitrary guessing strategies.
Correlations. There was no apparent correlation of the per-
centage correct with the participants' age or sex, as shown in
(7).
Furthermore, there was no direct correlation between the
children's ability to do well on the illustration pretest and their
ability to perform the style recognition task. This indicates
that performance on the style recognition task is not a simple
function of intelligence or ability to follow directions. Specifi-
cally, of the 10 children who completed the style recognition
task, the 3 children who did best on the illustration recognition
task (matched all 5 pictures correctly) got 0 or 1 correct on
the style recognition task. The children who did poorest (2
correct) on the illustration task, with one exception, got 0 or
1 correct on the style recognition task. But the children who
 Literary Style Discrimination in Five-Year-Olds /

did moderately well on the illustration task (3-4 correct) got

3-5 correct on the style recognition task.
(7) Comparison of number of correct responses with age
and with sex.

Average age of:

Total group 64. 3 months
3-5 Correct group 64.8 months
0-1 Correct group 63.8 months
Percentage of girls in:
Total group .38 (5/13)
3-5 Correct group .33 (2/6)
0-1 Correct group .43 (3/7)

A possible explanation for this is that the group that got

100 percent correct on the illustration task were accustomed to
attending much more to the illustrations in listening to stories
than to rhetorical and linguistic properties of the text, and
that most of the children in the group that did poorest on the
illustration task simply were not accustomed to attending to
either style or illustrations in listening to stories. But the
reason why the children who did best on the style recognition
task did only moderately well in recognizing illustrations may
be that their concentration on the aspects of literary style that
allowed them to recognize authorship precluded their paying
more attention to the illustrations.
In the absence, however, of confirmatory observations of the
individual children, it seems just as justifiable to attribute the
gap between the 0-1 correct group and the 3-5 correct group to
individual differences (e.g. sensitivity to language) or linguistic
maturity. Another possibility is that the children in the 0-1
correct group simply had less prior experience with the authors
whose style we chose to investigate. Logically, this would seem
to be a significant variable only if, in being read to before the
experiment, these children were made aware of the names of
the authors of the relevant books. I have no idea whether
this was true in the case of the children tested. I would guess
that the practice of reading the title page is not widespread,
but I know of no definitive investigations. Personally, I never
used to read aloud even the titles of the books I read to my
children, and as a consequence, they developed their own
designations for books. Thus, my daughter's name for The Cat
Who Stamped His Feet, by Betty Ren Wright (Golden Press,
1974) was 'the cat-in-the-attic book', and her name for The
Sheep of the Lai Bagh, (2) by David Mark (Parents Magazine
Press, 1967) was 'the Ramesh book'.
150 / Georgia M. Green

On the other hand, Cazden suggests (personal communication)

that prior exposure to other books by the same authors, even
when the author's name is not mentioned, might provide a child
with a frame in which to assimilate and categorize stylistic
properties of texts.
Before I go on to describe with more specificity the linguistic
and rhetorical aspects of text that these children must have
been attending to in order to make the correct judgments that
they made, I am going to describe how we selected the materials
for this task, because we took great pains to avoid using
materials that would allow a participant to make correct answers
based on text properties that we considered not particularly lin-
guistic, such as similarities of subject matter, or familiarly
named protagonists.
Selection of materials was not a matter to be taken lightly.
We knew that children might use subject matter or characters 1
names to decide authorship. For example, in a similar forced-
choice task, one child correctly chose 'the author of the Babar
books' as the author of an unfamiliar paragraph referring to an
individual1 named Arthur, and 'the author of Hi, Cat and Whistle
for Willie as the author of an unfamiliar paragraph referring to
a dog named Willie. When questioned, she replied that she had
made her judgments on the basis of the name Arthur and the
name Willie, respectively. Thus, our materials had to meet all
of the following criteria:
1. The author had to have a distinct style. If we were not
able, intuitively, to identify an author's works as stylistically
unique, we did not consider her or his works as candidates for
inclusion in the study. This eliminated a number of celebrated
children's authors, including Ezra Jack Keats and Robert
McCloskey.
2. The author had to have written at least three books which
were not all about the same unique subject matter. This ruled
out, for example, Jay Williams, among whose books we could
find only one that was not about princesses or kings.
3. The author had to have written at least two books with
nonover lap ping sets of characters. This, regrettably, ruled
out many authors with strongly individual styles--for example,
the de Brunhoffs, authors of the Babar books. We considered
including such authors, and changing the characters' names so
as not to 'give away' the authorship. We rejected this strategy
however, on the grounds that (a) the kinds of names an author
chooses are an aspect of style, and we did not want to compro-
mise the integrity of the experiment by meddling with even one
aspect of an author's style; and (b) if a child did know such
an author's works well, it might be unfairly confusing to ask
for judgment on a work that both is and is not that author's.
4. We had to have access to at least three books by the
author that shared a distinct style. This eliminated such stylis-
tically interesting authors as Maurice Sendak and Rosemary Wells,
 Literary Style Discrimination in Five-Year-Olds / 151

since we could not find three books (on the shelf at the local
library) that met our other criteria and shared the same style.
5. At least one of the books, and preferably all three, had
to have a text which could present the story independently of
the illustrations, so that (a) the familiarization stories could be
equally well assimilated by children sitting farther from the
teacher and by children clustered closely around her, and (b)
the taped story would not be incomprehensible.
The testing had to be done with tapes of the books rather
than exemplars, even exemplars that obliterated the author's
name, in order to eliminate the possibility that the children
might identify the authorship by identifying the illustrations,
which in most cases here were done by the author. Also, we
wanted to eliminate the graphics (type face, layout) as a possi-
ble source of identification, since we had observed that at least
some 2-year-olds can recognize these things and 'read' the
Crest, K-Mart, Sears, and Special K logos. (One 2-year-old
insisted for months that a certain supermarket was an ice cream
store, despite regular correction. Eventually, his mother
noticed that the lettering on the store's sign was very similar
to that used by the Baskin-Robbins chain, and made some head-
way in clearing up the confusion.) We figured that 5-year-olds
might also use such cues, and we wanted to eliminate them.
What we finally ended up with was the following: two authors
who wrote in rhymed couplets and used many long words: Dr.
Seuss and Virginia Kahl; and three authors who wrote about
anthropomorphized animals: Beatrix Potter, Margaret Wise
Brown, and Bill Peet.
Finding five authors who met all of our criteria was very
difficult. In the initial planning of the study, we feared that
including Dr. Seuss might bias the experiment in favor of the
hypothesis. However, the discovery of Virginia Kahl allowed
us to include both authors in the study, as both write verse
fantasy in similar meter. Samples are reproduced in (8) and
(9).
(8a) "Oh, help!" cried the Duchess. "Our children are gone!
They're not in the castle--they're not on the lawn--
They're not in the gardens. Are they down in the moat?"
"If they are," said the Duke, "let us hope they can
float."
"They have vanished, they've all disappeared from our
sight.
Our dear little daughters give one such a fright."
(Virginia Kahl: The Baron's Booty)
152 / Georgia M. Green

(8b) The message told what the men had seen:

An enormous beast of yellowy-green,
With a sinuous neck and a small fierce head
That had no hair but had horns instead.
(How Do You Hide a Monster?)
(8c) And everyone cried, "There's been an error.
That beast is never a cause for terror.
He'd never harm us; he's kind and true.
We must protect him; what shall we do?"
At last they announced, after due reflection,
"We'll send the men off in the wrong direction."
(How Do You Hide a Monster?)
(8d) They all ate their pancakes--the very last crumb;
But when they had finished, they all remained dumb.
Then said the Good Wife, "Now, why don't you praise
me?
Your manners are dreadful--you really amaze me.
You know that my pancakes are fluffy and flavory,
Tender and toothsome, incredibly savory--
Served with a syrup so pure and delightful,
That you've often swooned when you've bitten a
biteful."
(The Perfect Pancake)
(9a) Down slupps the Whisper-ma-Phone to your ear
and the old Once-ler's whispers are not very clear,
since they have to come down through a snergelly hose,
and it sounds as if he had smallish bees up his nose.
(Dr. Seuss, The Lorax)
(9b) But I'm also in charge of the brown Bar-ba-Loots
Who played in the shade in their Bar-ba-loot suits
and happily lived, eating Truffula fruits.
NOW . . . Thanks to your hacking my trees to the ground,
there's not enough Truffula fruit to go 'round.
And my poor Bar-ba-Loots are all getting the crummies
because they have gas, and no food in their tummies.
(The Lorax)
 Literary Style Discrimination in Five-Year-Olds / 153

(9c) I was real happy and carefree and young

and I lived in a place called the Valley of Vung
And nothing, not anything, ever went wrong
Until . . . well, one day I was walking along
And I guess I got careless, I guess I got gawking
At daisies and not looking where I was walking.
(I Had Trouble in Getting to Solla Sollew)
(9d) I dreamed I was sleeping in Solla Sollew
On the banks of the beautiful River Wah-Hoo
Where they never have troubles. At least very few.
Then I woke up. Arid it just wasn't true.
I was crashing downhill in a flubbulous flood
With suds in my eyes and my mouth full of mud.
{Solla Sollew)
(9e) Our camel, he said, had a bad case of gleeks
and should be flat in bed for at least twenty weeks.
(Solla Sollew)
(9f) I listened all night to the growls and the yowls
And the chattering teeth of those mice and those owls.
While the Midwinter Jicker howled horrible howls.
I tossed and I flipped and I flopped and I flepped.
It was quarter past five when I finally slept.
(Solla Sollew)
(9g) We're marching to battle. We need you, my boy.
We're about to attack. We're about to destroy
The Perilous Poozer of Pamplemousse Pass!
So, get into line! You're a Private, First Class.
(Solla Sollew)
(9h) They smell like licorice! And cheese!
Send forty Who-Bubs up the trees
To snip with snippers! Nip with nippers!
Clip and clop with clapping clippers!
Nip and snip with clipping cloppers!
Snip and snop with snipping snoppers!
(Happy Birthday to You!)
Similarly, by choosing three animal story authors, we hoped
to eliminate topic as a cue to authorship, and force the judg-
ments to depend on subtler cues. Indicative samples of the
three authors' texts are reproduced in (10)-(12).
154 / Georgia M. Green

(10a) "If you are a gardener and find me," said the little
bunny,
"I will be a bird and fly away from you."
"If you become a bird and fly away from me," said his
riiother,
"I will be a tree that you come home to."
(Margaret Wise Brown: The Runaway Bunny)
(10b) Once upon a time in the dark of the moon there was a
little raccoon.
(Wait Till the Moon Is Full)

(10c) "Does everyone sleep at night?" asked the little

raccoon.
"No," said his mother, "not everyone."
"Who doesn't?" asked the little raccoon.
"All things that love the night," said his mother.
"Wait till the moon is full."
"Is the moon a rabbit?" asked the little raccoon.
"No," said his mother. "The moon is a moon. A big-
round golden moon."
"Will I see it soon?"
"Wait," said the mother. "Wait till the moon is full."
(Wait Till the Moon Is Full)

(lOd) There was a little fur family

warm as toast
smaller than most
in little fur coats
and they lived in a warm wooden tree.
(Little Fur Family)
(lOe) Then the little fox climbed an apple tree. Along the
bark of the tree the eye of a tree toad closed suddenly.
The fox coughed, "Whiskerchew!" And the tree toad
knew that someone had seen him hiding there in plain
sight against the bark of the tree. Some children who
were supposed to be taking a nap in the afternoon
weren't sleeping at all. "Whiskerchew!" the fox
coughed. And the children knew that the fox knew
that they were not sleeping. All this the fox noted,
and he went on his way.
(Fox Eyes^
 Literary Style Discrimination in Five-Year-Olds / 155

(lla) Peter gave himself up for lost, and shed big tears;
but his sobs were overheard by some friendly sparrows,
who flew to him in great excitement, and implored him
to exert himself.
(Beatrix Potter: The Tale of Peter Rabbit)
(lib) I am sorry to say that Peter was not very well during
the evening.
(Peter Rabbit)
(lie) So that is the story of the two Bad Mice,--but they
were not so very very naughty after all, because Tom
Thumb paid for everything he broke.
(The Tale of Two Bad Mice)
( l i d ) "What a mercy that was not a pike!" said Mr. Jeremy
Fisher. "I have lost my rod and basket; but it does
not much matter for I am sure I should never have
dared to go fishing again!"
(The Tale of Mr. Jeremy Fisher)
(lie) And instead of a nice dish of minnows--they had a
roasted grasshopper with lady-bird sauce; which frogs
consider a beautiful treat; but I think it must have
been nasty!
(Jeremy Fisher)
(12a) "Where in blazes did you come from?!!" she shrieked,
giving the boulder a vicious kick.
(Bill Peet: Big Bad Bruce)
(12b) Once upon a time there was a lion named Eli who lived
in the faraway land of Kumbumbazango. He was a
decrepit old cat with a scruffy mop of mane, and most
of the thunder had gone out of his roar. Now, after
many long years as a proud king of beasts, the old
lion had finally become as meek as a mouse.
(Eli)
(12c) In one frantic leap, and with a wild swing of a paw,
Eli caught the jackal with a clout to the snout that sent
the little rascal yelping away with his tail between his
legs.
(Eli)
156 / Georgia M. Green

(12d) Raising his voice to a rumble to make sure all the

birds could hear, the lion let them have it. "You
good-for-nothing grubby old bone-pickers! You flea-
bitten beggars! You ugly old coots! You give me the
creeps! Skedaddle! Take off! Get a tree of your
own! Leave me be!"
(Eli)
(12e) "Wade out into that soup and scrunch down in the
gunkazunk grass. The Zoobangas will never look for
you there."
(£10
What might the children have been picking up on to make the
correct identifications that they made? Let us begin with the
verse selections. At first, the similarities between Kahl and
Seuss may seem more striking than the differences. Both write
obvious fantasy with a strong four-foot meter, mostly anapests.
And both do not hesitate to use words likely to be unfamiliar
to young children. But here the similarity ends. Seuss' un-
familiar words tend to be unfamiliar because they are invented
(slupps, snergelly, gleeks, flubbulous, snop), whereas Kahl's
are likely to be unfamiliar because they are drawn from the
formal, academic register of language, to which few young chil-
dren have been exposed, and hardly any have attended. Some-
times she uses basically academic or literary words in her
verses (error, toothsome, swooned), but much of the unfamiliar
word usage is just academic senses of words in common usage
in children's books (for example, reflection in the sense
'thought', due in the sense 'sufficient', true in the sense
'loyal', dumb in the sense 'mute'). Although the plots are com-
paratively simple and predictable, the whole tone of Kahl's stor-
ies is old-fashioned and/or mock academic, and this is reflected
in the syntax as well, in such phrases as cause for terror, and
after due reflection, and in the nonanaphoric use of the pronoun
one to mean 'a person', and the Germanic verb-second syntax of
Then said the Good Wife.
In contrast, the tone of the Dr. Seuss stories is very inti-
mate and conversational. This is reflected in the vocabulary,
where one finds such colloquial items as smallish, tummies, the
crummies, real used to intensify an adjective, and the contrac-
tion go 'round. The conversational tone shows up just as
strikingly in the syntax, in such locutions as the introductory
well, the hedge I guess, and the get + present participle con-
struction (got walking).
Then there are the Seuss trademarks--the made-up species
(Bar-ba-Loots, Truffula, Who-Bubs), and the coined place
names (Valley of Vung, River Wah-Hoo), and the novel com-
pound nouns (Bar-ba-loot suits, Super-Axe-Hacker,
 Literary Style Discrimination in Five-Year-Olds / 157

Key-Slapping Slippard), Finally, alliteration, assonance, and

consonance, as in selections (9g-i), are much more character-
istic of Seuss' verse than of Kahl's.
There are differences in length (the Seuss stories are longer)
and in plot construction: the Seuss stories involve more epi-
sodes, are less predictable, and generally involve a human
protagonist in interaction with nonhuman species (or only non-
human characters), whereas Kahl's stories involve almost ex-
clusively human protagonists (the sole exception is a Loch Ness-
type monster). However, I suspect that these global properties
of the texts were less salient to the children than the more lin-
guistic differences, and this feeling is supported by the fact
that several children made judgments (usually correct) before
they had heard one-tenth of a story. Without hearing a longer
selection, they could not easily have formed correct judgments
about such global properties as length and plot construction.
Furthermore, when the children mentioned reasons for their
choices, they were usually framed in terms like 'it sounded
like . . . ' , although one child, justifying an incorrect choice,
mentioned particular actions:
I think it's (by the author of) Peter Rabbit because they
were planting things and stuff. They were planting
carrots.
As it turned out, the Dr. Seuss story was identified correctly
7 out of 12 times; one Kahl story was identified correctly 4 out
of 11 times, the other once in 2 trials. Among the 6 children
who identified the authorship of 3 or more stories correctly,
the Dr. Seuss story was misidentified only once (as being
written by Kahl), and the Kahl story was misidentified twice.
What cues allowed the children to recognize stories as being
written by Brown, Potter, and Peet? First of all, although all
three begin their stories traditionally enough with Once upon a
time or Once there was or There once was, there are striking
differences in the register used to tell the stories. Peet's
stories have a colloquial (scrunch, clout, snout), even earthy
tone. He minces no words; his characters are scruffy, de-
crepit, crafty. They do not just say or cry or even shout
things, they shriek and let them have it. And his characters,
who tend to be rather bad-tempered, do not mince words
either. Roxy comes as close to cursing in (12a) as you can in
a picture book, and Eli sounds almost like a Marine drill in-
structor when he calls the vultures all those colorfully rude
names in (12d).
On the other hand, Potter's stories, written in Edwardian
England, sound like it. When Mr. Jeremy Fisher curses, it's
'What a mercy that was not a pike!' Some of the vocabulary is
very formal and literary (implored, exert). Many of the phrases
strike the modern ear as old-fashioned or maiden-auntish, for
158 / Georgia M. Green

example, shed big tears, so very very naughty, it does not

much matter, I should never have dared to.
The register of Brown's stories is that of bedtime storytell-
ing. As in the Bank Street College's 'Here and Now' stories,
for which Brown was a principal writer, these stories are al-
most exclusively dialog, with a little bit of narration and de-
scription, and the description is exclusively literal. In this
way, her comparatively plain prose contrasts with Peet's, which
makes copious use of figurative language: Eli has a mop of
mane, and the thunder had gone out of his roar. When the
vultures urge Eli to wade into the swamp (12e), they call it
soup. It also contrasts with Potter's, in that Potter almost al-
ways interrupts her narrative at the end and makes her pres-
ence felt with comments like those in ( l i b ) , ( l i d ) , and ( l i e ) .
I do not mean, by saying that Brown's prose is plain, to im-
ply that it is either colorless and boring, or lacking in style.
It has a lyrical rhythmicity, clear in the refrain 'Wait,' said
his mother. 'Wait till the moon is full', as well as in the selec-
tions in (10b), (lOd), and (lOe). And there is so much inter-
nal rhyme and half-rhyme, as in (10b), (lOd), and (lOe), that
some of the passages almost seem to be in verse. Furthermore,
Brown's prose has a cyclical structure that also marks it as
unique, at least among this group of authors. This structural
cyclicity shows up plainly in the conditional- counterconditional
repartee (exemplified in (10a)) that constitutes almost the whole
of The Runaway Bunny, and it is no less clear in the repeated
requests in Wait Till the Moon Is Full that are answered, every
page or so, with the refrain 'Wait,' said his mother. 'Wait till
the moon is full', as in (10c).
All of the Brown stories used in the experiment are quiet,
calm stories, with no violence and a comparatively low level of
suspense—what is going to happen is never a matter of life
and death. 2 In contrast to Brown's simple, almost plotless
stories, Peet's and Potter's stories involve unpredictable chains
of episodes, and in Peet's these involve embedded and conflict-
ing plans. All of the stories by these three authors that the
children heard have animals as the main protagonists, but
Brown's are almost always presented as juvenile and 'pedomor-
phized', while many of Potter's and all of Peet's are full-grown,
though not grown-up--they act and react like children. At
least one of the stories by each author also involves human be-
ings, though always as minor characters.
The Potter book in the test {The Tale of Two Bad Mice) was
correctly identified 4 times out of 11; the Brown books {Fox
Eyes and Little Fur Family) 7 times out of 13, and 1 time out of
4, respectively; and the Peet book (Eli). 2 times out of 6.
Among the 6 children who correctly identified the authorship of
3 or more books, Fox Eyes was correctly identified 5 out of 6
times, and Little Fur Family 1 time out of 2; The Tale of Two
Bad Mice was correctly identified 4 times out of 6, and Eli, 2
times out of 4.
 Literary Style Discrimination in Five-Year-Olds / 159

Let us turn for a moment to address the question of account-

ing for the errors that were made. What might have caused
some of the confusions? We can identify a number of cross-
author similarities that might account for some of the errors.
Both Peet and Seuss use very colloquial vocabulary and syntax.
And Peet, like Seuss, refers to obviously invented species and
places (gunkazunk grass, Zoobangas, Kumbumbazango), though
Seuss' are more often compounded of familiar morphemes than
Peet's. Both Potter and Kahl use a fairly formal and literary
vocabulary and syntax. Seuss as well as Potter intrudes into
the narrative and makes the author's presence felt. Happy
Birthday to You and The Lorax are specifically addressed to
the reader, the former as an extended wish, the latter as a
sort of reverie. / Had Trouble in Getting to Solla Sollew un-
surprisingly is a first-person narrative. Both Kahl and Brown
tell simple stories, with relatively predictable plots, though
Kahl's are more complex, and some of Brown's have hardly any
plot at all. Finally, while Potter's stories are not as lullaby-
like as Brown's, the suspense is muted, the action damped, by
the calm, matter-of-fact tone of the telling.
How well do these similarities account for the errors that were
actually made? If the errors had been random and evenly dis-
tributed, half of them would have been in cells predicted by
these similarities. In fact, 56 percent of the errors were in
these cells (56.5 percent of the errors by the 0-2 correct
group, and 55 percent of the errors by the 3-5 correct group).
And 4 children made symmetrical errors--for example, identify-
ing the Potter story as by Kahl and vice versa, suggesting that
the errors were not random, but were based on some perception
of similarity.
Implications. This study appears to show that at least some
5-year-olds have the ability to appreciate and discriminate among
the literary styles available in books intended for young chil-
dren. Indeed, several children found the challenge of testing
this ability exhilarating.
I cannot show that what the children were attending to when
they correctly identified the authorship of stories they had not
heard before was, in fact, the linguistic and rhetorical aspects
of literary style that I have indicated (I could not prove that,
even if the experiment had been conducted with well-read and
highly articulate adults), but it seems a good bet. In any case,
it means that the children understood a whole lot more than the
bare outlines (or even dressed-out outlines) of plot. Making
the correct judgments almost certainly entailed not only noticing
and abstracting from very fine details of wordcraft, but also
attending to and abstracting from global structural matters of
form and content.
If it is true that 5-year-olds generally, and by extension, 6-
and 7-year-olds, have the ability to make such fine discrimi-
nations, then it seems likely that they would be able to tell the
160 / Georgia M. Green

difference between the prose in ordinary children's books of

the sort I have been discussing, and the prose in their read-
ers, a sample of which is given in (13).
(13) Rabbit said, "I can run. I can run fast. You can't
run fast."
Turtle said, "Look Rabbit, See the park. You and I
will run. We'll run to the park."
Rabbit said, "I want to stop. I'll stop here. I can
run, but Turtle can't. I can get to the park fast."
Turtle said, "I can't run fast. But I will not stop.
Rabbit can't see me. I'll get to the park."
Such prose is edited to conform to readability formulae which
impose strict limits on sentence length and vocabulary. Owing
to the strict constraints imposed by the publishers of basal
readers 3 on sentence length, vocabulary, and story length,
these works end up being designed in such a way that they
are devoid of most characteristics of individual style. If it is
generally true that at the age when reading instruction begins,
children attend to and appreciate stylistic differences, then it
would seem to follow that expecting them to read such basal
readers i s , to say the least, inconsiderate. At best, it is
pointless; at worst, it is counterproductive. It wastes valuable
time that could be spent in more profitable ways and risks bor-
ing the children and conveying to them that there is nothing
interesting to be learned in books, or even in school. Is it
possible that Johnny does not learn to read because there is no
thrill in being able to read texts like (13), which is from what
is supposed to be a version of Aesop's fable about the hare
and the tortoise?
The objection is likely to be raised that the fact that 5-year-
olds can appreciate the differences between works by Beatrix
Potter and Margaret Wise Brown does not mean that 7-year-olds
could read the works of either author independently, that 7-
year-olds have enough trouble reading the admittedly anemic
prose in the basals. It is certainly true that there is no direct
entailment from what 5-year-olds can comprehend orally to what
7-year-olds can independently read, but I think this study sug-
gests that 7-year-olds might be able to read Margaret Wise
Brown and Beatrix Potter; the fact that some have trouble with
second-grade basals might be due to stylistic properties of the
basals that are introduced in the process of writing a graded
reader. Work at the Center for the Study of Reading (Davison,
Kantor, et al. 1980) has shown that many of the devices used
in adapting a text to meet sentence-length, vocabulary, and
passage-length requirements contribute to a marked decrease in
the coherence and interest of the text. In addition, it is a
 Literary Style Discrimination in Five-Year-Olds / 161

basic principle of attention theory that perceptual activities

which demand more mental processing tend to be favored over
less demanding activities (Hardiman and Zernich 1978). Suc-
cessfully meeting a challenge is itself a source of pleasure and
satisfaction. If some 7-year-olds have trouble with grade-level
basal readers, it may be a problem of motivation; it may be
that they would do better on more complex, more difficult, more
challenging, more rewarding material.
If the ability to discriminate literary styles is general among
primary-grade children, then it may be that by editing their
readers to meet someone's preconceived notions of what is easy,
we are depriving children of the satisfaction of meeting a chal-
lenge, and contributing to making learning to read an un-
pleasant experience.
NOTES

Permission for use of illustrations and quotations in this paper

is gratefully acknowledged, as follows:
Illustrations:
Examples (1) and (2): By Lionel Kalish, from The Cat and
the Fiddler, by Jacky Jeter (copyright © 1968 by Parents Mag-
azine Press), and The Sheep of the Lai Bagh, by David Mark
(copyright © 1967 by Lionel Kalish), respectively. Englewood
Cliffs, N . J . : Four Winds Press. Used with permission.
Quotations:
Example (8a): From Virginia Kahl, The Baron's Booty. Copy-
right © 1963 by Virginia Kahl. (New York: Charles Seribner's
Sons, 1963) Reprinted with the permission of Charles Scrib-
ner's Sons.
Examples (8b), (8c): From Virginia Kahl, How Do You Hide
a Monster? Copyright © 1971 by Virginia Kahl. (New York:
Charles Scribner's Sons, 1971) Reprinted with the permission
of Charles Scribner's Sons.
Example (8d): From Virginia Kahl, The Perfect Pancake.
Copyright © 1960 by Virginia Kahl. (New York: Charles
Scribner's Sons, 1960) Reprinted with the permission of
Charles Scribner's Sons.
Examples (9a), (9b): From The Lorax, by Dr. Seuss. Copy-
right © 1971 by Dr. Seuss and A. S. Geisel. Reprinted by
permission of Random House, Inc.
Examples (9c) through (9g): From I Had Trouble in Getting
to Solla Sollew, by Dr. Seuss. Copyright © 1965 by Dr. Seuss.
Reprinted by permission of Random House, Inc.
Example (9h): From Happy Birthday to You! , by Dr. Seuss.
Copyright © 1959 by Dr. Seuss. Reprinted by permission of
Random House, Inc.
Example (10a): Text excerpt from The Runaway Bunny by
Margaret Wise Brown. Copyright © 1942 by Harper and Row,
Publishers, Inc. Text renewed © 1970 by Roberta Brown Rauch.
By permission of Harper and Row, Publishers, Inc.
162 / Georgia M. Green

Examples (10b), (10c): Text excerpts from Wait Till the

Moon Is Full by Margaret Wise Brown. Copyright 1948 by
Margaret Wise Brown. Text copyright renewed 1976 by Roberta
Brown Rauch. By permission of Harper and Row, Publishers,
Inc.
Example (lOd): Text excerpt from Little Fur Family by
Margaret Wise Brown. Copyright 1946 by Harper and Row,
Publishers, New York and Evanston. Renewed 1974 by Roberta
Brown Rauch and Garth Williams. By permission of Harper and
Row, Publishers, Inc.
Example (10e): From Fox Eyes, by Margaret Wise Brown.
Copyright © 1952 by Pantheon Books, Inc. Reprinted by per-
mission of Pantheon Books, a division of Random House, Inc.
Example ( l l a ) , ( l i b ) : From Beatrix Potter, The Tale of
Peter Rabbit. 1902. New York: Frederick Warne and Co.,
Inc. Used with permission.
Example ( l i e ) : From Beatrix Potter, The Tale of Two Bad
Mice. 1904. New York: Frederick Warne and Co., Inc. Used
with permission.
Examples ( l i d ) , ( l i e ) : From Beatrix Potter, The Tale of
Mr. Jeremy Fisher. 1906. New York: Frederick Warne and
Co., Inc. Used with permission.
Example (12a): From Bill Peet, Big Bad Bruce. Copyright
© 1977. Boston: Houghton Mifflin Co.
Examples (12b) through (12e): From Bill Peet, Eli. Copy-
right © 1978. Boston: Houghton Mifflin Co.
Example (13): From A Duck Is a Duck of the Reading 720
series by Theodore Clymer and others. © Copyright 1976,
1969, by Ginn and Company (Xerox Corporation). Used with
permission.
1. For testing kindergarteners' ability to recognize literary
style, we considered a number of tasks. A simple recognition
task, wherein a child would be asked if a passage had been
heard before, was rejected as not directly tapping the abilities
we wanted to test. A 2x2 forced-choice task (matching un-
familiar (or familiar) passages with familiar authors' names two
at a time) was rejected as not very informative, since making
one incorrect answer practically entailed making another, and
vice versa, one correct answer practically entailed making an-
other correct answer. A 2-out-of-3 (or more) matching task,
where a child would be asked to say which two passages out of
a group were by the same author, was rejected as logistically
unfeasible for nonreaders: the passages would have to be pre-
sented orally, and we judged that it would be asking too much
to ask children to remember three or more passages and their
order of presentation, in order to say which two were most
alike.
We wanted to make the task as difficult as we could and still
get better-than-chance performance so that it would test the
limits of the children's ability and so that the results would be
as informative as we could manage. For this reason, we settled
 Literary Style Discrimination in Five-Year-Olds / 163

on a l-out-of-5 multiple-choice style-matching task, with the

test materials containing as few non-style-related clues as possi-
ble.
2. Brown did not limit herself to 'lullaby' stories. A bizarre
and aggressive picture-book called The Steamroller: A Fantasy
(Walker Publishing Company, 1974) shows an entirely different
side of her.
3. In response to perceived demands from text book selection
bodies.
REFERENCES
Davison, A., R. N. Kantor, J. Hannah, G. Hermon, R. Lutz,
and R. Salzillo. 1980. Limitations of readability formulas in
guiding adaptation of texts. Technical Report No. 162.
Urbana: University of Illinois, Center for the Study of Read-
ing. ERIC Document Reproduction Service No. ED 184 090.
Hardiman, G., and T. Zernich. 1978. Basic research:
Aesthetic behavior in the visual arts. Viewpoints 6.1:29-38.
 TOPICS WITHIN TOPICS
Joseph E. Grimes
Cornell University and
Summer Institute of Linguistics.

When you and I talk, much of the time we talk 'about' some-
thing. Some of us here, myself included, are not too sure
what 'about' means, but we have a few observations we can
build on:
1. There are interesting properties of the form our
sentences take that seem to depend on what we are
talking about. Clefting would be one example.
2. Conversations and other discourses that do not
succeed in establishing some agreement on what
they are about do not get very far as vehicles
for communicating information, though they may
succeed as phatic speech acts.
3. Often, while talking about one thing, we put it
aside temporarily and talk about something else for
a while. After we finish with the new thing, we
may come back to the first one, but we do not
always do so.
4. What we talk about does not have to be a thing
in the physical sense or even in the grammatical
sense of being encoded as a noun. Places, times,
activities, and states of affairs are also candidates.
5. Linguists use terms like 'topic', 'theme', 'focus',
and 'foregrounding' for some of these things I
have just mentioned, but no two linguists use the
terms in the same way.
I invite you to concentrate with me on the third observation:
that while talking about one thing, we often put it aside and

164
 Topics Within Topics / 165

talk about something else before we return to pick up the

main point and finish whatever we set out to do regarding it.
I warn you that I am not far enough along in my under-
standing of this phenomenon that I can lay out very many sig-
nals for you, even for English, and that I am far from being
able to give an algorithm that can recognize shifts to embedded
topics or to following topics. Both are part of a total effort
that a number of people are engaged in. We hope it will
eventually show how referential orientation is communicated.
What I can do is to help you see that such a phenomenon is
plainly there, and that even now we can discern a consider-
able richness in the way it works. My argument is from
weaker to stronger: if we can learn this much about one
aspect of language with nothing but the crude concepts I have
used, we can certainly expect to get somewhere by refining
them some more.
Let me illustrate concretely this setting aside and resuming
of reference. Many of us studied Abraham Lincoln's Gettys-
burg Address of 1863 in school; we may even have memorized
it. There is broad agreement that whatever we may mean by
a well-formed discourse, this qualifies as one. Furthermore,
it is brief enough that it is not hopeless to understand, yet
the text is complex enough that our reward for doing so will
be something more than a sense of having found out the ob-
vious. Schank and Carbonell (1978) have called attention to
some of the complex inferences that have to be made in order
to understand this speech; I focus my attention here on differ-
ent aspects.
The text itself, with the sentences numbered, is given in
the appendix to this paper. The spelling and punctuation
reproduced there are from Lincoln's own hand, as preserved
in the Cornell University Library's copy. Asterisks indicate
points of audience reaction, taken from a New York Tribune
account reproduced in Everett (1863).
This text is typical of a large class of texts in which the
things we are talking about at any point have properties of
identity, place, and time, and may be indexed by any combi-
nation of these. The identities in this text are not hard to
sort out. There are the speaker and around 15,000 hearers,
taken as a group. There is the nation with which that group
is affiliated. There are soldiers who fought during a period
of war, some of whom were killed. The world, in the sense
of its inhabitants, is mentioned.
Spatial reference is not as prominent in this text as it is in
some. There is the continent on which the nation was estab-
lished, a battlefield, and 17 acres out of that field that in-
clude the actual site of the speech. At the end there is a
reference to the earth as a whole.
The most complex part of the referential system is time.
Time references center around the war period already referred
to, and in particular around two events within that period: a
166 / Joseph E. Grimes

significant battle, and the event of the speech itself four and
a half months after. In addition, there is the time when the
nation to which all the participants belong began, 'fourscore
and seven years' before the time of the speech.
Related to these are four time periods that extend without
limit into the future but share a boundary on the other end
with one of the concrete events. The first begins with the
founding of the nation and refers to the things its founders
were trying to accomplish in what to them was the future.
The second is to begin as soon as the occasion of the speech
ends and has to do with its results. The third began when
the speech did and has to do with attitudes that the speaker
and his hearers share and that the speaker is encouraging
them to continue to hold. The fourth is expected to begin
when the period of war ends.
Another unbounded period reaches back into the past by
way of gnomic or timeless reference--something that is inde-
pendent of any particular time framework.
Figure 1 diagrams the topology of the time framework, using
Litteral's conventions. On that time line 'before' lies to the
left and 'after' to the right, and significant stretches of time
numbered with even numbers are separated by vertical bound-
ary marks numbered with odd numbers, with no attempt to
represent actual duration.

Figure 1. Topology of time in Lincoln's Gettysburg

Address.

11 13 15
8 10 12 14 16

NEW I |BATTLE! SPEECH

NATION July 1-3, November 19,
1776 1863 1863
CIVIL WAR 1861-? [1865]

When each portion of the text is related to this overall

referential framework, its parts sort themselves into four con-
nected subsystems. In establishing these subsystems, time
reference seems to be the most important factor, though I
cannot yet define what I mean by 'most important' with either
a structural or a quantitative measure.
The first group is in Figure 2. It consists of those refer-
ences for which time is not a factor. Their time reference is
indicated by a single unbounded line that can be identified by
its first and last parts as (0,16). Sentence numbers in paren-
theses identify fragments that illustrate the time reference.
 Topics Within Topics / 167

Figure 2. Atemporal elements in*Lincoln's Gettysburg

Address.

9 11 13 15
8 | 10 12 I 14 I 16
NEW BATTLE SPEECH!
NATION
the proposition that all men are created equal (1) any
nation so conceived and so dedicated (2) in a larger
sense (6)

The second group is in Figure 3. It has to do with the

nation: its founding (1,3), the intent of its founders (3,16),
its testing (5,15), and its future (15,16). Here are a set of
time references that share common boundary lines, including
also the time displacement (3,11) and the possible end of the
nation (3,15).

Figure 3. The NATION complex in Lincoln's Gettysburg

Address.

3 5 7 11 13 15
4 I 6 8 10 I 12 14 I 16
NEW BATTLE SPEECI
NATION

brought fourscore and seven

forth years (1)
conceived
dedicated
(1) can long endure (2)
new
birth
engaged in ... war (2) not
testing (2) perish
of that war (3) (10)
unfinished work (9)
cause (10)
(10)
government of the people
168 / Joseph E. Grimes

The third referential group is in Figure 4. It has to do

with the battle Lincoln i s speaking of ( 7 , 9 ) , along with its
short-range outcome (9,11) and its long-range consequences
(.9,16).

Figure 4. The BATTLE complex in Lincoln's Gettysburg

Address.

3 5 7 11 13 15
2 I 4 I 6 I 10 12 I 14 I 16

NEW | BATTLE SPEECH!

NATION

gave l i v e s (4) n a t i o n might l i v e ( 4 )

struggled (7) have c o n s e c r a t e d ( 7 )
did (8) not . . . i n v a i n ( 1 0 )
fought (9)
gave (10)
died (10)
thus far . . .
advanced (9)

Finally, Figure 5 gives the fourth referential group. It has

to do with the occasion of Lincoln's speech (11,13) and its
expected outcome (11,16) and (13,16).
Now that we have an idea of the referential framework we
are dealing with, we can begin to look at the way one referen-
tial complex within it is laid aside temporarily for another,
then picked up again later. This speech begins and ends in
what we could call the referential complex of the NATION,
given in Figure 3. The first sentence makes implicit reference
to the situation by speaking, via the phrase ending in ago
that moves the reference back to the time of the founding of
the nation. This displacement away from the situation of
speaking has been commented on for other types of text, in-
cluding Labov and Waletzky's (1967) analysis of narratives
done 14 years ago.
The second sentence moves into the middle part of the
referential complex of the NATION by two paths. The first
is through a return to the situation of speaking using now.
along with the situational expression are engaged, which is in
the present tense and defines a time span broader than that
of the situation of speaking, namely, the Civil War. That
period had begun before the speech started, and it was not
yet over when the speech was given.
 Topics Within Topics / 169

Figure 5. The SPEECH-related complex in Lincoln's

Gettysburg Address.

1 3 5 7 9 11 13 15
0 I 2 I 4 I 6 I 8 I 10 I 12 I 1 4 I 16
NEW I BATTLE SPEECH
NATION

be dedicated (9, 10)

take (10)
highly resolve (10)

met (3) dedicate (4)

have come (4) resting
fitting (5) place (4)
can not (6)do this (5)
add or detract (7) dedicate/
say (8) consecrate/
hallow (6)
note (8)
remember (8)
never
forget (8)
task (10)

The second approach takes off from the preceding reference

to the NATION, and speaks of the possible termination of that
entity in the words can long endure. There is a complex pro-
cess of inference that ties the termination of the Civil War
with the possible termination of the concept of the NATION;
this is the point where a precise statement of what is going
on gets sticky, but where we ordinary users of language not
unreasonably recognize that a referential connection does exist.
Following the second sentence there are only three minor
references to the NATION complex until the end. They are
all embedded grammatically within sentences whose major refer-
ence is laid but in Figure 5, associated with the SPEECH.
The three minor references: in Sentence (3), of that war
refers briefly to the NATION complex; in Sentence (9), the
unfinished work refers to the sequel to the founding of the
NATION, which Lincoln hopes will outlast the war itself; and
in Sentence (10b), that cause refers to the same sequel.
At the end, however, and at a point where the audience
responded overtly, the whole reference pattern shifts back
to the point in time where Sentence (2) left off. At the time
indexed by 15, which is later than the SPEECH and at which
170 / Joseph E. Grimes

it will be possible to confirm or deny the content of the words

can long endure in Sentence (2), the last two segments of
Sentence (10) pick up the hoped for continuity of the referen-
tial complex of the NATION and the idea behind it: that this
nation, under God, shall have a new birth of freedom—and
that government of the people, by the people, for the people
[the manifestation of the NATION concept] shall not perish
from the earth.
In other words, the referential pattern of the speech as a
whole takes the perspective of the nation: its founding, its
uncertain continuation, and hope for its perseverance. This
referential pattern, however, is laid aside in Sentences (3) to
(10c), which talk about more immediate things, as Figure 6
shows.
Figure 6. Suspension and resumption of reference in
Lincoln's Gettysburg Address.

NATION
brought can it shall not
forth (I) endure perish (10d-e)
(2)
SPEECH
met to ... dedicate
dedicate / ^ X ourselves
(3-6) / \ (8-IOc)
BATTLE
have
dedicated
(7)
In this central section, all but one of the matrix sentences
belong to the SPEECH complex of Figure 5. Sentences (3),
(4), and (5) are stative and perfective in form and refer to
the immediate occasion: for example, the time reference of
 Topics Within Topics / 171

we are in (3) is only part of the temporal scope that is associ-

ated with the same words in (2). Sentence (3) provides a
transition from the NATION orientation of the beginning sen-
tences. The words of that war in Sentence (3), which I take
it have to be read as part of the tail of the intonation, in
Ladd's (1980) terms, are the context within which battlefield
and we are met point to circumscribed events, the BATTLE
complex and the SPEECH complex of Figures 4 and 5, respec-
tively.
Sentences (3), (4), and (5), and I think the word but with
which (6) begins, anchor to the occasion of the speech itself.
Sentence (3) places that occasion in its larger context, while
(4) through (6) continue on to talk about what is to happen
following the speech, which in the case of (6) also includes
the activity of the speech itself.
At this point there comes what I have elsewhere called a
composite operation in the progression of changes in participant
orientation. The relation of the participants in Sentences (3)
through (6) is 'We...are doing something about a place...for
those who fought'. Sentence (7) reverses this orientation:
'Those who fought... did something about that place.. .regard-
less of us'. In the study of participant orientation systems,
I have noticed a high correlation between composite shifts of
orientation like this one and surprise points (Grimes 1975:
266).

Figure 7. Changes in participant orientation, adapted from

Grimes (1975:264).

srs (we...place...those who fought)

A B
= (those who f o u g h t . . . p l a c e . . . u s )
C B A
ABC

r / \s r (reversal)
BACX XACB (12) ( 3 )
I I ABC •BAC
s r
I J s (switch)
BCA\ /CAB (1)(23)
r \ / s ABC »-ACB

The next four sentences, taking the parts of the last

orthographic sentence as separate divisions, all have matrices
that refer to what is likely to happen after the SPEECH.
172 / Joseph E. Grimes

Sentence (8) is referentially a recapitulation of (4) through

(6) in its first half, and of (7) in its second half.
Sentences (9) and (10b) are parallel to each other. They
organize the same referential complexes in the same relation.
Each touches base with the main triple: the SPEECH and its
sequel, the unbounded sequel to the founding of the NATION,
and the BATTLE period. Sentence (9) also fills the gap be-
tween BATTLE and SPEECH, which is otherwise untouched,
in the words have thus far so nobly advanced. Either of
these sentences could stand as a minimal statement of the con-
tent of the speech by virtue of bringing together its three
referential complexes; yet if either were taken to be the sum-
mary, that would miss the perspective of the speech because
neither of these sentences has the NATION complex as its
framework, and the speech as a whole has.
Sentence (10c) appears to provide the transition from the
center section oriented around SPEECH references back to the
outer NATION complex. It does so by having the SPEECH
and its sequel be the referents for the matrix sentence, then
picking up the unbounded sequel to the BATTLE. This open
segment (9,16) shares its final segment, the unbounded one,
with the remainder of the speech.
I have laid out a fairly complicated set of interrelationships
that I have noticed so far in the referential structure of one
text. I have done it without recourse to any particular theo-
retical model. I have found none yet that is quite rich
enough to allow me to sort out what I observe, though some
of the proposals of our next speaker [Teun van Dijk] offer a
partial model. To me, recognizing these relationships brings
up several questions that ought to be asked:
1. Is this linguistics or something else?
2. Could anybody else get a comparable analysis out of
the same data?
3. What have we got when we finish?
4. Have we actually proved anything about one referential
complex being laid aside while another takes over,
followed by reinstatement of the first?
5. Do all texts behave this way?
Since I for one have never thought of linguistics as some-
thing sharply circumscribed, but rather as a characteristic
way of thinking about language, I would say that I am oper-
ating somewhere near the edge of what others might restrict
linguistics to, but I am doing so as a linguist would. Spe-
cifically, I am not making assertions about language that are
not required by the kinds of forms that are actually said, nor
am I making assertions for which no parallels can be found in
other texts or in other languages.
 Topics Within Topics / 173

As for whether other people can be brainwashed into seeing

things my way well enough to get a comparable analysis out of
the same data, all I can say is that I think I am closer to it
than the last time I tackled this subject (Grimes in press).
The results of this kind of analysis are a more precise idea
of what referential elements are being dealt with at each stage
of a text, together with a notion of suspension and resumption
of a set of referents, indicated in some cases (but not in all)
by transitional sentences that tie one referential complex with
another. Whether this hypothesis about nesting one referential
complex inside another is confirmed by texts like this one is,
of course, open to interpretation; one alternative would be
that we simply ramble around from one sentence to another,
and occasionally meander back to some referential complex we
have already taken up, perhaps with some broad constraints
to the effect that we ought to try to end up in the same area
we started out from.
This is possible, but I think an argument for nesting can
be constructed from those cases in which referential shift goes
along with grammatical subordination. I would ask why any
language needs to have any mechanism of subordination at all
in order to carry out its communicative tasks. Most of the
shifts in reference that are made within sentences in this text
and others I have examined coincide with the boundaries of
grammatically dependent or embedded syntactic units. I would
take it, then, that subordinate structures could be thought of
as one explicit means of communicating referential shift.
So far, the differences I have found among texts involve
two things: the kinds of elements on which I find referential
continuity or discontinuity, and the depth of nesting. In a
movie review from Time magazine, for example, I find that all
the referential shifts are on identifiable elements: the script,
the actors, the color print, and so on. The only shifts that
involve time or place are embedded within sentences, and have
nothing to do with the text as a whole or with major sections
of it. On the other hand, in Koine Greek texts I find con-
siderable suspension of one referential complex in favor of
another, and that of another, and so on until the chain of
resumptions starts up in the other direction. Much of this
nesting is made explicit by grammatical embedding and subordi-
nation, with the result that the average syntactic depth of
Greek sentences is greater than would be natural for English.
So we are in the intriguing position of being able to recog-
nize the temporary replacement of one set of referents by an-
other. To a certain extent we can match this with shifts in
syntax; but we have some distance to go before we can say
we really understand either the mechanisms of this replace-
ment or the part it plays in human communication.
174 / Joseph E. Grimes

APPENDIX

Gettysburg Address
Abraham Lincoln
November 19, 1863
(1) Four score and seven years ago (3,11 NATION related
to SPEECH)
our fathers brought forth, on this continent, a new
nation, conceived in Liberty, and dedicated to
the proposition (1,3 NATION)
that all men are created equal. (0,16 TIMELESS)*
(PARAGRAPH)
(2) Now we are engaged in a great civil war, testing
whether (5,13 WAR to limited sequel to NATION)
that nation, (3,15 NATION + limited sequel)
or any nation so conceived, and so dedicated, (0,16
TIMELESS)
can long endure. (3,15 NATION + limited sequel)
(3) We are met on a great (11,13 SPEECH)
battle-field (7,9 BATTLE)
of that war. (5,15 WAR)
(4) We have come (11,13 SPEECH)
to dedicate a portion of that field, as a final
resting-place for (13,16 sequel to SPEECH)
those who here gave their lives, (7,9 BATTLE)
that that nation might live. (9,16 sequel to BATTLE)
(5) It is altogether fitting and proper (11,13 SPEECH)
that we should do this. (13,16 sequel to SPEECH)
(PARAGRAPH)
(6) But, (11,13 SPEECH)
in a larger sense, (0,16 TIMELESS)
we can not dedicate--we can not consecrate--we
can not hallow--this ground. (11,16 SPEECH +
sequel)
(7) The brave men, living and dead, who struggled
here, (7,9 BATTLE)
have consecrated it (9,16 sequel to BATTLE)
far above our poor power to add or detract. (11,13
SPEECH)*
(8) The world will little note, nor long remember (13,16
sequel to SPEECH)
what we say here, (11,13 SPEECH)
 Topics Within Topics / 175

but it can never forget (13,16 sequel to SPEECH)

what they did here. (7,9 BATTLE)*
(9) It is for us the living, rather, to be dedicated
here to (11,16 SPEECH + sequel)
the unfinished work which (3,16 unlimited sequel to
NATION)
they who fought here (7,9 BATTLE)
have thus far so nobly advanced. (9,11 sequel to
BATTLE up to SPEECH)*
(10a) It is rather for us to be here dedicated (11,16
SPEECH + sequel)
to the great task remaining before us (13,16
sequel to SPEECH)
(10b) --that from these honored dead we take increased
devotion to (11,16 SPEECH + sequel)
that cause for which (3,16 sequel to NATION)
they here gave the last full measure of
devotion (7,9 BATTLE)
(10c) --that we here highly resolve (11,16 SPEECH +
sequel)
that these dead (9,16 sequel to BATTLE)
shall (11,16 SPEECH + sequel?)
not have died in vain (9,16 sequel to BATTLE)*
(lOd) --that this nation, under God, shall have a new
birth of freedom (15,16 sequel to WAR)
(lOe) --and that (15,16 sequel to WAR)
government of the people, by the people, for the
people, (3,16 sequel to NATION)
shall not perish from the earth. (15,16 sequel to
WAR)*
REFERENCES

Everett, Edward. 1863. An oration delivered on the battle-

field of Gettysburg. New York: Baker and Godwin.
Grimes, Joseph E. 1975. The thread of discourse. The
Hague: Mouton.
Grimes, Joseph E. (in press) Reference spaces in text.
In: Nobel symposium on text processing. Edited by
Sture Allen. Gothenburg: SprSkdata.
Labov, William, and Joshua Waletzky. 1967. Narrative
analysis: Oral versions of personal experience. In:
Essays on the verbal and visual arts. Edited by June Helm.
Seattle: American Ethnological Society and University of
Washington Press.
176 / Joseph E. Crimes

Ladd, D. Robert, Jr. 1980. The structure of intonational

meaning. Bloomington: Indiana University Press.
Lincoln, Abraham. [1863] The Gettysburg address of Abra-
ham Lincoln: Holograph manuscript presented by George
Bancroft with accompanying letter by Lincoln, February 29,
1864. MS.
Litteral, Robert. 1972. Rhetorical predicates and time
topology in Anggor. Foundations of Language 8.391-410.
Schank, Roger C. , and Jaime G. Carbonell, Jr. 1978. Re:
the Gettysburg address: Representing social and political
acts. Yale University Department of Computer Science,
Research Report 127.
van Dijk, Teun A. 1977. Text and context: Explorations
in the semantics and pragmatics of discourse. London:
Longman.
 EPISODES AS UNITS OF DISCOURSE ANALYSIS

Teun A. van Dijk

University of Amsterdam

1. Units of discourse analysis. One of the tasks of a sound

theory of discourse is to explicate the analytical units postu-
lated in the abstract description of textual structures at vari-
ous levels. In addition to the usual morphophonemic, syntactic,
semantic or pragmatic units or categories of sentence grammars,
the theory1 of discourse has introduced new notions, such as
'coherence , 'cohesion', 'topic', or 'theme1--explicated in so-
called 'macrostructures', whereas the analysis of conversation
makes use of such notions as 'turn' or 'move'.
Thus, more or less at a 'meso-level' in between the unit of
a clause or sentence on the one hand, and the unit of a text,
discourse, or conversation as a whole, the notion of 'para-
graph' or 'episode' has recently been discussed in various
branches of discourse analysis (Chafe 1980, Longacre 1979,
Hinds 1979). Roughly speaking, paragraphs or episodes are
characterized as coherent sequences of sentences of a dis-
course, linguistically marked for beginning and/or end, and
further defined in terms of some kind of 'thematic unity'--for
instance, in terms of identical participants, time, location or
global event or action.
In this paper I would like to contribute to a further defini-
tion of the notion of paragraph or episode, and will thereby
focus on their semantic properties. It is argued here that an
explicit account of these notions also requires a characteriza-
tion in terms of semantic macrostructures, of which the various
'surface manifestations' often function as typical paragraph or
episode markers.
For the sake of theoretical clarity, I make a distinction be-
tween the notion of 'paragraph' and the notion of 'episode'.
An episode is properly a semantic unit, whereas a paragraph
is the surface manifestation or the expression of such an

177
178 / Teun A. van Dijk

episode. Since I would like to pay attention especially to se-

mantic issues, the discussion is mainly about episodes, rather
than about paragraphs and their grammatical properties (as
have been studied by Longacre 1979, Hinds 1979, and others).
Although this paper has a predominantly 'structuralistic'
nature, the assumption that episodes are semantic units raises
the possibility that they also have psychological relevance, as
units in a cognitive model of discourse processing. Recent
work in that area seems to suggest, indeed, that episode-like
units have processing relevance in reading, representation,
arid memorization of discourse (see, for example, Black and
Bower 1979; Haberlandt, Berian, and Sandson 1980). It is
briefly shown at the end of this paper that my linguistic/
semantic observations may indeed be relevant for a cognitive
discourse model.
Before I begin my discussion, a methodological remark is in
order. There has been considerable controversy--mostly out-
side discourse analysis or text grammar, and often directed
against these--about the linguistic and in particular the gram-
matical status of postulated discourse categories or units. In
order to keep linguistics and especially grammar nice and
clean, not only have many linguists preferred to remain within
the seemingly safe boundaries of the sentence, but at the same
time they have tried to discredit as linguistically or grammati-
cally 'foreign' most of the specific units, categories, or levels
used in various kinds of discourse analysis, admitting these at
most to a theory of language use, to pragmatics, to rhetoric,
or to other theories or disciplines outside their scope of re-
sponsibility.
The style of the last sentence suggests that I do not share
that opinion. It is certainly true that many properties of dis-
course cannot and should not be accounted for in the format of
a linguistic grammar; for example, rhetorical or narrative struc-
tures require separate—but integrated--treatment. Yet, many
other discourse phenomena are properly linguistic or even 'gram-
matical' , that is, they can be fully accounted for in terms of
the usual levels, categories, or units that are familiar in the
account of sentences. This does not mean, of course, as some
have suggested, that therefore a linguistic theory of discourse
can safely be reduced to that of a theory of sentences (plus
some theory of language use, a pragmatic or a cognitive model).
New notions and specific phenomena, such as coherence, macro-
structure, or episode, are certainly necessary, but they can
be described in terms of familiar theoretical notions.
Although I do think that linguistics and grammar have spe-
cific tasks and hence theories of their own, I would like to sug-
gest more generally that the boundaries between 'grammar' and
other linguistic theories, or between linguistic accounts of lan-
guage and language use, and psychological or sociological ones,
are not and should not be too sharp. In a general functional
approach to language (Dik 1978, Givon 1979a, 1979b) which is
 Episodes as Units of Discourse Analysis / 179

now a central paradigm in linguistics, it is stressed that on the

one hand units, categories, rules, and structures at one level of
analysis are systematically linked with those at other levels,
and that on the other hand, all these 'linguistic' structures are
functionally linked (both ways: determining and depending on)
to the cognitive and social processing and use of language in
communicative interaction. Thus, properties of the sentence--
morphophonemic, syntactic, semantic, and pragmatic—appear
also to have functions within the discourse (Givon 1979b). The
same holds for a theory of the episode. Some properties of
episodes can conveniently be formulated within a linguistic and
even a grammatical framework; others need additional or alter-
native description in cognitive, interactional, and social terms.
If attention is focused on the first, it should be kept in mind
that the theory is essentially partial, and that the interdepen-
dencies with the cognitive and social processes and functions
are an indispensable further explanation of the 'grammatical'
structures of a more abstract account of episodes in discourse.

2. Intuitive notions of 'episode 1 . The notion of episode oc-

curs not only in a theory of discourse, but also in everyday
discourse. We speak about an 'episode' of our life, an 'epi-
sode' during a party, an 'episode' in the history of a country,
or about episodes in stories about such episodes. In this
sense an episode is first of all conceived of as a part of a
whole, having a beginning and an end, and hence defined in
temporal terms. Next, both the part and the whole mostly in-
volve sequences of events or actions. And finally, the episode
should somehow be 'unified' and have some relative indepen-
dence: we can identify it and distinguish it from other epi-
sodes. Thus, a war can be an episode in the history of a
country, a battle an episode in a war, and some brave action
of a group of soldiers an episode during the battle. Appar-
ently, the 'unifying' aspect of such sequences of events or
actions conceptually appears in global event or action notions,
such as 'war', 'battle', 'attack', and so on, as well as in the
identity of the participants of such events and actions (a
country, armies, groups of soldiers, individual persons, e t c . ) ,
and finally, in the temporal identification of beginning and end.
This intuitive notion of episode corresponds to the notion of
episode in a story or account of such actions and events: one
speaks about an episode in a novel or in a history textbook,
and the meaning of that notion is similar to its corresponding
world episode: a sequence of sentences (or of propositions
expressed by such sentences) denoting such an episode, hence
with marked beginning and end and some conceptual unity. It
is this intuitive notion that has been taken over in the theory
of discourse. In this paper, then, I try to explicate this theo-
retical notion, and investigate whether it can be applied also to
other discourse types, that i s , not only to event or action
discourse such as stories.
180 / Teun A. van Dijk

3. The semantics of episodes. Since episodes are taken to

be semantic units of discourse, one must be able to define them
in semantic terms, for example, in terms of propositions. I
will indeed so do, and characterize an episode of a discourse as
a specific 'sequence of propositions'. Just like the discourse as
a whole, such a sequence must be coherent according to the
usual conditions of textual coherence (van Dijk 1972, 1977).
That is, the respective propositions should denote facts in some
possible world, or related possible worlds, that are--for exam-
ple, conditionally--related. Besides this so-called local co-
herence, the sequence should be globally coherent, that is, be
subsumed under some more global macroproposition (van Dijk
1972, 1977, 1980). Such a macroproposition explicates the over-
all unity of a discourse sequence as it is intuitively known under
such notions as 'theme', 'topic', or 'gist'. Macropropositions are
derived from sequences of (local, textually expressed) proposi-
tions of a discourse by means of some kind of semantic mapping
rules, so-called macrorules, which delete, generalize, or 'con-
struct' local information into more general, more abstract, or
overall concepts.
These macrorules are recursive, so that one may have several
layers of macroproposition sequences, together forming the
macrostructure of a discourse. Such a macrostructure can
typically be expressed by (individually varying) summaries of
a discourse. Macrorules not only have textual propositions as
input, but since much of the information of a text is implicit
for pragmatic reasons also need information from the 'knowledge'
and 'belief sets of language users, for example, from frames or
scripts which organize this knowledge according to criteria of
stereotypical usage. By definition, a macroproposition features
a central predicate and a number of participants, denoting
either an important or global property, event, or action and
central participants in a discourse. The textual 'basis' of each
macroproposition, thus, is a sequence of propositions of the
discourse. It is precisely this sequence which we call an 'epi-
sode'. In other words, an episode is a sequence of proposi-
tions of a discourse that can be subsumed by a macroproposi-
tion.
Because macrorules operate recursively, and we therefore may
have macropropositions at several levels of generality, there may
also be episodes of varying length or scope in a discourse.
Theoretically, even the discourse as a whole, as a limiting case,
is an episode. The overall 'unity' we intuitively postulated for
world episodes and discourse episodes is precisely defined by
the subsuming macroproposition(s). The beginning and end of
an episodic sequence are then theoretically defined in terms of
propositions which can be subsumed by the same macroproposi-
tion, whereas the previous and the following proposition of,
respectively, the first and the last proposition of an episodic
sequence should be subsumed by another macroproposition.
 Episodes as Units of Discourse Analysis / 181

Further on, I show that these 'breaking points' are interest-

ingly marked by linguistic (and other) means.
Although there may be episodes of varying length or scope
in a discourse, it might be more relevant to restrict the notion
of an episode to those sequences which have some specific fur-
ther 1properties, intuitively characterized in terms of 'impor-
tance . That i s , perhaps lowest level macropropositions do not
always define what is intuitively called a discourse episode, so
that only higher level and sometimes nonreducible macroproposi-
tions subsume textual episodes. It must be seen, therefore,
whether there are additional constraints on the identification of
episodes.
The theoretical relevance of the notion of episode first of all
lies in the fact that we now have a text-base unit correspond-
ing to the earlier notion of macroproposition, that i s , the se-
quence of propositions from which the macroproposition is de-
rived. Secondly, one may assume that this textual unit has
cognitive and linguistic properties. As the theory predicts
and as has been confirmed in descriptive analysis (Chafe 1980,
Longacre 1979, Hinds 1979, and the analysis to be given fur-
ther on in this paper), the following grammatical 'signals' may
be expected for the beginning of episodes:
1. pauses and hesitation phenomena (fillers, repetition) in
spoken discourse;
2. paragraph indentations in written discourse;
3. time change markers: in the meantime, the next day,
etc. and tense changes;
4. place change markers: in Amsterdam, in the other room;
5. 'cast' change markers: introduction of new individuals
(often with indefinite articles) or reintroduction of 'old'
ones (with full noun phrases instead of pronouns);
6. possible word introducing or changing predicates (tell,
believe, dream, etc.);
7. introduction of predicates that cannot be subsumed under
the same (macro-) predicate, and/or which do not fit the
same script or frame;
8. change of perspective markers, by different 'observing'
participants or differences in time/aspect morphology of
the verb, (free) (in-)direct style.
Such markers signal the beginning of a new episode and hence
at the same time the end of a previous one. In other words,
as soon as there is a change of time and place (a scene), a
different cast of participants, and a different global event or
action now being initiated--according to scriptal or frame-like
world knowledge about the components of such events or actions
--one may assume that there is a beginning of a new episode.
It goes without saying that such markers play an important role
in a cognitive model for the strategies of discourse
182 / Teun A . van Dijk

comprehension in which the language user has to derive a

macroproposition from the propositions in the text.
In the foregoing I have assumed that perhaps not every
macroproposition would qualify as a good candidate for defining
episodes. In a story about a party, the sequence of events
and actions that constitute the beginning of the story, and
which can be subsumed under such macropropositions as 'I was
invited to Peter's party', or 'I left', or 'I arrived at Peter's
party', would not usually be qualified as episodes. Of course,
the theoretical notion of an episode could have wider applica-
tion, but I prefer to explicate an intuitively relevant concept.
On the other hand, one does speak of an episode, both in the
discourse and in the denoted world, if I get drunk at the party,
or if as a consequence of getting drunk I have a car accident
afterwards. In that case, the episode seems to have a specific,
e.g. narrative, function in the discourse. In my example, for
instance, it would be the interestingness criterion defining the
Complication category of a story. The initial macropropositions
would also have a function, e.g. that of a Setting, but they
rather define the theoretical correlate of an episode, like a
scene or background of other episodes, such as place, time,
participants, and so on.
In addition, it seems that in order to be able to really define
and identify a sequence, the global events or actions should
not, as such, be stereotypical or normal--according to our world
knowledge and beliefs. And finally, those global actions or
events that are only preparations for or components of more
global and interesting actions and events, should also not be
identified as episodes. 'Leaving to go to a party' is not a goal
on its own but part of a higher level action (going to a party
and participating in i t ) . Getting drunk or having an accident,
however, is not stereotypical and hence may be qualified as
episodic. It follows that those macropropositions have episodic
nature (i.e. define a textual episode) which are not stereotypical
according to scripts or frames, which cannot be subsumed by
higher level macropropositions, and which have a specific func-
tion in the discourse as a whole. In other words, episodes
typically require global goals of participants, or actions and
events that frustrate, thwart, or menace the realization of such
goals, so-called 'incidents'. Thus, studying psychology may be
a global goal in my life and hence the sequence of events or
actions defined by it may be an 'episode' of my life, whereas
the incident of flunking an exam or being seduced by my
teacher may be an episode within this more general episode:
this holds both for world episodes and for the episodes in the
discourses about them.
We now have a number of specific semantic criteria--and some
brief suggestions for surface manifestations of these constituting
the episode markers--for the identification of episodes in a dis-
course. Yet, just as intuitive episodes are usually distinguished
especially in event and action sequences, it seems that especially
 Episodes as Units of Discourse Analysis / 183

event or action discourses, e.g. stories, have episodes. How

about other discourse types?
I am going to show that 'newsstories' also have episodic struc-
ture: they are about (locally, nationally, or internationally)
important events, feature important participants, are not stereo-
typical or only in part predictable, and can be identified in
time and place.
Similarly, in history textbooks, parts of the discourse may be
episodes because they are about important historical events and
actions of a country or of the world: that i s , actions or events
that have broad social, economical, political, or cultural conse-
quences, and which are characterized by the same cast of par-
ticipants and limited in time and place.
Whether poems, advertisements, psychological theories, or
concert programs have 'episodes' of this kind, however, remains
to be seen. These certainly have (functional) 'parts', but they
are not necessarily defined in terms of a global event or action,
a cast of participants, or identical time or location parameters.
Further research is necessary to see whether episode-like units
can or should be identified for these and other discourse types.
So my observations provisionally hold only for the large class of
event and action discourses, of which the set of stories is only
a subset.
4. Analysis of an example: A newsstory. To make the theo-
retical remarks of the previous sections more concrete, let us
analyze a text sample in terms of episodic structure. I have
chosen a newsstory from Newsweek about the American foreign
policy in Latin America after the election of Reagan as presi-
dent. This article (see Appendix) is mainly about the various
opinions, both in the United States and in Latin America, about
this assumed foreign policy. This means that one cannot simply
analyze the text in terms of events and actions, but rather
should also account for different 'opinions'. Since, however,
different participants, different opinions, and different 'loca-
tions' are involved, an episodic analysis seems possible.
Table 1 lists the episodes, with the sentences which are part
of them, and the respective segmentation criteria. Since there
may be several layers of macrostructure, one may also distin-
guish different episodes, which are given in Tables 2 and 3,
respectively. In this way the total number of episodes drops
from 39 to 22 and to 13. The latter 13 episodes seem to corre-
spond with the 13 nonreducible macropropositions of the text.
Macropropositions 1 and 2 are, as usual in newsstories, the
more general 'summaries', highest in the macrostructure, of
which the other macropropositions are specifications.
At the level of surface structure one may observe that these
13 macropropositions and their episodes correspond more or less
with the 11 paragraphs of the text.
184 / Teun A. van Dijk

Table 1. Episode segmentation of 'A new team's Latin test'.

Episode:
lines Sentences Segmentation criteria
1:4-7 Nowhere.. .Latin America General thematic intro-
duction: USA
2:7-8 And nowhere.. .passion General thematic intro-
duction : LA
3:8-11 Many governments... Specification: attitude
White House LA governments
4:11-15 In Chile...friends Specification: attitude
government Chile
5:15-20 and on a tour. ..found Specification: La capi-
it tals/Rockefeller (USA)
6:21-23 Most human-rights... Specification: HR activ
region vists
7:24-27 Leaders in Cuba... Specification: LA left
Washington. countries
8:27-31 But for R. administra- Reagan administration
tion. . . .on the list policy
9:31-35 'The most...policy' Specification: statements
Kirkpatrick
10:36-43 Traditional wisdom... R's policymakers: the
K. herself. team
11:43-51 Kirkpatrick.. .practices Kirkpatrick: presentation
and policy
12: 52-54 Reagan's... governments Reagan's traditional policy
13:55-57 'For four.. .Buenos Official opinion Argentina
Aires'
14:57-68 'That will end.. .respect' Opinion Kirkpatrick:
better relations, respect
15:69-72 Among the Carter... Opinion Carter supporters
liberalize in LA
16:72-76 Said Eduardo.. .every- Specification: Opinion
where Brazilian C. supporter
17:76-92 Kirkpatrick.. .C.'s HR- Opinions K.: realistic HR
policy. policy
18:93-97 Green Light.. .Meza Foreign ministries in LA:
Bolivia policy USA?
19:97-104 'It would be...from us' Opinion Bolivian diplomat
in exile
20:104-110 Reagan aides. ..nations' Opinion aides, K. about
Bolivia
21:111-113 The real test.. .Carib- General statement: policy
bean in Caribbean
22:113-120 The problems.. .Belize K's opinion about Carib-
bean
23:120-121 Some LA officials... Opinion LA officials about
activist. Reagan
 Episodes as Units of Discourse Analysis / 185

Table 1. Continued.

Episode:
lines Sentences Segmentation criteria
24:122-124 The presidents.. .region Specification: Opinion
presidents M. and P.
25:124-129 And one analyst... Opinion analyst in Rio
America about anti-communism
26:129-132 But.. .them Opinion Kirkpatrick:
help against communism
27:133-136 Fourth Place.. .sales General statement: how
does R. help?
28:136-144 'Our whole.. .demili- K's opinion: sell arms
tarize.'
29:145-151 The R. administration... General statement: will
conditions R. help unfriendly
nations?
30:151-155 We must.. .United States Opinion K.: only under
strict conditions
31:156-157 Moderating influence... General statement: this
controversial is controversial
32:157-171 Some analysts.. .Soviet Specification: opinion
Union some analysts
33:172-174 Similarly... Cuba General statement: policy
about Cuba
34:174-177 Some LA e x p e r t s . . . Specification: opinion LA
Havana experts
35:177-182 On the basis. ..Third Opinion Newsweek: un-
World likely
36:183-188 Some LA's worry... Attitude in LA
exists
37:188-191 "People.. .diplomat Opinion LA diplomat
38:191-199 But Kirkpatrick.. .big Opinion K: democratic
way regimes get help
39:199-204 Such talk.. .action Evaluation Newsweek
186 / Teun A. van Dijk

Table 2. Second level episodes in 'A new team's Latin test'.

Episode 2:
lines Segmentation criteria
1:4-7 General summary statement: USA foreign policy
change in LA
2:7-8 General summary statement: reactions in LA to
R's election
3:8-20 LA governments: relief
4:21-27 Human rights activists' opinion: severe setback
5:27-35 Insiders R's policy: priorities already set
6:36-39 R's transition team at work on LA policy
7:39-48 Kirkpatrick important member of the team
8:48-68 Opinion K: traditional policy, better relations
with LA
9:69-76 Opinion Carter supporters in LA: less pressure on
conservative regimes
10:77-92 Opinion K.: more realistic human rights policy
than Carter's
11:93-104 Foreign ministries: what will happen with Bolivia,
recognition?
12:104-110 Reagan aides: different criteria for recognition
13:111-120 R's policy in Caribbean: against unrest
14:120-129 LA's reaction: crusade against communism will be
dangerous
15:129-132 K's opinion: we are going to help regimes against
communism
16:133-155 R's policy: sell arms, but not to leftist countries
17:156-171 Carter's help is moderating policy in Nicaragua
18:172-177 Plea to continue good relation with Cuba
19:177-182 Reagan will have different policy towards Cuba
20:183-191 LA's worries about help to democratic regimes
21:191-199 Kirkpatrick denies this: we will nurture democ-
racies
22:199-204 Opinion Newsweek: reassurance, but how from
thought to action?
 Episodes as Units of Discourse Analysis / 187

Table 3. Third level episodes in TA new team's Latin test'.

Episode 3 :
lines Segmentation criteria
1:4-7 General summary statement: USA foreign policy
change in LA
2:7-8 General summary statements: reactions in LA to
R's election
3:8-20 LA governments: relief
4:21-27 Human rights activists' opinion: severe setback
5:27-48 R's LA team for LA policy, with Kirkpatrick as im-
portant member
6:48-68 K's opinion: traditional policy, better relations
with LA
7:69-92 Opinions about human rights policies of Carter and
Reagan
8:93-110 Policy towards Bolivia: recognition by R
9:111-132 Policy in Caribbean: against communism
10:133-155 Help by selling arms
11:156-182 No moderating help for leftist regimes like
Nicaragua and Cuba
12:183-199 Democratic countries will be helped
13:199-204 Newsweek's evaluation

But what are the semantic properties which define the epi-
sodes on these respective levels? Let us examine the first
(most detailed) level first (see Table 1). A first criterion for
segmentation appears to be level of description: the first sen-
tences express rather general (macro-) propositions, which sum-
marize the text as a whole. They are so-called thematic sen-
tences, often appearing at the beginning of newspaper articles.
In dailies they are sometimes printed in bold characters (as the
'lead' of the story). So, the general theme is 'Change in US
foreign policy towards Latin America after Reagan's election as
president'. Then, the subsidiary main theme is: 'Various re-
actions to this policy in Latin America'. From there on, the
general structure of the article is as follows: some topic from
Reagan's Latin American policy is mentioned (mostly through
the mouth of his adviser Kirkpatrick), and then the reactions
(opinions, fears, etc.) to this point both by left wing (pro
Carter) officials and right wing (pro Reagan) officials, mostly
conservative governments.
The relatively large number of episodes for this short text
comes from this recurrent switch between a policy statement by
Reagan's aides and reactions from various people in Latin Amer-
ica, or vice versa. In between are found the general state-
ments of the Newsweek journalists, introducing a new theme or
new aspect of a theme (mostly policy points). So, if level of
description is a first distinguishing criterion--because the
188 / Teun A. van Dijk

statements cannot be reduced to the same macrostructure (i.e.

have different participants, etc. from the subsequent sentences),
we must have a change of level as a mark for the next episode.
That is, we change from the general theme 'change of policy' to
the more specific 'consequences of this change'.
Similarly, in the third episode, we now get the various speci-
fications: who is reacting how? So, we first go down to the
collective group of (conservative) governments, then find a
specification of the reaction in Chile, then a statement by
Rockefeller in Argentina. In other words, we first witness a
change of participants: we go either from a general set to a
member, or change between members of different sets (repre-
sentatives of Carter vs. those of Reagan, left wing vs. right
wing Latin American officials). This also means in our text
changes in the local scene of the respective opinions being
given: the various countries are passed in review. The epi-
sode change markers in many of these episodes are simply the
first noun phrases (sentence topics) of the sentences: Many
governments, In Chile, Most human-rights activists, Leaders in
Cuba, etc. These may be subjects, often indicating the se-
mantic agent, or complementizers of place or time. A further
indication are the connectives: differences in opinion may be
introduced with but: But for Reagan Administration insiders-
to-be... (line 27). Another typically journalistic device is not
to first introduce the next speaker or opinion, but to introduce
it with a direct quotation, followed by name or function or
group of the speaker. Finally, we have the usual bold printed
headings, indicating new main themes, as well as the paragraph
identations, as indications of episode change.
We now have, for this kind of text, the following episode (and
hence episode change) criteria:
(a) Level of description (general vs. particular)
(b) Major participant(s): Reagan camp, Carter camp,
Latin American left, Latin American right, Newsweek
(c) Place (in this text hardly time: all present)
(d) Different main themes about Latin American policy
(e) Contrasting, conflicting opinions about these themes
Also we see what kind of coherence is at work here. Besides
the usual type of conditional relations between actions or events
(X says p and therefore Y says q ) , we witness various func-
tional relationships between sentences or between episodes (and
hence between macropropositions): we have many Specification
relations: a sentence or episode specifies or gives an Example
of a more general point. This Specification may be a specifica-
tion of a theme, a specification of a country (location), or a
specification from group to members of the group. Furthermore,
we have already observed that Contrast plays an important role:
several spokesmen give their conflicting opinions about the re-
spective policy items. More specifically, the Contrast may take
 Episodes as Units of Discourse Analysis / 189

the form of Counterargument. Not only do the two respective

camps try to counterargue each other's opinions, but also the
reporters of Newsweek try to fomulate (rather moderate)
counterarguments, or at least doubts about several policy
points.
Now, if we analyze the further reduction of episodes--that i s ,
episodes with a larger scope, as pictured in Tables 2 and 3--
which of the criteria mentioned earlier remain, and which should
be relaxed? First of all, the General-Specific dimension, that
is, the level of description, remains: very general statements
cannot be reduced nontrivially (that i s , if they are not reduced
to themselves, as highest macropropositions) and subsumed
under another macroproposition with their more specific subse-
quent episodes. However, we may abstract from the most par-
ticular participants: instead of talking about a Chilean official,
we may just segment an episode with respect to 'Chile' or
'Chilean government' as participant. The same holds for the
respective pro-Carter groups in Latin America. In this text,
this leaves us with the various groups with different opinions
in the respective countries.
Next, we still have the respective main themes of Reagan's
Latin American policy: better relations with conservative
regimes, more realistic human rights policy, opposition to com-
munist regimes, encouragement of 'young democracies', etc.
Again, the different groups and different or even conflicting
opinions seem to be the main cirteria for the establishment of
episode boundaries at this higher (or more embracing) level of
episode structure. The scene change is from the United States
to some Latin American countries, and the participant change is
from one camp to the other.
As a provisional conclusion, we may assume that this kind of
newspaper or weekly articles about (foreign) policy is episodi-
cally organized according to the following dimensions: (1) main
tenets of the policy; (2) opinions of those who endorse the
policy; (3) opinions of those who are against it; (4) groups of
people who are affected by the policy (positively or negatively),
and their respective contrasting opinions; (5) in relation to
these: the varying locations of the respective groups or people.
This kind of article, typical for weeklies such as Newsweek, is
merely a review of various opinions. It does not give an inde-
pendent critical analysis; it gives little background and few
facts or good arguments, or historically motivated sketches of
prospects. This means that there are few historical explanations
(background) and hence few past tensed episodes; nor are there
prospects or predictions, and hence future tensed episodes
(there i s , however, some hint of these in the last sentence of
the article, and the passage beginning with line 133); the only
future tense passages are those which are about the plans of
the Reagan camp, but Newsweek itself hardly mentions its own
predictions.
190 / Teun A. van Dijk

Similarly, in this example, the functional Contrast relations

between the episodes hold between the respective opinions of
the two camps, but not between these opinions, on the one
hand, and critical opinions of the reporters, on the others.
The weekly just reports in a more or less 'balanced' way, what
the United States government's or the president's policy will
be, and the global consequences of this in the respective
countries, where the consequences are given in terms of opin-
ions of officials. The article does not investigate the more im-
portant possible consequences for the social and political situ-
ation of the various peoples of Latin America, let alone give a
critical evaluation of the new president in terms of these social
values and norms (will the degree of suffering of more people
be higher?).
This very brief and superficial characterization of a typical
American weekly article also can be deduced from its episodical
structures, because the kind of themes, the kind of groups of
participants, the kind of functional relations (contrast between
conflicting parties rather than critical counterarguments against
opinions), and the role of the opinions of the reporters them-
selves , give a different overall picture of the episodes and their
connections from the one that is presented when other partici-
pant groups, other time aspects (past, for instance), other
type of opinions (criticism), etc. are given.
Most important for this analysis, however, is the fact that for
rather 'static' text types also, such as policy reports in week-
lies, an episodic analysis makes sense. Such an analysis speci-
fies how macropropositions are realized in the text itself, how
episodes can be unified during comprehension, how episodes and
hence the (sub-)theme can change, how further organization can
be assigned to the text base (e.g. by unity of place, time,
participant, theme), and what kind of episode, and hence macro-
proposition, changes are explicitly marked in the text (new
paragraphs, headings, sentence topic, e t c . ) .
5. Some implications for a cognitive model. The theoretical
linguistic analysis of episodes which I have given here may have
interesting implications for a cognitive model of discourse proc-
essing. Black and Bower (1979) have already shown that story
statements tend to cluster in episodes and that such chunking
has cognitive relevance: if we add propositions to an episode,
this 'total load' does not affect recall of other episodes. Simi-
larly, if we add unimportant propositions to an episode, this in
general enhances memory for the important episode propositions.
And Haberlandt, Berian, and Sandson (1980) have shown that
episodes are a 'macrounit' of discourse; the encoding load at
the boundaries of an episode is higher than at other nodes of
a story schema. In ongoing experimental research in Amster-
dam (see, for example, den Uyl and van Oostendorp 1980), it
was also discovered that at the beginning of one-episode stories,
the comprehension time for first sentences is significantly higher
 Episodes as Units of Discourse Analysis / 191

than for other sentences of the episode (say, 800 milliseconds

vs. 600 milliseconds). The explanation for this phenomenon
seems obvious: the reader not only must understand the sen-
tence, but also needs to actualize relevant world knowledge,
e.g. frames or scripts, which may stay 'active' in the compre-
hension of the next sentences. Also, as has been shown in
detail in my work with Kintsch (Kintsch and van Dijk 1978;
van Dijk and Kintsch 1977, 1982), the first sentence is strategi-
cally used to derive a macroproposition. This macroproposition
remains in Short Term Memory for the rest of the interpreta-
tion of the same episode. As soon as propositions are inter-
preted that no longer fit that macroproposition, a new macro-
proposition is set up.
The various linguistic markers I mentioned earlier serve as
strategic data for this change of macroproposition, and hence
of episode: as soon as the cast of participants, time, place,
circumstances, and (global) event or action seem to change, a
new macroproposition can or should be formed, and these se-
mantic changes are often expressed in surface structure: para-
graph indentation, pauses, macroconnectives, full noun phrases
(cf. Marslen-Wilson, Levy, and Tyler 1981).
Against this experimental and theoretical background, the
episode has several cognitive functions:
(a) As an additional unit in the organization of textual
sequences of propositions, it assigns further 'chunking'
possibilities, i.e. further organization, to the text, which
in general allows more structured representation in memory
and especially better recall.
(b) Episodes are the textual manifestation of macropropositions;
properly marked, they therefore strategically allow an
easier derivation of macropropositions and hence allow
better and faster understanding of the text as a whole,
as well as better retrieval and recall.
(c) Episodes may be associated with various textual and
cognitive functions, e.g. narrative categories of a story,
or as the bearers of 'interestingness' or 'importance' for
certain text segments, and maybe--for certain discourse
types--of pragmatic functions: the Conclusion of an
argument or the Coda of a story may indicate what
general practical inference should be drawn, or what
should be known, believed, done.
(d) Episodes may be the 'locus' for local coherence strategies:
coherence relations between facts, the (re-)identification
of referents by means of pronouns, the possibility to keep
place or time indications implicit, may take place within
the boundaries of an episode: language users therefore
need to search for the relevant information not in the full
preceding discourse representation in memory, but only in
the representation of the current episode.
192 / Teun A. van Dijk

Of course, further theoretical and experimental work is neces-

sary to specify and test these assumptions. Earlier work, how-
ever, suggests that discourse chunks such as episodes do in-
deed have relevant cognitive properties in terms of short-term
memory interpretation; long-term memory representation, re-
trieval, and recall; hierarchical differences between important
and less important information; the application of macrostruc-
ture formation strategies; the application of local coherence
strategies and their corresponding information searches in
memory; and the further organization of the discourse in terms
of functional categories.
6. Conclusions. Episodes appear to be linguistically and
psychologically relevant units of discourse structure and proc-
essing. They are taken as semantic units, which can be de-
fined as sequences of propositions of a text base which can be
subsumed under a macropreposition. In surface structure they
are expressed by sequences of sentences which usually corre-
spond with paragraphs, and signalled by various phonetic,
morphological, lexical, and syntactic means. Semantically, they
can be identified in terms of (changes of) global predicate, de-
noting a global event or action, a specific cast of participants,
and time and place coordinates. Discourse episodes are taken
to denote world episodes, that is, sequences of events or actions
of some participants in some specific period. In general, both
for world episodes and for discourse episodes, we constrain the
identification of episodes to those propositions (or actions) that
are important, interesting, or 'incidents'--that i s , not stereo-
typical or normal. This means that especially the higher level
macropropositions which have a specific function, e.g. narrative
or pragmatic, cover episodes in a text.
In a cognitive model, episodes appear to function mainly as
further organizers of the text base in short-term processing
and long-term representation, allowing the strategic derivation
and application of a macroproposition, and restricted information
search in local coherence strategies, as well as better recall due
to this more elaborate organization of the discourse.
In addition to earlier linguistic work on episodes, we now have
a somewhat better insight into their semantic and cognitive
status. However, much additional work is necessary. First,
more work is necessary regarding episode markers in surface
structures. Second, we should try to be more explicit about
which macropropositions can be singled out as episode subsum-
ing. Third, the internal organization of episodes needs further
attention, for example, its 'development'. Fourth, the corre-
spondence between episodes and functions or categories of
stereotypical discourse schemata should be further investigated.
Finally, the corresponding cognitive properties need further
empirical research. In general, it remains to be seen whether
the notion of episode is also relevant for other discourse types
and not only for action or event discourse.
 Episodes as Units of Discourse Analysis / 193

APPENDIX

DIPLOMACY

A New Team's
Latin Test
N owhtn will U.S. fortifn policy change
more abruptly—or radically---duriag
the Realm Administration than in Latin
America. And nowhere did the American
election aroue greater passion. Many gov-
emments in the region have breathed a sigh
10 of relief at the prospect of Ronald Reagan
in the White House. In Chile last week
Interior Minister Sergio Fernanda happily
predicted that "the new United States Oov-
eminent will treat its friends as true
IS friends," and on a tour of several Latin
American capitals. Chase Manhattan Bank
chairman David Rockefeller told smiling
audiences that Reagan would be a realistic
President, that he would "deal with the
ttworld as he found it"
Most human-rights activists in Latin CSattiMecktfHUr with Arftntima-i PrtOdtnl Jotfi VU*U.-Antum m •rmlltmT
America viewed the election as a severe
setback for democracy in the region, and contends. "Above all, we should treat them rights. "Carter's poUcy was concerned only
leaden in Cuba and Nicaragua worried that with more respect" with violations of human rights that derive SS
i s Reagan's landslide victory would preclude Among the Carter Administration's sup- from governments and no other sources—
any chance of improvement in bilateral re- porters in Latin America, the greatest fear TO terrorists,forinstance," she argues. "What
lations with Washington. But for Reagan is thai more "respect" will mean leu pres- this has meant in practical terms is that
Administration insiders-to-be, priorities sure on regimes to liberalize. Said Eduardo any government that has forcibly attempted
are already being shaped, and rebuilding SeabraFagundes, president of the Brazilian to suppress terrorism and guerrilla action M
3c links with conservative regimes is high on Lawyers' Association, "Reagan's election has tended to run afoul of the Carter hu-
the list. "The most important issue is to will certainly have negative effects every-7S man-rights policy."
repair relations in the region," vows Jeane where." Kirkpatrick denies that. "We want Oraea Light) Foreign ministries in Latin
J. Kirkpatrick, Reagan's top adviser and a human-rights policy that is realistic and America arc watching attentively to see
designated spokesman on Latin American focuses on reasonably attainable goahs such bow Reagan handles one policy choice: »a
policy. 35 as the protection of personal and legal whether or not to recognize the Bolivian
Traditional Wiadacat The Reagan tran- rights," she says. Calling Carter poUcksao regime of Gen. Luis Garcia Meza. "It
sition team is already at work putting to- "more offensive than effective," Reagan's would be like aflashinggreen light to every
gether a task force to reflne Latin American adviser maintained that future policy will itchy Latin American general who has ever
policy. Among the group's likely members: involve a moreflexibledefinition of human dreamed of mounting a coup," warns one too
Georgetown University's Roger Fontaine,40 Bolivian diplomat in exile. "It's a way of
Pedro San Juan of the American Enterprise Kirkpatrick: A row to npmlr rtlatioru saying, 'If yon overthrow a constitutional
Institute, a Washington think tank, and government, you will not bear any com-
Kirkpatrick herself. Kirkpatrick. a 53' plaints from us'." Reagan aides see it dif-
year-old professor of government at ferently, and think that diplomatic recog-ios
Georgetown and a lifelong Democrat—48 nition is inevitable. "I would not make
b expected to play a key role on the tran- conformity to democratic practices a con-
sition task force and in the new Republican dition of our continued relations with Bo-
government itself. "The Reagan Admin- livia," says Kirkpatrick. "We do not do
istration," she said last week, "will have that with most other nations." no
higher regard for traditional wisdom and 80 The real test of Reagan's Latin American
traditional practices." policy, however, will probably come in Cen-
Reagan's foreign policy will almost cer- tral America and the Caribbean. "The
tainly be "traditional" in the way it treats problems in Central America must be dealt
many of the region's military governments. with immediately," Kirkpalrick says. In 118
"For four years we have been treated a s u addition to the "near-civil war" in El Sal-
an enemy by the United States," says one vador and the growing insurgency in Oua-
official in Buenos Aires. "That will end temala, she sees the danger of unrest and
in January." Kirkpatrick concurs, and ac- violence in Costa Rica, Honduras and Be-
cuses the Carter Administration of causing lize. Some Latin American officials worry 120
the "rapid deterioration" of relations withM that Reagan will be too much of an activist.
all the nations of Latin America. During The presidents of both Mexico and Panama
Reagan's Administration, she says, the em- recently issued warnings against U.S. in-
phasis will be on bilateral relations and tervention in the region. And one analyst
reciprocity. "We treated the Mexican Oov- in Rio de Janeiro warns that "the United 125
eminent outrageously in our negotiations 68 Slates is likely to And itself isolated if ii
on the natural-gas contracts," Kirkpatrick seeks to carry out a crusade against what
NEWSWEEK/NOVEMBER 24. 1910

Copyright 1980 by Newsweek Inc. All rights reserved. Re-

printed by permission.
194 / Teun A. van Dijk

by the former refine ofPre

MaSomotaDebayle. Managua Is going to need
help Iron the outside worid. If Reagan
vetoes aid. they maintain, h would only
serve to alienate the Sandlntsta leadership
further—and probably force Nicaragua to
nomove closer to PkW Castro's Cuba or the
Soviet Union.
Similarly, the incoming Reagan Admin-
istration will have to clarify its stance to-
ward Cuba. Some Latin American experts
ITSare urging the Reagan team to continue
Carters tentative esbrts to improve rehv
tiosa with Havana. On the basis of cwrest
readings, that is unlikely. Reagan can be
to end any modest good-neighbor
put m place under Carter and to
take a tougher fine against Cuba's military
and diplomatic activity in the Third World.
Some Latin Americans worry that the
Reagan Administration will get so caught
I Mup in its new realpolitik that it will not
do enough to reinforce democracy in those
Vlotnt dnlh in El Salmion Vndtr *mf*. « "JUxW itfiniHon of hums* rigku countries of Latin America where it now
exists. "People like President Jsime RoMos
it sees as the spread of Communism in Cen- to be much less enthusiastic—and perhaps of Ecuador are going to be looking over
tral America." But, retorts Kirkpatrick, downright opposed—to aiding nations that NOtbeir shoulder from now on," predicts one
130 "all the countries seem to be quite vul- it considers "unfriendly." Reagan will have Latin diplomat. But Kirkpatrick adamant-
nerable [to Communism] and we are going to decide scon after taking office whether ly denies that the incoming Administration
to have to help them." MOto help Nicaragua—and if so, on what con- will be lax m supporting democratic re-
Fourth Plac« Just how the Reagan Ad- ditions. "We must have guarantees about gimes like the recently restored ones in Peru
ministration intends to help is unclear. Al- where the aid will go," says Kirkpatrick. Wand Ecuador. "You will see very great ef-
OS most certainly, it will lift Carter's 1977 ban "It should not be used to assist in the con- fora by us to nurture democracies in the
on arms ssles. "Our whole military-sales sobdation of power in a one-party state region," she insists. "That will involve both
and training-assistance
g policy
py hi overdue mthat is hostile to the United States." moral and economic support and ins very
for review," ssys " Kirkpatrick. "The~ - - - - - - -
i That attitude is big way." Such talk is bound to i
amount off arms acquired
id in i Latin
Li AmericaAi certain to prove controversial. Some an-aooWaahington's friends in the hemisphere •
I4ois st a higher level than ever while the alysts say the Carter Administration's offer but it remains to be seen just how quickly
United Slates has simply fallen to fourth of $75 million in aid is exerting a moder- and effectively the incoming Reagan Ad-
place as a supplier behind France, WestMOadng influence on Nicaragua's Sandinista ministration can translate such thoughts
Oennany and the Soviet Union. That's not leaders. They argue that given the Nica- into action.
progress toward demilitarization." raguan Government's massive economic DOUOLAS RAMSEY m WKMUWIIM
148 The Reagan Administration is expected problems and the destruction left behind lA*M.y%OKTt*.mtbi>£i—mn

REFERENCES
Black, John B . , and Gordon H. Bower. 1979. Episodes as
chunks in narrative memory. Journal of Verbal Learning and
Verbal Behavior 18.309-318.
Chafe, Wallace. 1980a. The deployment of consciousness in
the production of a narrative. In: Chafe (1980b:9-50).
Chafe, Wallace, ed. 1980b. The pear stories. Norwood, N.J.:
Ablex.
van Dijk, Teun A. 1972. Some aspects of text grammars.
The Hague: Mouton.
van Dijk, Teun A. 1977. Text and context. London: Long-
man.
van Dijk, Teun A 1980. Macrostructures. Hillsdale, N . J . :
Erlbaum.
van Dijk, Teun A . , and Walter Kintsch. 1977. Cognitive
psychology and discourse. In: Current trends in text lin-
guistics. Edited by W. U. Dressier. Berlin/New York:
de Gruyter. 61-80.
van Dijk, Teun A . , and Walter Kintsch. 1982. Strategies of
discourse comprehension (in preparation).
Dik, Simon C. 1978. Functional grammar. Amsterdam: North
Holland.
 Episodes as Units of Discourse Analysis / 195

Givon, Talmy. 1979a. From discourse to syntax: Grammar as

a processing strategy. In: Givon (1979b:81-112).
Givon, Talmy, ed. 1979b. Syntax and semantics 12. Dis-
course and syntax. New York: Academic Press.
Haberlandt, Karl, Claire Berian, and Jennifer Sandson. 1980.
The episode schema in story processing. Journal of Verbal
Learning and Verbal Behavior 19.635-650.
Hinds, John. 1979. Organizational patterns in discourse. In:
Givon (1979b: 135-158).
Kintsch, Walter, and Teun A. van Dijk. 1978. Toward a model
of discourse comprehension and production. Psychological
Review 85.363-394.
Longacre, Robert E. 1979. The paragraph as a grammatical
unit. In: Givon (1979b: 115-134).
Marslen-Wilson, William, Elena Levy, and Lorraine Tyler. 1981.
Producing interpretable discourse: The establishment and
maintenance of reference. To appear in: Language, place
and action. Edited by R. J. Jarvella and W. Klein.
Chichester: Wiley.
den Uyl, Martyn, and Herre van Oostendorp. 1980. The use
of scripts in text comprehension. Poetics 9.275-294.
 DISCOURSE THEORY AND
THE INDEPENDENCE OF SENTENCE GRAMMAR
J. L. Morgan
University of Illinois

0. Introduction. During this time of remarkable growth in

the study of discourse, an important question has been posed
for the generative grammarian and the discourse theorist alike:
what is the relation between, on the one hand, the mental sys-
tems that underlie human ability to understand and produce
connected discourse, and, on the other hand, theories of
grammar of Chomskyan coloration that take just sentences as
their domain—what I call 'sentence grammar'.
There are a number of positions one might take on the rela-
tion between discourse competence and sentence grammar, on
either prioristic or empirical grounds. A couple of these posi-
tions are potentially threatening to theories of sentence gram-
mar, and I would like to examine some arguments that might
be advanced for them, to see whether the arguments are
persuasive as criticisms of sentence grammar.
The literature of discourse studies by linguists, from at
least van Dijk (1972) to de Beaugrande (1980) and Givon
(1980), contains a number of proposals with a common thrust:
that close study of problems of discourse analysis and theory
shows that sentence grammar is deficient in important respects,
hence incorrect; therefore it ought to be abandoned in favor
of a general theory of discourse, in whose terms sentence
grammar is reformulated or entirely subsumed. But in fact,
such proposals contain two major claims that are logically inde-
pendent, since one can be true and the other false. The
weaker of the two is the claim that certain facts about dis-
course show that generative sentence grammar cannot be the
correct account for sentence syntax and semantics. The
second, stronger position is that the right discourse theory
(or complex of theories) will do the work of a generative sen-
tence grammar, making the latter superfluous as a distinguish-
able component of overall linguistic ability, thus debunking

196
Discourse Theory and Independence of Sentence Grammar / 197

such important hypotheses as the autonomy of syntax, the

independence of sentence grammar, perhaps even the existence
of a logically independent 'language faculty'. These are not
trivial issues, as one can see from the large and combative
literature in several fields that has grown around Chomsky's
work. So it is worthwhile to consider whether the arguments
against sentence grammar that are so common in the literature
of discourse theory actually are a threat to sentence grammar
theory.
In this paper I want to examine both the weak and strong
positions to determine the strength of the arguments for each.
I try to show that arguments for the weak position have no
force, since they overlook a central claim of sentence grammar
theory. I argue that the strong position, though unconvinc-
ing in present forms, is more promising, and worth pursuing.
I also discuss what it would take to make the strong position
a more persuasive one.
1. The weak position. Arguments for the weak position--
that is, against sentence grammar--are generally taken as argu-
ments for the complete abandonment of sentence grammar in
favor of some other theory. But it is not always clear which
crucial property of sentence grammar it is that the argument
is directed against. So it might be useful to list some central
properties of sentence grammar. I believe that the following
five properties--that i s , claims about the nature of grammar--
are characteristic of more or less standard versions of genera-
tive grammar.
(1) The independence of sentence grammar: Linguistic
competence contains as a distinguishable component a
cognitive system whose domain is the sentence (and,
implicitly, smaller expressions that make up sentences).
The fact that this system interacts with other cognitive
systems in performance is not an argument against the
claim that it is an independent system. The boundaries
of this system--that is, what properties of sentences
fall in its domain--is an open empirical question, not
determinable by a priori means.
(2) Language uniqueness: At least some aspects of this
cognitive system--or, from the linguist's viewpoint,
the conceptual vocabulary for describing it--are unique
to language, not found in other cognitive systems.
(3) Grammar is formal: Sentence grammar is organized
and, from the linguist's viewpoint, can be described,
purely as a matter of form, entirely independent of
questions of communication, speaker's intention, and
other matters of language use.
(4) Structure is Chomskyan: The proper treatment of the
notion 'sentence structure' is in the terms of the familiar
phrase markers in the work of Chomsky and many others
198 / J . L. Morgan

before and since; that is, syntactic structure (at a

given level) is reducible to relations of dominance,
precedence, and syntactic category, and other notions
defined in these terms.
(5) More than one level of structure: An empirically ade-
quate grammar requires that each sentence be assigned
at least two levels of structure. The exact number of
levels is an open question.
Obviously, these five properties do not exhaust the con-
ceptual content of generative grammar; Chomsky's more recent
work on extended standard theory, to pick a single example,
is far richer. And, in fact, there is not unanimity on all five
properties even within the generative camp. Properties (4)
and (5) are presently controversial within generative grammar,
under attack by proponents of 'arc-pair grammar' (Johnson and
Postal 1980), and in recent work on 'constituent structure
grammar' by Gazdar (1980) and his collaborators. Abandoning
or replacing these properties would indeed constitute an im-
portant theoretical shift; but the result would still count, as
far as I can see, as a sentence grammar of the sort under
attack from discourse theorists. Unless I am mistaken, it is
properties (1) through (3) that are relevant to the criticism
of generative sentence grammar vis-a-vis discourse grammar.
The question is, then: do the arguments of the weak posi-
tion succeed as arguments against the first three properties?
There are two lines of argument for the weak position.
The first, apparently intended as an argument against the
independence of sentence grammar, is that the only 'naturally
occurring unit' for linguistic analysis is the discourse, not
the sentence. This argument is far from convincing. If it is
taken as a rather bizarre a priori methodological restriction,
then it has no bearing on the question of independence.
Even accepting this restriction, there is no reason to believe
that it precludes the possibility of discovering, from close
analysis of discourse, motivation for an independent sentence
component. Taken more seriously, as leading in some obscure
way to an a priori argument against the independence of sen-
tence grammar, again it has no force. The question of inde-
pendence is an empirical one--either there is an independent
sentence grammar or there is not--and cannot be decided a
priori.
The second line of argument is to criticize sentence grammar
as inherently incapable of providing an analysis of phenomena
of various sorts. Such observations can be construed either
as arguments that sentence grammar is incomplete, or as argu-
ments against one or more of the first three properties.
Van Dijk (1972:7) uses them to show, if I understand correctly,
that sentence grammar is incomplete, that there are discourse
phenomena that cannot be accounted for by a sentence gram-
mar:
Discourse Theory and Independence of Sentence Grammar / 199

As long as S-grammars cannot provide satisfactory,

general and consistent, descriptions of the structures
underlying discourses, by formulating the rules which
must be mastered by native speakers to be able to per-
form the different tasks, we have to consider them
empirically inadequate.
But one cannot criticize an owl for not being a partridge.
This kind of criticism ignores a central thesis of sentence
grammar theory. Chomskyan generative grammar and its
descendants are based on a hypothesis of great theoretical
and empirical importance: the existence of an independent
language faculty, within which there is a sentence grammar
with rather narrow domain. Insofar as sentence grammar in
the Chomskyan spirit can offer only an incomplete account of
some language-related phenomenon, it amounts to a claim about
the nature of the phenomenon--that it is outside the domain of
sentence grammar, perhaps outside the domain of the language
faculty altogether, the product of some other cognitive system
or systems. The incompleteness is not an incompleteness by
default or omission, but a claim, correct or incorrect, about
the facts.
Thus arguments of this type against sentence grammar, on
the grounds that it cannot account for certain discourse phe-
nomena, have no force, unless accompanied by a demonstration
that the phenomena in question must be considered to be in the.
same domain as phenomena that are central to sentence grammar
theory. But such a demonstration is a very difficult task—
phenomena are not pre-sorted by nature in this way. There
is no basis from which to argue one way or the other, save to
provide support for what I have called the 'strong position':
to present an alternative theory which treats sentence phe-
nomena and discourse phenomena in a single unified theory,
without a distinguishable sentential subcomponent. One could
then attempt to compare such a theory with a sentence grammar
as an explanation of sentential phenomena. Lacking such an
alternative, there is no ground for comparison. Along these
lines, observations taken to be arguments for the incomplete-
ness of sentence grammar could be construed as arguments
against properties (2) and (3) (hence potentially as arguments
against independence), if it could be shown that the phenomena
in question require a uniform treatment of discourse and sen-
tential properties, and that the sentential properties concerned
are central to sentence grammar.
One such argument is the claim that there are important
parallels between properties of discourse and properties of
sentences. In fact, there are at least three parallels one
might see between sentences and discourses. First, they
both have structure. The point has been made again and
again, and quite correctly, that one key to the understanding
of discourse is the idea that discourses have structure. Then
one might propose 'grammars' of some kind for discourses, to
200 / J . L. Morgan

generate texts and assign them structure. From here it is

only a short step to the hypothesis that one can provide a
single grammar that treats both sentence and discourse struc-
ture.
But the parallelism turns out to be rather tenuous on closer
inspection. The kind of structure commonly attributed to sen-
tences (and not just by Chomskyans) is not the same kind of
structure commonly attributed to discourses (see Morgan and
Sellner 1980 for more discussion). What is needed is a demon-
stration that the system for determining discourse structure
can be extended to give a complete treatment of the syntax of
sentences, a demonstration so far lacking, though not incon-
ceivable.
A second apparent parallel is that texts, like sentences, have
'meaning' in some sense of this perniciously vague term. For
example, de Beaugrande (1980:37) points out that meaning re-
lations that can hold within a sentence can hold between inde-
pendent sentences in a discourse. He offers the following
pair from Isenberg (1971:155) as illustration of this point:

Peter burned the book because he didn't like it.

Peter burned the book. He didn't like it.

But again the observation of the parallelism is misleading, in

that it obscures an important difference between the two cases
in the illustration. In the first, sentential case, we under-
stand the relation to hold because of the parts of the sentence
and their mode of combination. But in the second, discourse
case, as in all cases of understood relations between inde-
pendent sentences in a discourse, we must infer that the rela-
tion is to be understood to hold. Two different mechanisms
are involved: in the sentential case, our knowledge of gram-
mar--of the conventional meaning of the word because, and of
just how the meanings of larger English expressions are related
to the meanings of their parts; in the discourse case, our
ability to make common sense inferences. The latter can be
cancelled by contextual factors, in the manner of Grice's (1975)
conversational implicature; the former cannot. It would be a
mistake to ignore the difference.
The third parallel--anaphoric relations like some antecedent -
anaphor relations, the interpretation of definite noun phrases,
and so on--is similar to the previous one, in that such rela-
tions can hold either between elements within a sentence or
between elements in two separate sentences in a discourse.
In this case, though, the evidence suggests the necessity for
a unified treatment. It is fairly clear that such matters need
to be treated in extrasentential, perhaps discourse terms.
But a perfectly coherent response to such arguments is avail-
able to the proponent of the independence of sentence gram-
mar: namely, just to yield the territory—to conclude (cor-
rectly, to my mind) that such cases are outside the domain of
Discourse Theory and Independence of Sentence Grammar / 201

sentence grammar. Nothing in sentence grammar theory entails

that everything that can conceivably be labelled a property of
sentences must be accounted for in sentence grammar.
Another kind of criticism of sentence grammar is based on
the observation that there are discourse explanations for
apparently syntactic facts. There are a number of interesting
attempts in the literature (see Givon 1980 for some recent ex-
amples, especially the papers by Garcia and Erteschik-Shir).
But generally, the sentence grammarian can respond to such
analyses in the same way as to the previous case: by yielding
the territory, concluding that the existence of convincing dis-
course explanations shows that the problem was not a syntactic
one to begin with.
A related kind of argument is based on the observation that
there are expressions whose meaning and /or syntactic distri-
bution is clearly to be given in terms of discourse or func-
tional terms. One might conclude from the existence of mor-
phemes that function as topic or focus markers, for example,
that a complete theory of sentence grammar must incorporate
a treatment of notions like 'focus', 'topic', and the like. But
there is no reason to accept this conclusion. It is no more
necessary than this one: since English has pronouns like he
and she whose meaning properties (hence use) are determined
in part by natural gender, the theory of sentence grammar
must contain a theory of physical gender. More plausibly,
these are instances of the interaction of sentence grammar with
other linguistic or nonlinguistic cognitive systems. The exist-
ence of such interactions in no way provides arguments for the
identity of the interacting systems.
A similar approach is available for dealing with 'optional'
rules or constructions with clear discourse value--constructions
like 'Y-movement1, for example, that have different discourse
appropriateness conditions from their unmoved counterparts.
Sentence grammar need only specify which orders are possible,
i.e. grammatical. The rest should follow from language-
specific discourse rules, from general principles of communi-
cation, or from the interaction of grammar with other cognitive
systems.
In short, such empirical arguments against the independence
of sentence grammar are not convincing, taken one by one.
Conceivably, they could become persuasive cumulatively, by
gradually reducing the domain of sentence grammar to empti-
ness; but that day is hardly on the horizon. In the meantime,
the sentence grammarian can fairly comfortably continue to
take such observations as progress in the empirical determi-
nation of the domain of sentence grammar.
3. The strong position. The strong position is that the
right discourse theory will entirely subsume sentence grammar
or provide a radical reformulation of it in discourse terms.
Obviously, one persuasive way to argue for the abandonment
202 / J . L. Morgan

of a theory is to present an alternative that gives a superior

treatment of a significant portion of the domain of the theory
under attack; for example, to show how a fairly well articulated
theory of discourse provides an account of some phenomenon
that is central to sentence grammar. What would be required
of the theory of discourse, then, is that it be complete enough
to allow examination of its consequences at the sentence level
for naturally occurring or constructed discourse, with a de-
gree of formal detail that approaches that of existing sentence
grammars. Unfortunately, there is no theory of discourse
that is that well developed. This state of affairs is hardly
surprising, given the almost miraculous complexity of the men-
tal systems that underlie our ability to produce and understand
discourse. At this point such a comparison is impossible.
Still, it might be useful to sketch in hypothetical terms some
directions that such an enterprise might take.
To begin, I need to narrow down a bit what I mean by 'dis-
course theory'. I mean any theory that attempts to satisfy
two minimal conditions:
(1) It attempts at least a partial account of the most
striking aspect of discourse comprehension: how an
understanding of a discourse is so much greater than
the logical sum of the parts (i.e. sentences) that make
it up.
(2) It offers a definition or explication of indispensable
but elusive notions like 'topic', 'focus', 'given/new',
'relevance', 'coherence', and 'text structure'.
Such a theory might fruitfully be framed in terms of communi-
cative actions, i.e. rules or strategies for the activity of com-
munication, rather than rules of well-formedness. Matters of
ill-formedness would be recast either as actions that violate
rules of communication, or as inefficient or self-defeating
communicative actions, given principles of communicative
efficiency. Such a theory would also need to include (or ap-
peal to interaction with) a theory of common sense reasoning,
and would likely include a component of language-specific
conventions of discourse and of other aspects of language use
(see Morgan 1978a,b for discussion).
If such a theory were available, then, it would be possible
to attempt to recast central aspects of sentence grammar in
terms of the independently motivated discourse theory. Again,
lacking a detailed theory of this type, discussion is necessarily
speculative. But a couple of illustrations help make clear what
kind of attempt I have in mind. The strategy would be to
determine how much of the semantics and syntax of sentence
grammar could be treated by discourse theory, leaving only
matters of morphology and the lexicon to sentence grammar.
The possibilities for semantics are rather dubious, it seems
to me. A theory whose goal is to explain discourse
Discourse Theory and Independence of Sentence Grammar / 203

comprehension must give a central role to inference, though

perhaps with a language-specific, conventional component as
well (by language-sped fie here I mean principles that differ
from language community to language community, and must be
learned). Then it is not out of the question that this system
could provide a parallel treatment for problems of composition-
ality within sentences, yielding then a single uniform system
for all aspects of meaning analysis, both at discourse and
sentence-internal levels. But such an attempt faces large
obstacles. It would be necessary to show how the composition-
ality that is a central tenet of sentence grammar could be dis-
pensed with. The claim of compositionality is that any ade-
quate theory of semantics must analyze meaning as depending
not only on the elements that make up expressions, but on
their syntax--on the way they are combined to make up the
expression. It is hard to see how to extend an inference-based
understanding system to deal naturally with the difference be-
tween the dog bit the cat and the cat bit the dog without in
the process reinventing sentence syntax, let alone how to con-
quer the well-known problems of the relation of syntactic
properties to scope of logical operators. The likelihood of suc-
cess of such an attempt is very implausible, I think, though
it cannot be ruled out a priori.
The attempt to recast syntax in discourse terms is perhaps
slightly less implausible. Take the syntax of noun phrases,
for example. Given a theory well developed enough to include
treatment of the action of referring, and given that language
communities can differ in their conventional rules of discourse,
one could attempt to recast the syntactic rules for English noun
phrases as English strategies for the act of referring.
The possibility of such reformulation of syntax in terms of
communicative function is tantalizing, since there are obvious
correlations between syntactic form and communicative function.
For example, in language after language, the unmarked posi-
tion of the restrictive relative clause is adjacent to the 'head
noun' that it 'modifies', as in The woman who invented the
wheel died in 70,000 B.C. Viewed purely formally, this seems
just an unexplainable (though widespread) quirk. Viewed
functionally, on the other hand, it is hardly surprising, since
the head and accompanying relative are uttered in pursuance
of a single purpose--to pick out a referent by describing its
properties. Uttering the head and the relative constitutes a
single communicative act, and the temporal adjacency is un-
surprising, assuming some intuitively obvious principles of
efficiency. From this viewpoint, it is the cases where the
relative is detached from the head that are surprising.
But this kind of analysis, though tantalizing, also faces
serious obstacles, insofar as form does not always follow func-
tion. For example, how could such a theory explain cases of
apparent functional disunity like extraposed relatives, as in
The woman died in 70,000 B.C. who invented the wheel or
201 / J . L. Morgan

verb-particle constructions like John put the cat out, to say

nothing of the numerous apparently purely formal conditions
and constraints proposed by generative grammarians from Ross
(1967) to Chomsky (1981)? The burden is clearly on the dis-
course theorist to show that at least a significant fraction of
these problems have explanations in discourse and/or functional
terms. Personally, I am skeptical that such explanations will
ever be achieved. But I think the knowledge to be gained in
the attempt is worth the effort.
REFERENCES

de Beaugrande, Robert. 1980. Text, discourse, and process.

Norwood, N . J . : Ablex.
Chomsky, Noam. 1981. Lectures on government and binding.
Dordrecht, Holland: Foris Publications.
van Dijk, Teun. 1972. Some aspects of text grammars. The
Hague: Mouton.
Erteschik-Shir, Nomi. 1980. Discourse constraints on dative
movement. In: Givon (1980).
Garcia, Erica. 1980. Discourse without syntax. In: Givon
(1980).
Gazdar, Gerald, (to appear) Phrase structure grammar.
In: The nature of syntactic representation. Edited by
G. K. Pullum and Pauline Jacobson. Boston: Reidel.
Givon, Talmy. 1980. Syntax and semantics, Vol. 12: Dis-
course and syntax. New York: Academic Press.
Grice, H. P. 1975. Logic and conversation. In: Syntax
and semantics, Vol. 3: Speech acts. Edited by Peter Cole
and Jerry Morgan. New York: Academic Press.
Isenberg, Horst. 1971. Ueberlegung zur Texttheorie. In:
Literaturwissenschaft und Linguistik: Ergebnisse und
Perspektiven. Edited by Jens Ihwe. Frankfurt: Athenaeum.
Johnson, David, and Paul Postal. 1980. Arc pair grammar.
Princeton: Princeton University Press.
Morgan, Jerry. 1978a. Two types of convention in indirect
speech acts. In: Syntax and semantics, Vol. 9: Prag-
matics. Edited by Peter Cole. New York: Academic Press.
Morgan, Jerry. 1978b. Toward a rational model of discourse
comprehension. In: Theoretical issues in natural language
processing. Edited by D. Waltz. Urbana, 111.: University
of Illinois.
Morgan, Jerry, and Manfred Sellner. Text structure from a
linguistic point of view. In: Theoretical issues in reading
comprehension. Edited by Rand J. Spiro, Bertram C.
Bruce, and William F. Brewer. Hillsdale, N . J . : Erlbaum.
Ross, John R. 1967. Constraints on variables in syntax.
MIT Ph.D. dissertation. Bloomington: Indiana University
Linguistics Club. Mimeo.
 STRATEGIES FOR UNDERSTANDING FORMS
AND OTHER PUBLIC DOCUMENTS

V. Melissa Holland and Janice C. Redish

American Institutes for Research

Since 1978, we and our colleagues in the Document Design

Center at the American Institutes for Research have been
studying the problems that people have in understanding and
using public documents. Our interest ranges across a wide
variety of documents—for example, legal contracts, insurance
policies, warranties, regulations, instructions for forms, and
the forms themselves. In this paper, we are concerned with
forms as discourse: how are forms similar to other types of
text? how do they differ? how can we best study forms as a
type of text? what have we learned in our work so far?
From other studies that we have done, we have both objec-
tive and subjective evidence that forms are critical in people's
lives and that they have difficulty dealing with them. Yet
very little research has been done on them. Our interest in
conducting this research is threefold. We hope to (1) help
writers and designers improve forms so that they better fit
the skills and strategies of the target audience; (2) help edu-
cators to improve the skills and strategies of students who will
have to use these forms; and (3) add to linguists' and psy-
chologists' understanding of how people process written material.
A model for reading forms and other functional documents.
Documents have pragmatic contexts and immediate consequences
that the more traditionally studied texts may not have. Read-
ing a document is a functional necessity for the reader; docu-
ments usually require immediate action rather than long-term
memory storage. We, therefore, use the term 'functional read-
ing' to contrast document-reading with reading for pleasure
(stories) and reading to learn (expository prose).
The model pictured in Figure 1 is an illustration of some of
the factors we see as critical to understanding 'functional

205
 206
Figure 1 . Critical aspects of functional reading.

User characteristics User behavior Document characteristics

Background characteristics Execute tasks with the Text (message)
document Organization
• intelligence (0

lissa Ho
* yaneral knowledge Syntax
* cognitive style Lexicon
• reading Iqyel Physical characteristics
© perceptual & motor

t
0)

nd and Janice C. Redis

skills Formulate questions and
a social & demographic strategies for making
characteristics sense of the document; User's task
• experience with form internal representation
of the document 9 read
documents
• fill in
• sign
• transfer information
\ • etc.
Expectations about
i
documents
r 3"

Purpose of document
within system
I Specific needs and
1
motivations
r

Social and physical

context of using
document
Strategies for Understanding Forms and Public Documents / 207

reading'. x As the model shows, successful use of a document

is an interaction between the document's characteristics and the
user's characteristics as they influence the user's behavior.
The user employs strategies and processes to make sense of
the document and to fulfill the required tasks. We propose
that by viewing forms as discourse, we can stimulate a pro-
ductive line of research--that i s , we find that forms exhibit
identifiable text characteristics and elicit strategies for compre-
hension that call on these characteristics, on context, and on
the user's prior knowledge about the world--as do other types
of text. We further propose that simply carrying over what
we know about other types of text is not sufficient for under-
standing forms—that i s , we find that forms have unique charac-
teristics and that they require processes and strategies which
differ in systematic ways from those needed to understand
stories or textbooks.
We want, however, to bring previous research to bear on our
studies of forms as discourse.
What has research on other text types shown? Linguists
analyzing discourse have found a range of variables that serve
as distinct levels in the description of stories, expository
prose, and the more traditional forms of text. Psychologists
studying text comprehension and memory have found that these
variables are robust predictors of readers' performance. These
variables include:
(1) Cohesion: the ties between sentences identifiable in
surface structure such as ellipsis, anaphora, word
repetition (Kintsch 1974, Grimes 1975, Halliday and
Hasan 1976).
(2) Message structure (staging): the patterning of infor-
mation in sentence sequences by theme-rheme or given-
new relationships (Halliday 1967, Chafe 1970, Grimes
1975, Clark and Haviland 1977).
(3) Rhetorical predicates or coherence relations: the deeper
logical and semantic connections between sentences
(Grimes 1975, Meyer 1975, Hobbs 1979).
(4) Text structure: the overall organizational form, depend-
ing on the text type or genre (van Dijk 1972, 1980;
Kintsch and van Dijk 1978). Grammars have been formu-
lated to characterize the structures of texts that fit
highly steroetyped cultural norms, e.g. simple stories
(Rumelhart 1975, Mandler and Johnson 1977, Stein and
Glenn 1979). For example, Adams and Collins (1979)
suggest that to interpret a simple tale as a fable, the
reader must have a schema for fable specifying that
some statement will be interpreted as a moral.
(5) Content or message base: the level that draws on world
knowledge, such as is represented in scripts (Shank and
Abelson 1977) and frames (Minsky 1975). Scripts
208 / V. Melissa Holland and Janice C. Redish

represent people's knowledge about frequently en-

countered episodes with predictable participants and
scenes (for example, a restaurant script). Knowledge
frames define elements and relationships in more static
arrays. At the very least, the presented text, in order
to be understood, must engage some part of the reader's
prior knowledge about the world (Bransford and Johnson
1973, Bransford and McCarrell 1974). This knowledge,
as revealed by work in artificial intelligence, semantic
memory, and discourse comprehension, appears to be
organized into richly structured, global configurations--
not only scripts and frames, but looser, more abstract,
and more flexible structures known as 'schemas1 (Ander-
son 1977) (a distinction drawn by Winograd 1977 and
developed by Rumelhart and Ortony 1977).
Tannen's work (1979) on frames in the discourse of
oral reporting fits here also; it conforms to what might
be called dynamic schemas. Tannen's work shows that
one can operate on several complex and subtly distinct
levels at once--from adopting a general rhetorical frame
(subject-of-the experiment vs. telling-a-story) to frames
for interpreting content.
(6) The cooperative contract: a pragmatic convention de-
scribed by Grice (1975) in which speaker and hearer (or
writer and reader) enter into a tacit agreement to cooper-
ate, and for each to assume that the other is cooperating
toward the purpose of communicating. Cooperating, ac-
cording to Grice, means being relevant, clear, truthful,
and informative. This contract underlies all discourse.
As we show later, one of the problems that unsuccessful
form users have is that they do not assume the forms
designer had a communicative intent.
This very productive work needs to be expanded in two ways:
(1) to see how these variables affect performance with other
text types beyond expository prose and stories; (2) to delve
deeply and more directly into the box in our model labeled
'strategies' or 'processes'. We are exploring both of these
directions in studying application forms.
Forms as discourse—textual analysis. It might be objected
that forms are not text. After all, they typically consist of
numbered, fill-in-the-blank items rather than complete sentences
in a linear flow. The connections between items appear to be
largely procedural rather than logico-semantic or rhetorical.
And the cognitive result of using a form is not the construction
of an internal text base, or macrostructure, as Kintsch and
van Dijk (1978) have posited as the product of reading prose.
However, work indicates that forms do display properties of
connected prose, although the connections are looser and
sparser than in more traditional texto (On the continuum
Strategies for Understanding Forms and Public Documents / 209

suggested by Halliday and Hasan 1976, forms would fall some-

where between an index and a narrative.) We have both
analytical and empirical indications that forms are text. That
is, the indications can be seen in the document itself and in
the behavior of document users.
We have discussed analytical evidence that forms are text in
other papers (Charrow, Holland, Peck, and Shelton 1980; Hol-
land 1980, 1981; Campbell and Holland forthcoming). There-
fore, we only briefly review some of this evidence here. Con-
sider the Medicaid form shown as Figure 2.
like most forms, this one displays 'cohesion' by using word
repetition and anaphora. As an example of repetition, note
that terms like family and family member have special meanings
defined in the form. These terms can be seen in the heading
of the box on the front of the form. They are repeated
throughout the document with the same special meaning. If
form users do not keep the same special meaning in mind
throughout the form, if they do not grasp the intended co-
hesion, then they may make errors in filling out the form.
As an example of anaphora, note the pronouns he(she).
your, and their connecting the separate items under Questions
3 and 4, referring to support from spouses. Anaphora is more
common in new 'plain English' forms we have designed, in
which the pronouns we and you are defined as specific parties
in the beginning of the form and then occur throughout the
document. Obviously, users must look for and recover the
special meanings of repeated terms and the anaphoric connec-
tions if they are to understand and respond accurately on the
form.
This Medicaid form also displays a 'given-new' patterning of
information. This pattern underlies local hierarchies of items
in several places on the form. For example, the items in Ques-
tion 5 'Income from work' are all hierarchically related: 'Amount
Taken Out' and 'Expenses' refer to 'Income from Work'. 'Ex-
penses', in turn, subsumes 'Transportation' and 'Other'. The
appropriate response to 'Transportation' and 'Other' is only the
expenses in these categories that are related to work--not just
any 'transportation' or 'other' expenses in one's life. Many of
the Medicaid recipients who were subjects in our study of this
form did not observe these topical links. They answered all
items as if they were independent and unconnected. Under-
standing the hierarchy in Question 5 requires operating under
the conventions for staging information in texts. This means
seeing each label or phrase as the 'new' information at the
level where it is introduced, and then as the implicit 'old' in-
formation for the next lower level. The hierarchy is recover-
able by applying a given-new strategy.
Analyzing forms is one of the techniques that we and our
colleagues are using to understand and improve public docu-
ments. We are also conducting empirical studies in which we
210 / V. Melissa Holland and Janice C. Redish

Figure 2. Medicaid form.

RECERTIFICATION FOR MEDICAL ASSISTANCE

<i CENSUS TRACT:

SERVICE AREA:

0. C. I. 0. NO:

UPON RETURN SEND TO:

I b t M M U lunUM proof «l

II this lorm Is not returned completed within 15 days your eligibility (or Medical Assistance will terminate.
PLEASE PRINT CLEARLY. ANSWER ALL QUESTIONS ON BOTH SIDES OP THIS FORM. WHEN THE FORM IS COMPLETED AND SIGN

THIS COLUMN IF RECV1NO WELFARE

LIST NAME OF Afl

T3FF-- CITY:
IF YOU HAVE MOVED: PT * Q _ STATE- 2IP CODE:

3. Does the SPOUE;e ot any person listed abov e live outside t h e - a ™ . , . D YESQ NO

V/nle in Ihe amounl lhat he Ishe) give SlCWi3rd yoiur support each mon,ms

n NO
'I yes. f)ive narr,e ana address:
NOW FILL IN BACK
Write in ihe amount that he ishe) gives tc-.vard their support per month: S
OF THIS FORM
 Strategies for Understanding Forms and Public Documents / 211

Figure 2. Continued.

9.

meom. Betoro Deductions: Weakly: S Bi-Wkty: $

UPfiNBIg DO NOT WRITE

s s Transoo'tation (Daily)
s 0. C. Tai I Child Care (Wesxtv) IN SHADED AREA
t Secur'iy om
FfH
oort oavmenti etc
B n workino
Bi-Wkiy: J
Employers Na nn and Addrmi:

s '" ""fVw""'" 1 "' i Tra ni orialion (Oau

n r. T » . * Chi

s s Oth er lEioia.nl

J Bonds A Crodil Union Support Payments, etc.

6. o HER INCOME: Do Any family members hving wiih you and navmg a meaicai ance
owing:

SOL RCE
i WORKER'S SSR CODE:
.-,, Service Remmem $ INSURANCF COOF

_Milila.y Rei.rerrnnl S (Ois'egarded Income) 1 :OAA 1 1 AS

t ! lAFOOl !OTH

*
s TOTAL RESOURCES

7
IN ings & Loan lesa Htdlh Ins. (« months)

M oney m Savings s. INCOME SCALE

Real estate; (Other than Home "• wmch we live) 1

Signature ol Cefhlying Qtln

Caih Vatue

And pay $ . .(Circle one) weekly B.-Weekly Monthly Quarterly Insurance Company

C. Medicaid Inlorr h you receiving Medxaro?

_ Person! s) covered

sucn proof. 1 w<;irepo ,,.o,on

•rzrriovrv n,»,,5.n,,da or address change to i:

3.
._

SIGNATURE OF ESS
ADDRESS
TO -X" MARK A!IOVE CF
MONTH VEAR YEAR WITNFS4-

A
5
OO.S«:t,on 3.3.2161.

S,na.Ur,o,Aoo wan IS Quaro.an. or OO'lon »no Hewed III! 0U1 MO Telephone Numoe.

I lh« Pair Hasrlngt OtvUhjn, 1h* T«toplion« Numb«r la 639-4701-

212 / V. Melissa Holland and Janice C. Redish

observe performance to locate sources and types of errors,

and in which we revise and test forms to incorporate more
discourse level features.
Another empirical method that we are using to understand
public documents involves analyzing oral protocols of subjects
'thinking aloud' as they fill out forms. In the rest of this
paper, we present some very preliminary results from a proto-
col study of the processes and strategies used by experts
and novices to fill out an application form.
Forms as discourse—processes and strategies. Analysis of
thinking-aloud protocols is not a technique that has been
widely used in studying discourse. In searching for strategies
that readers use to understand text, researchers have gener-
ally used recall or recognition tests and then inferred what the
reader is doing from the patterns of omissions or intrusions.
Kintsch and van Dijk (1978), for example, constructed a model
of the way in which readers restructure a text to make it com-
prehensible. They acknowledge that 'the model does not
specify the details of the processes' (1978:364).
Looking again at our model (Figure 1), one can see that the
area we are interested in here is the box with the heavy
border--the processes and strategies that constitute the user's
behavior with the document. Traditional discourse studies only
infer what might be in that box; protocol analysis allows us to
peek inside. A thinking-aloud protocol is a record of the se-
quence of thoughts and behaviors a person engages in while
performing a task. In a thinking-aloud protocol, the subject
is asked to say whatever comes to mind as he or she does the
task (in this case, filling out the form). The protocol is tape-
recorded, transcribed, and analyzed by methods that assume
objectivity and interrater reliability.
Developed by Newell and Simon (1972) to look at problem-
solving, thinking-aloud protocols have also been a rich source
of information on writing process (Flower and Hayes 1979,
1980, 1981). As colleagues on the Document Design Center's
major project (funded by NIE), Flower and Hayes have used
protocols to look at how people read and interpret Federal
regulations (Flower, Hayes, and Swarts 1980).
We have been collecting protocols of subjects instructed to
'think aloud' as they attempt to complete the Federal govern-
ment's job application form--SF 171. This form ranked very
high in difficulty among 54 critical documents, selected in a
scientific sampling of critical and frequently used public docu-
ments. Subjects in this study vary in education, work experi-
ence, and English language proficiency. We are particularly
interested in the differences in the protocols of subjects who
are successful--who complete a form that is highly likely to be
well received (we call these people 'expert form users')--and
subjects who cannot complete the form or whose form is not
likely to be accepted by the government service agency (we
 Strategies for Understanding Forms and Public Documents / 213

call these people 'novices1 or 'inexperienced form users'). In

looking at this distinction, it is important to note that we judge
novices as unlikely to be successful not by their inherent quali-
fications, but by how they represent these qualifications on the
form: is it the appropriate and best representation of them-
selves for getting a government job?
We have only begun to analyze our protocols, but it appears
that the expert form users are operating on several levels as
they try to understand and complete the form--specifically, it
seems that at least three levels of strategies are involved in
successfully filling out this form. Figure 3 is a list of strate-
gies we have identified from the protocols we have thus far
analyzed. We have tentatively categorized these strategies into
three levels, as Figure 3 shows.

Figure 3. Strategies of form users (from a preliminary protocol analysis).

1. Decoding 2. Form-using 3. tWetacomments/
strategies: strategies: reality-testing:
lexicon creating examples about forms in general
syntax selecting among examples about the rhetorical
creating scenarios situation
proposing definitions about the way to fill
calling on memory out forms
rereading instructions about the writer's in-
editing/reviewing/reflect- tentions
ing on what is already
written
being concerned about
consistency

The lowest level is represented by 'decoding' statements--

indicating that the user was devoting attention to deciphering
the lexical and syntactic aspects of the form, attempting to
figure out word meanings and to disambiguate sentences.
Typical responses at this level include:
(1) Subject reads the instruction to 'estimate the amount
of time in each type of work' and then says, 'That's
gonna be a problem--what is a "type of work"?'
(2) Subject reads the item heading, 'lowest pay or grade
you will accept', and then says, 'I don't understand
about grade'.
At this level, we could say that the reader is in a word-
bound frame.
At the second level are comments reflecting what we call
'form-using strategies'. These are generally text-level pro-
cesses for understanding in which the user goes beyond words
and sentences and attempts to relate items across the form or
to draw on personal knowledge to clarify the meaning of items.
Processes we have found at this level include:
214 / V. Melissa Holland and Janice C. Redish

(1) Instantiating: in deciding how to answer the question

'When will you be available?' one subject commented,
'Let's say I feel that I must have at least, must give
my present employer 30 days notice . . . ' To understand
the question and decide on a response, this subject
conjured up a specific plausible situation.
(2) Calling on memory: 'My discharge from the reserves . . .
Let me see if I can resurrect it.'
(3) Watching for consistency: '[That answer] would cause
me a dilemma because it sounds like I'm saying "Yeah,
I'll take a part-time job", when over here someplace I
said I'm not going to take a part-time job'.
It is at this level that we would also put the 'scenarios' that
Flower, Hayes, and Swarts found in studying readers trying to
understand regulations. That is, subjects trying to interpret
an abstract definition or condition would try to express it as
a concrete event in which someone does something (like a piece
of a story). The work on scenarios shows that readers create
strategies for interpreting difficult documents by borrowing
from discourse frames that are familiar arid natural--here, the
narrative frame.
At this second level of comments, we can say that the reader
is in a form-bound frame.
Finally, we observed a third, higher level class of statements
which we call metacomments. These reflect the global strategies
that arise as the reader puts the document in a societal and
institutional context. The reader at this level is in a context -
bound frame, looking for the intention behind the questions,
predicting how answers will be interpreted, and showing aware-
ness of the rhetorical situation and the text type. These
strategies call on a very complex set of cognitive operations
which involve taking the perspective of another--specifically,
of the agency that produced and will make use of the returned
form.
Some of the linguistic cues to these global strategies in the
protocols of our form users are:
'I assume that if anything appears on a form it's there for
a reason.'
'I have to stop and think and guess at what the implications
[of different answers] are.'
'I interpret this in terms of what I think they're asking.'
'You're always gonna answer with what will get you the
most points.'
'The intent of the question is . . . '
'If I were applying to XXX, I would probably not put that
down. If it were YYY I probably would.'
Strategies for Understanding Forms and Public Documents / 215

These strategies clearly invoke the rhetorical setting of the

document and assume a two-party discourse between user and
agency.
From this preliminary analysis, we believe that three levels
of processing can be clearly distinguished in the protocols of
expert form users. The protocols indicate, however, that
subjects move rapidly from one frame to another; clearly, the
frames are interconnected. For example, subjects call on their
world knowledge for the information they use in the second-
level strategy of creating examples. They use the same kind
of knowledge at the top level, in predicting the purposes of
the agency and the criteria by which institutional decisions are
made.
Differences in processes of expert and novice form users.
Of greatest interest in our preliminary results is this finding:
the comments of the expert form user are far more likely to re-
flect the higher two levels of strategies than are the comments
of the novice form user. Comments fi*ora novices are more
likely to reflect the first level, decoding. Novices are likely
to say, rI don't know my Social Security number' and leave the
question blank. Experts use various recall strategies or go
get the card and copy down the numbers. Novices are likely
to say 'this word is ambiguous' and not try to disambiguate it.
Novices are far more likely than experts to make no meta-
comments at all--to ignore the rhetorical situation or intent.
This failure to make metacomments is reflected in other be-
havior. If the expert form users have a piece of information
they think will win points, they are likely to find a place to
put it down. The novice is far more constrained by the indi-
vidual items on the form, leaving out potentially point-winning,
personal information because 'they don't ask about that'.
We are speculating that the experts are successful because
they have available and use these higher level discourse
strategies; and, further, that these macrolevel strategies are
probably essential for coping effectively with forms--at least
with complicated forms like the SF 171.
Two reasons can be suggested for the finding that novices
tend to respond mainly on the first level. They may be
spending all of their energy in decoding, because just dealing
with the words and sentences overtaxes them. As Norman and
Bobrow (1975) have pointed out, humans appear to have a
limited capacity central processor which distributes a finite
pool of cognitive resources to ongoing mental tasks according
to need. These resources may be entirely depleted by the
words and sentences for poorer readers.
Of more interest to u s , a second reason for failing to oper-
ate in larger frames may be that the novice form users have
not developed the appropriate strategies, or have developed
them but do not apply them to forms. We hypothesize that we
can take the successful strategies we uncover in our protocol
216 / V. Melissa Holland and Janice C. Redish

analysis and teach them to the less successful form users.

Direct teaching of higher level strategies has been a successful
educational practice in reading (Bartlett 1978; Jones, Monsaas,
and Katims 1979).
Although our analysis of the protocols in this study is not
complete, we can foresee two practical outcomes of the study:
(1) we can use our results to recommend or develop instruc-
tion to teach people how to approach forms; (2) we can use
our results to recommend principles for designing forms.
How might the second application work? We think that forms
can be designed to facilitate the use of macrolevel strategies
and to exhibit more explicitly the discourse features--cohesive
devices, scenarios, instantiation--that we find successful form
users attending to in forms or building into their subjective
structuring of forms.
For example, in the Medicaid study discussed earlier in this
paper, many subjects not only failed to recognize some of the
critical text-level variables in the form, but they failed to see
the form as any kind of purposive communication at all. They
did not appear to understand what a form is. We rewrote the
introduction as a letter, on the theory that a letter format is
a more familiar communicative frame for this audience. The
letter frame seemed to help.
Both the analytical work we have done and the protocol
studies we are conducting show that forms can be productively
studied as a type of discourse. Although the work we have
reported on in this paper is barely a beginning to an under-
standing of forms and how people use them, we believe that
this research will eventually show us how to help both those
who must develop forms and those who must fill them out.
NOTE
1. The model in Figure 1 is a refinement of an earlier model
developed by Robbin Battison and Melissa Holland of the Docu-
ment Design Center. Brown, Campione, and Day (1981) have
recently presented a similar model for thinking about problems
that students have in learning from instructional text.
REFERENCES

Adams, M. J . , and A. Collins. 1979. A schema-theoretic

view of reading. In: New directions in discourse process-
ing. Edited by R. O. Freedle. Norwood, N . J . : Ablex.
Bartlett, B. J. 1978. Top-level structure as an organizational
strategy for recall of classroom text. Unpublished doctoral
dissertation. Arizona State University, Tempe, Ariz.
Bransford, J. , and M. Johnson. 1973. Considerations of
some problems of comprehension. In: Visual information
processing. Edited by W. G. Chase. New York: Academic
Press.
Strategies for Understanding Forms and Public Documents / 217

Bransford, J . , and N. McCarrell. 1974. A sketch of a cog-

nitive approach to comprehension; Some thoughts about what
it means to comprehend. In: Cognition and the symbolic
processes. Edited by W. Weimer and D. Palermo. Hillsdale,
N.J.: Erlbaum.
Brown, A. L., J. C. Campione, and J. D. Day. 1981. Learn-
ing to learn: On training students to learn from texts.
Educational Researcher 10.14-21.
Campbell, L. J . , and V. M. Holland, (in press) Understand-
ing the language of public documents, because formulas
don't. In: linguistics and the professions. Edited by
R. Di Pietro. Norwood, N . J . : Ablex.
Chafe, W. 1970. Meaning and the structure of language.
Chicago: University of Chicago Press.
Charrow, V. R., V. M. Holland, D. G. Peck, and L. V. Shel-
ton. 1980. Revising a Medicaid recertification form: A case
study in the document design process. Washington, D.C.:
American Institutes for Research.
Clark, H. H., and S. E. Haviland. 1979. Comprehension and
the given-new contract. In: Discourse production and com-
prehension, I. Edited by R. O. Freedle. Norwood, N . J . :
Ablex.
Flower, L. S., and J. R. Hayes. 1979. A process model of
composition. Document Design Project Technical Report No.
1. Pittsburgh: Carnegie-Mellon University. [Also published
in: Cognitive processes in writing. Edited by L. W. Gregg
and E. R. Steinberg. Hillsdale, N . J . : Erlbaum. 1980.]
Flower, L. S., and J. R. Hayes. 1981. The pregnant pause:
An inquiry into the nature of planning. Paper presented at
Annual Meeting of AERA, Los Angeles.
Flower, L. S., J. R. Hayes, and H. Swarts. 1980. Revising
functional documents: The scenario principle. Document
Design Project Technical Report No. 10. Pittsburgh.
Carnegie-Mellon University.
Grimes, J. E. 1975. The thread of discourse. The Hague:
Mouton.
Halliday, M.A.K. 1967. Notes on transitivity and theme in
English: II. Journal of Linguistics 3.199-244.
Halliday, M.A.K., and R. Hasan. 1976. Cohesion in English.
London: Longman.
Hobbs, J. 1979. Coherence and coreference. Cognitive
Science 3.67-90.
Holland, V. M. 1980. Revising a government document: The
case of the Medicaid recertification form. Paper presented
at the Symposium on Writing and Designing Documents: Re-
search and Practical Solutions, Carnegie-Mellon University,
October.
Holland, V. M. 1981. Psycholinguistic alternatives to reada-
bility formulas. Explanations from psycholinguistics. Docu-
ment Design Project Technical Report No. 12. Washington,
D. C.: American Institutes for Research.
218 / V. Melissa Holland and Janice C. Redish

Jones, B. F . , J. A. Monsaas, and M. Katims. 1979. Improv-

ing reading comprehension: Embedding diverse learning
strategies within a mastery learning instructional format.
Paper presented at Annual Meeting of AERA, San Francisco.
Kintsch, W. 1974. The representation of meaning in memory.
Hillsdale, N . J . : Erlbaum.
Kintsch, W., and T. A. van Dijk. 1978. Toward a model of
text comprehension and production. Psychological Review
85.363-394.
Mandler, J. M. , and N. S. Johnson. 1977. Remembrance of
things parsed: Story structure and recall. Journal of
Verbal Learning and Verbal Behavior 9.111-151.
Meyer, B . J . F . 1975. The organization of prose and its effect
on recall. Amsterdam: North-Holland.
Minsky, M. 1975. A framework for representing knowledge.
In: The psychology of computer vision. Edited by P. H.
Winston. New York: McGraw-Hill.
Newell, A., and H. Simon. 1972. Human problem serving.
Englewood Cliffs, N . J . : Prentice-Hall.
Norman, D. A., and D. G. Bobrow. 1975. On data-limited
and resource-limited processes. Cognitive Psychology 7.44-
64.
Rumelhart, D. E. 1975. Notes on a schema for stories. In:
Representation and understanding: Studies in cognitive
science. Edited by D. G. Bobrow and A. M. Collins. New
York: Academic Press.
Rumelhart, D. E . , and A. Ortony. 1977. The representation
of knowledge in memory. In: Schooling and the acquisition
of knowledge. Edited by R. C. Anderson, R. J . Spiro, and
W. E. Montague. Hillsdale, N . J . : Erlbaum.
Schank, R. C. , and R. P. Abelson. 1977. Scripts, plans,
goals, and understanding: An inquiry into human knowledge
structures. Hillsdale, N . J . : Erlbaum.
Stein, N. L., and C. G. Glenn. 1979. An analysis of story
comprehension in elementary school children. In: New di-
rections in discourse processing, Vol. 3. Norwood, N . J . :
Ablex.
Tannen, D. 1979. What's in a frame? Surface evidence for
underlying expectations. In: New directions in discourse
processing. Edited by R. O. Freedle. Norwood, N . J . :
Ablex.
van Dijk, T. A. 1972. Some aspects of text grammars. The
Hague: Mouton.
van Dijk, T. A. 1980. Macrostructures: An interdisciplinary
study of global structures in discourse, interaction, and
cognition. Hillsdale, N. J . : Erlbaum.
Winograd, T. 1977. Discourse. In: Cognitive processes in
comprehension. Edited by M. Just and R. Carpenter.
Hillsdale, N. J . : Erlbaum.
 SPEECH ACTIONS AND REACTIONS
IN PERSONAL NARRATIVE

William Labov
University of Pennsylvania

Through the better part of this century, linguists have taken

as their major focus the internal relations of linguistic struc-
tures, and1 they have drawn considerable profit from this en-
gagement. There is no indication of a weakening of the focus
on internal structure in the last several decades. Yet there
are several critical issues where linguistic analysis is necessar-
ily brought into confrontation with physical reality. I am not
speaking here of speech acts that work on the social under-
standings of speakers and listeners, where people can be seen
to suffer embarrassment, boredom, or insult. I am considering
rather the interrelations of language with violent acts that can
terminate conversational turns abruptly and bring an end to
any and all linguistic analysis.
In recent months, the Philadelphia newspapers have run ac-
counts of several violent events, where people have been killed
by their fellow-citizens without any apparent reason. Three of
these incidents took place on the highways. One case involved
James Harkins, a Philadelphia policeman of 20 years' standing.
Cam den police said that Harkins drove from a downtown
Camden parking lot into the path of a Philadelphia-bound
Transport of New Jersey bus at about 10:15 p.m.
Harkins' car cut in front of the bus, and driver Enrique
Cardona honked his horn, police said. Harkins then
stopped his car in the middle of the street, in front of
the bus, and demanded that Cardona open the door, they
said. When Cardona refused, Harkins walked to the other
side of the bus, drew a gun and fired, police said. Two
Camden police officers, Albert Ruderow and Howard Cad-
well, reportedly heard the gunshot and rushed to the bus,
where they encountered Harkins holding a gun. Harkins

219
220 / William Labov

allegedly aimed the gun at them when they ordered him to

drop it. The officers opened fire and wounded Harkins in
the chest, abdomen and arm.
--Philadelphia Inquirer, 3/12/81, B:l-2.
In another incident, a man shot and killed a truck driver
after an argument on the Schuylkill Expressway. Everyone
who has heard about these cases wonders what could be the
cause of this sudden and apparently senseless violence. People
suggest that there is a general climate of violence, a sense of
frustration and despair in American life, or that Americans are
simply going crazy these days. There may be far-reaching
sociological or economic factors that go beyond the details of
face-to-face interaction. But there are also causal sequences
that are embedded in the structure of the immediate confron-
tation of one person and another. And in all the cases re-
ported here, words were exchanged in that confrontation.
Things were said, and immediately after, things were done--a
close connection between speech acts and actions that implies
a causal relation. (Though speech acts are, of course, kinds
of actions, I refer throughout to nonverbal physical actions as
'actions' and refer to verbal actions as 'speech acts'. 2 )
We do not know just what was said in any of these cases, so
as linguists we are in no better position than anyone else to
guess at the relations between language and action from the
newspaper accounts. But we can draw inferences from com-
parison with a wider range of data concerning similar incidents,
drawn from narratives of personal experience obtained in
various studies of the speech community.
In this discussion I draw on three narratives of events where
speech acts alternate with violent physical actions. Each of
these narratives has been observed to hold the attention of
listeners to a remarkable degree. I analyze the sequential
structure of these narratives in order to get at the general
principles that underlie the relations of speech and action and
the particular sequences where speech leads to violence. In
doing so, I try to keep in the foreground the fact that we are
dealing with reports of events, not observations of the events
themselves. The complexity of the many-layered relationship
between reality and reported reality cannot be overestimated.
I have already indicated some of the ways in which narratives
are thoroughgoing transformations of reality (Labov 1972:
Chapter 9). How then, can narratives of personal experience
be used to illuminate relations of speech and action in the
world reported by narrative?
Goffman (1974) gives us the most sophisticated view of the
many-sided relationship between reality and reports of reality.
On the one hand, reports of events are framed in ways that
are highly conventionalized. On the other hand, behavior it-
self incorporates imitations and replayings, strips derived from
those conventional representations (Goffman 1974:560-562).
 Speech Actions and Reactions in Personal Narrative / 221

Narrative accounts are not unreal accounts in the sense of be-

ing unrelated to reality. They are framed accounts, and with
proper attention to those frames and the rules of transforma-
tion, we can begin to reconstitute their relations to the wider
frames outside of the narrative context.
Attention to the social setting of the narrative is then an
essential part of this analysis. All of the narratives to be dis-
cussed here are drawn from tape-recorded interviews. They
therefore include the effects of formal observation, as speakers
adjust their speech to the norms appropriate for such observa-
tion. At the same time, all of our field work indicates that
these effects are minimal in narratives of personal experience,
highly dramatized and objective accounts of events actually ex-
perienced by the speaker. For this reason, many of our field
methods (Labov 1981) are concerned with techniques for elicit-
ing such narratives. In face-to-face interviews, they yield the
closest approach to the phonology and syntax of the vernacular,
the form of language used among intimate peers when the, mini-
mum degree of attention is given to superposed norms of speech.
As we will see, they also yield information on vernacular norms
of behavior; not in the normative rules or explanations that are
consciously reported, but in the unstated assumptions that
govern sequences of reported events. 3
1. Narratives of violent events. In a discussion of violence
and the danger of death, Harold Shambaugh of Cleveland gave
me an account of something that had happened to him in Buenos
Aires, when he was in his early twenties and in the Merchant
Marine. He was 31 at the time I talked with him.
(1)
(What happened in South America?)
1 Oh I w's settin' at a table drinkin'.
2 And--uh--this Norwegian sailor come over
3 an' kep' givin' me a bunch o' junk about I was sittin 1
with his woman.
4 An' everybody sittin' at the table with me were my
shipmates.
5 So I jus' turn aroun'
6 an' shoved 'im,
7 an1 told 'im, I said, 'Go away,
8 I don't even wanna fool with ya.'
9 An' next thing I know I'm layin' on the floor, blood
all over me,
10 an' a guy told me, says, 'Don't move your head.
11 Your throat's cut.'
There are many things that might be said about Shambaugh's
narrative in the attempt to explain its powerful impact on
222 / William Labov

audiences, in the concise use of language, in its extraordinary

shift of visual perspective. For my present purpose, the most
important feature is the pattern that relates speech acts to
actions.
This narrative shows the intimate relations of speech acts and
actions that is the main focus of this discussion. In this case,
speech and action are simultaneous in lines 5 and 6, associated
more or less like word and gesture. Shambaugh's refusal to
deal with the Norwegian sailor is conveyed by the imperative,
'Go away', the statement 'I don't even wanna fool with ya', and
the action of shoving the other away. There is nothing
problematic about this association: it is normal for words to
be reinforced by gestures. The problem in understanding
Shambaugh's narrative is to account for the sudden increase in
the level of violence from line 8 to line 9. The linguistic ques-
tion is then whether there are any properties of the speech
acts reported that can contribute to this explanation.
Among the various narratives that I have dealt with over the
last decade, one stands out in its strong effect on listeners.
It is an account given to me by Jacob Schuster, a retired post-
man from New York City, of a sudden and violent conflict that
broke out in his family just after his father's death. In a
hundred retellings, to audiences of one or one thousand, the
result is the same: after the first few sentences, the small
movements, whispers and coughs that establish the normal
noise level of an audience come to an end, and there inter-
venes the total silence that marks the undivided attention of
listeners.
(2)
(What happened?)
1 My brother put a knife in my head.
2 Like kids, you get in a fight,
3 and I twisted his arm from behind him.
4 This was just a few days after my father had died,
5 and we were sitting shive. **
6 And the reason the fight started,
7 --this was in Coney Island--
8 a rat ran out in the yard,
9 and he started talk about it,
10 and my mother had just sat down to have a cup of
coffee.
11 And I told him to cut it out.
12 'Course kids, you know, he don't hafta listen to me.
13 So that's when I grabbed him by the arm
14 and twisted it up behind him.
15 When I let go of his arm,
16 there was a knife on the table,
17 he just picked it up
 Speech Actions and Reactions in Personal Narrative / 223

18 and he let me have it.

19 And I started bleed like a pig.
20 And naturally, first thing was, run to the doctor,
21 and the doctor just said, fOh just about this much more,'
22 he says, 'and you'd a been dead.'

An understanding of this narrative and its effect on audiences

demands a study that goes beyond the scope of this analysis.
Here I am primarily concerned with the relations of speech acts
to action. The escalation of violence in Schuster's narrative,
Example (2) , follows the same pattern as that outlined for
Shambaugh in Example (1).
In this case, the first physical action is not simultaneous
with a speech act, but follows in sequence. Line 14 is a r e -
sponse to 12, which does not report a speech act directly, but
implies one. Again, the problem is to explain the escalation of
the level of violence from 14 to 18. We might say for both
Shambaugh's and Schuster's narratives that it is appropriate
for one action to follow another; but in the absence of a gram-
mar of action, we can say from our own knowledge of American
culture that stabbing someone with a knife is not a typical re-
sponse to having an arm twisted. The challenge for us then
is the same: are there any characteristics of the speech acts 9,
11, and 12 that would motivate or explain the passage from 14
to 18?
The third narrative that deals with a sudden outbreak of vio-
lence was told me by Joanna Williams, 31, of Morgantown, West
Virginia. We were not talking about the danger of death, but
about this very subject, the level of violence in Appalachia.

(3)

They didn't believe in calling the law or anything like that.

They just took things in their own hands. (Did you ever
see any shooting of that sort?)

1 Oh yes, I can remember real well.

2 I was just a girl.
3 In fact, it stayed with me quite a while.
4 Well there w's a fellow, his name was Martin Cassidy and
Bill Hatfield.
5 Mr. Hatfield's mother give him some money
6 and told him to go get a bushel of peaches.
7 And he went down to Martin's house.
8 And Martin had some moonshine there.
9 Back down there they make their own liquor, you know.
10 So--we call it moonshine.
11 Today they call it white lightnin'.
12 But at that time we called it moonshine.
13 And--I remember real well what happened.
14 Bunch of us kids was out there playin'.
224 / William Labov

15 And no one meanin' any harm about it.

16 but anyway Mr. Hatfield--Mrs. Hatfield come down
17 And took away her money from Mr. Hatfield,
y'know, for the peaches,
'cause she know t h a t - -
he was gonna buy drinks with it.
18 And Mr. Cassidy was layin' out t h e r e in the y a r d .
19 And Mr. Cassidy just looked up
20 and he said to Bill, just--just jokin', just in a kiddin'
way, he said,
'Ah h a h , ' he s a y s , ' t h a t ' s another dollar bill you
won't get to spend for a drink, h a h . '
21 And Bill said, 'I'll fix you, you s o - a n d - s o . '
22 So he walked to Martin Cassidy's HOUSE, his own house,
23 come out with a double-bitted axe,
24 hit him down across t h e head once,
25 t u r n e d over
26 and hit him again,
27 then throwed the axe down
28 and r u n down through the woods.
29 J u s t over two dollars
that he was sent for peaches with.

This account is from a t h i r d - p a r t y witness of the violence

r a t h e r than one of the main actors. But the effect on a young
girl was profound. Through her eyes we can isolate an alter-
nation and combination of speech acts and action. Joanna
Williams' account does not give us a direct view of the speech
act implied in line 5, but by inference t h e r e must have been a
verbal request with the t r a n s f e r of money. There may have
been a speech act when the money was taken away in line 17,
but it is not r e p o r t e d . The speech acts of 20 and 21 are r e -
ported in clear detail, and the final action, lines 22-28, as well.
Here the analytic problem is to account for the escalation of
violence in response to a speech act: the insult ( 2 0 ) , i n t e r -
preted as a joke by the o b s e r v e r , does not seem in any way
enough motivation to explain the killing. We do not know the
past history of Hatfield and Cassidy. But the question for us
is whether t h e r e are any characteristics of t h e speech acts in-
volved that would make t h e whole sequence part of an u n d e r -
standable pattern instead of incomprehensible violence.
The p a t t e r n s presented so far have been independent of any
theoretical framework. To advance f u r t h e r , it is helpful to
review briefly the characterization of narrative that has been
presented in previous publications, and to develop a further
framework that facilitates the connection between the rules for
speech acts of Labov and Fanshel (1977) and the analysis of
narrative as a means of recapitulating and transforming ex-
perience. The approach to speech acts as forms of action that
is characteristic of Goffman (1976) is followed t h r o u g h o u t . The
more abstract interpretation of speech acts and their sequences
 Speech Actions and Reactions in Personal Narrative / 225

then yields some insight into the relation between speech and
action, the emergence and escalation of violence.
2. The structure of narrative. The view of narrative
structure that is used here begins with the definition of narra-
tive developed in Labov and Waletzky (1967), their analysis of
the temporal organization of narrative, and the role of evalu-
ation in narratives of personal experience. The concept of
REPORTABILITY presented there is developed further, and
the notion of a MOST REPORTABLE EVENT is introduced as
the generating center of narrative structure. I then analyze
more closely the other events that make up the complicating
action, with the help of the distinction between OBJECTIVE
and SUBJECTIVE events. A sequence of objective events is
isolated. This leads in the following section to the recognition
of speech acts as elements in the narrative sequence, and the
analysis of the higher level interpersonal actions that operate
on social identity and social status.

2.1 Temporal organization. As conceived by Labov and

Waletzky (1967), NARRATIVE is a technical term, referring to
one of many linguistic devices available to speakers for the
recapitulation of past experience. Narrative does this through
the basic rule of narrative sequencing (Labov and Fanshel
1977), which allows the listener to infer the reported temporal
order of past events from the temporal sequence of clauses in
the report of those events.
2.1.1 Temporal juncture. If two clauses occur in a given
order, and a reversal of that order leads to a change in the
semantic interpretation of the order of the reported events,
they are said to be separated by a TEMPORAL JUNCTURE.
2.1.2 Narrative clauses. Only a limited number of clause
types participate in temporal juncture and so serve as NAR-
RATIVE CLAUSES. Subordinate-independent pairs like 'When
I let go of him arm . . . he picked it up' (Schuster, in Example
(2), lines 15,17) may occur in the same temporal order as the
original events, but reversal of that order produces no change
in the order of semantic interpretation, e.g. 'He picked it up
when I let go his arm'. Narrative clauses are independent
clauses with verbs in the indicative mood and (in English) one
of three tenses: the preterit, the historical present, 5 or the
past progressive. Beyond the test of temporal juncture, narra-
tive clauses can be identified by the criterion that they are
appropriate answers to the criterial question, 'And then what
happened?' The sequence of narrative clauses forms the
COMPLICATING ACTION.
226 / William Labov

2.1.3 Definition of narrative. A NARRATIVE is then a se-

quence of two or more narrative clauses, that is, a sequence
of clauses separated by one or more temporal junctures.
2.2 Orientation. Most narratives give orienting information
on four types of data: the time, the place, the participants in
the action, and their general behavior before or at the time of
the first action. This information is usually concentrated at
the beginning, in an ORIENTATION section--Shambaugh,
Example (1), line 1; Schuster, Example (2), lines 4-7;
Williams, Example 3, line 4. But some orienting information
can be placed later in the narrative. The location of the knife
in Schuster's story is not given until line 16. In the Williams
narrative, Cassidy's behavior is not described until line 18,
Though the displacement of orientation can sometimes be ac-
counted for on simple cognitive grounds, it often appears to
serve an evaluative function.
2.3 Evaluation. Narratives of vicarious experience and
narratives of young children are often limited to complicating
action. Such narratives are often heard as not having a point.
As narrators mature, they use an increasing number of evalua-
tive clauses that do not refer to an event that occurred, but
rather to one that did not occur (Labov 1972). The contrast
between what did occur and what did not but might have
occurred serves to evaluate the narrative. Negatives, futures,
modals, and comparatives thus enter into narrative structure.
Other evaluating devices group several actions or behaviors in
the same time unit, using participles and other nonfinite verbs.
The main point or focus of the narrative, as told, is often
indicated by the concentration of a number of evaluating clauses
in an EVALUATION SECTION directly before a particular narra-
tive clause.
In Examples (1) and (2), the evaluation section is placed at
the end of the narrative, and merged with the final action or
RESOLUTION. In Shambaugh, lines 8-9, we observe the sus-
pension of the action (with progressive and adverbial phrase).
The seriousness of the situation is then stated by a negative
imperative: 'Don't move your head'. In Schuster, line 21, we
find a comparative ('this much more') and a complex modal
('would have been'), evaluating the seriousness of the situation
by comparison with what might have taken place.

2.4 Abstracts. Many narratives are preceded by a brief

summary statement of the substance of the narrative as viewed
by the narrator. The focus of the abstract is normally the
same as the point of the narrative, but not necessarily, since
it is more closely linked with the preceding utterance of the
other person and the insertion of the narrative in the conver-
sation. An ABSTRACT of this kind is given in Schuster,
lines 1-3.
 Speech Actions and Reactions in Personal Narrative / 227

2.5 Analysis of the complicating action. Much of the atten-

tion of previous analyses has focused on the elaboration of
narrative beyond the fundamental sequence of narrative clauses.
The main thrust of this discussion is in the other direction: to
reduce the narrative to its skeletal outline of narrative clauses,
and to outline 4he generating mechanism that produces the
narrative backbone.
2.5.1 Reportability. When people tell narratives, they
occupy a larger portion of social time and space than in most
other conversational turns. There have been various discus-
sions of how narrators alert listeners to the impending narrative.
Here one must be concerned with the other end of the question:
AFTER the narrative is finished, do listeners accept this occu-
pation of conversational time as justified? If the response to a
narrative is 'So what?' or 'What are you getting at?', it must
be considered a failure, and much of narrative is organized to
forestall such a response. The evaluation section can contribute
a great deal to this end, but the fundamental burden of achiev-
ing acceptability is on the character of the events being re-
ported .
We can classify all narrative clauses into two types in terms
of the appropriateness of complementary sets of responses from
listeners:
In response to Type A, expressions of ordinary under-
standing: 'I see', r Uh-huh', 'Naturally' . . . 6
In response to Type B, expressions of ordinary surprise:
'Really?', 'Is that so?', 'You don't mean it! 1 , 'No kidding!',
etc.
Thus, Williams' first clause, line 1, is Type A:
1 Oh I w's settin' at a table drinkin'.
(I see.)
And the same holds true for the clause in line 2, but not for
line 3, which is Type B:
2 And--uh- -this Norwegian sailor come over
(Uh huh.)
3 An1 kep' givin' me a bunch o' junk about I was sittin'
with his woman.
(I see.)
Here, a response of 'Really?' would be more appropriate.
Shambaugh's last sentence, in line 11 of (1), is even more
clearly Type B. After 'Your throat's cut', it would be out of
the question to say, 'Naturally', 'I understand', or 'Uh-huh'.
Ordinary expressions of surprise like 'Really?' are not strong
enough here. A good number of listeners respond with an
228 / William Labov

ingressive gasp of breath. My response on tape is, in fact,

a devoiced 'Wow!' Evidently, reportability is not a binary
dimension but a scalar one.
These reactions are not to expressions but to the report of
events. Events of Type B are REPORT ABLE events. If a
narrative contains such an event, it is not possible for some-
one to respond at the end with 'So what?' A normal narrative,
then, which has succeeded in the tasks of holding the atten-
tion of an audience and justifying the time taken to tell it,
contains at least one reportable event. The reportability of a
narrative is equivalent to that of the maximally reportable
event in it.
The assessment of REPORTABILITY does not rest on the
objective grounds that I would like, first, because it is evi-
dently relative to the culture of the narrator. For cultures
very different from our own, we may have only a dim view of
what is reportable. The relativity of this concept is also
obvious when we listen to children's narratives. Second, re-
portability is relative to the social occasion. Higher degrees
of reportability are required to hold the floor when other re-
portable matters are on hand than when nothing else is happen-
ing. 7 Third, the judgments on reportability that I use here
are intuitive. Though I have some objective evidence from
participant-observation and the reactions of audiences, I am
primarily using my own reactions in judging the appropriate-
ness of responses.
Against this subjectivity of response there is the generaliza-
tion that death and the danger of death are among the two or
three major themes of human concern and interest for all of
the cultures that we know. It is for this reason that these
themes play an important role in our interviews. The central
events of the three narratives given here are highly report-
able; they are far from the margins of reportability where the
subjective character of such judgments becomes a problem. It
is only under very special and terrible circumstances that
someone's death is not reportable: where we would agree with
Macbeth that 'there would have been a time for such a word'.
2.5.2 Credibility. Reportable events are almost by definition
unusual. They are therefore inherently less credible than non-
reportable events. In fact, we might say that the more re-
portable an event i s , the less credible it is. Yet credibility is
as essential as reportability for the success of a narrative. A
narrative that is judged entirely false, 'nothing but a big lie',
does not have the impact or acceptability of a narrative that is
considered essentially true. And except for certain special
storytelling traditions, 8 the reputation of the narrator suffers
if he or she is judged to be a liar. For narratives of personal
experience, this situation raises the question of whether there
is an inherent contradiction between reportability and credi-
bility.
 Speech Actions and Reactions in Personal Narrative / 229

There are two main directions in which we can look for the
resolution of this question. One has to do with the credibility
of the evaluation provided in the narrative, and the scale of
objectivity of evaluation (Labov and Waletzky 1967). In this
discussion, I look in the other direction: at the construction
of the complicating action that precedes the reportable event.
Credibility is seen to rest on a series of causal relations that
lead from the initial situation to that maximally reportable
event.
2.5.3 The generation of a narrative. Every narrative is
about something. The abstract, if there is one, tells us what
it is about. It is usually about the maximally reportable
event, though the relocation of the evaluation section can
alter this perception on the part of listeners (Labov and
Fanshel 1977). When someone decides to tell a narrative, he
or she has normally decided to tell a story about that event.
Though that decision has not yet created a narrative by my
definition (Section 2.1.3), the reportable event holds a cen-
tral position in any formal characterization of a narrative. One
can write N •> E ( r ) , where E is a series of ordered events in
temporal sequence and r is the index number of the most re-
portable action. The basic solution to the problem of achieving
credibility of E(r) is to provide an account of the events that
led up to it, that i s , the real-world conditions that gave rise
to this unusual event. It might be possible to construct a
narrative by a series of adjuncts to E ( r ) , each one answering
the question 'How did that come about?', of the form E(r) •*•
E(r-l) + E(r), where each E(r-l) would be an event that led
to the simple or complex series of events that follow. Yet the
major problem is to know where to stop in this process, often
expressed by the phrase, 'Where should I begin?' The selec-
tion of the orientation section by the narrator is one of the
crucial steps in the construction of the narrative and the
theory of causality that supports it. In general, the speaker
searches for the first set of general conditions where the
question 'How did that come about?' might be appropriately
answered by, 'That's the kind of thing that we usually did
then'. Thus, Shambaugh's story begins with him sitting with
his shipmates at the table in Buenos Aires drinking, the kind
of thing they usually did in port.
Given these initial conditions, the speaker has to find a
credible way to bridge the gap between them and the report-
able event with a series of intervening actions. This suggests
a set of four rules:
(1) N + Or + E(r)
(2) Or + Or + E(l)
(3) E(i) -+ E(i) + E(i+1)
(4) E(i) -»• E(i-l) + E(i)
230 / William Labov

These rules have no intrinsic order. They register the fact

noted earlier, that the first decision in forming a narrative is
to select the orientation that ultimately makes understandable
the reportable event. An orientation is frequently coupled with
or superimposed on a first event. Then there is no control
over whether actions are picked as consequences of previous
actions by Rule (3) or as the causes of later events by Rule
(4). At first glance, it might seem necessary to impose an ex-
ternal condition that no E(i) can be earlier than an E ( i - l ) , but
there is no need for such a condition. If E(i) precedes E(i-l)
in time, the past perfect tense is used, as the result of just
such an overlap in the construction of a chain of events.
2.5.4 Objective and subjective events. The process I have
described characterizes the generation of a skeleton of events
that forms the complicating action of a narrative, each event
separated by temporal juncture from the other. Each event
can be thought of as the answer to a question, 'And then what
happened?' or 'How did that come about?1 Up to this point,
the process can be thought of as relatively neutral to the many
transformations of these events that are characteristic of narra-
tives as delivered. Narratives often show the insertion of
evaluative material throughout the text, not only in the evalu-
ation section, additions that are not directly related to the
business of telling what happened. But other functions of
narrative--the presentation of self, the maximization of the
position of the narrator, the polarization of antagonist and
protagonist, are also facilitated by the selection of orientation
and the arrangement of narrative clauses.
In completed narrative, one can distinguish two kinds of
events, which I call OBJECTIVE and SUBJECTIVE EVENTS.
The distinction can be seen most clearly if we refer to potential
testimony in a court of law. A report of an objective event
can be contradicted by a witness who was present at the time;
a report of a subjective event cannot. Things that narrators
feel or say to themselves are then subjective ('I had this feel-
ing that . . . ' , 'I said to myself, "There'll be times I can't put
up with this . . . " ' ) . What is said aloud to others is objective.
Most physical events are objective, but there are subjective
ones. The report, 'I took this girl and started to move her
away' cannot be contradicted, since the mere intention to move
is enough to justify 'start + FOR/TO + Verb', as opposed to
'I started9
moving her away', which demands some physical
motion. Actions of turning and orientation are often subjec-
tive. 1 0
The three narratives considered here show a remarkably high
proportion of objective events. Subjective events that can be
identified are Shambaugh, line 5, 'I jus' turn aroun", and
Williams, lines 1, 13, rI (can) remember real well (what
happened)'.
 Speech Actions and Reactions in Personal Narrative / 231

2.5.5 Instrumental events. The sequence of narrative

clauses that appears in the surface text of a narrative often
shows events that are predictable means of implementing other
events. Thus Schuster, line 13, ' . . . I grabbed him by the
arm' is an inevitable antecedent of line 14, '(I) twisted it up
behind him'. Such instrumental events are not appropriate
responses to the generating question, 'How did that come
about?' If that question is addressed to 'I twisted it up be-
hind him', the answer would be line 12, 'He didn't listen to
me'.
2.5.6 The basic sequence of complicating action. This mode
of analysis can be used to reconstitute the complicating action
for the three narratives as an objective event sequence (OES),
leaving aside evaluative clauses, subjective events, and in-
strumental events. Events located as simultaneous, between
the same two temporal junctures, are restated as a single com-
pound event. Each event is then separated from the others
by a temporal juncture. This OES then represents the cogni-
tive framework that is provisionally accepted as a true repre-
sentation of the events reported in the narrative. If the se-
quence is then coherent, each event will be an appropriate re-
sponse to the criterial question addressed to the preceding
event, 'What happened then?' (forward sequencing) and the
criterial question addressed to the following event, 'How did
that come about?' (causal sequencing).
We can now see whether an OES constructed for each of the
three narratives is more coherent than our original reading of
the text. I represent each of the events by the independent
narrative clause(s) with preterit heads.
OES for Shambaugh narrative, Example (1):
Or 1
E(l) 2 A Norwegian sailor come over
E(2) 3 he kep' givin' me a bunch o' junk
about I was sittin' with his woman.
E(3) 6,7 I shoved 'im an' told 'im, 'Go away'
E(r) 11 A guy says, 'Your throat's cut.'
Certainly, E(l) through E(r) answer the first test of forward
sequencing, but the causal sequence fails in the same way when
we address the question 'How did that come about?' to E(r).
OES for Schuster narrative, Example (2):
Or 4-7
E(l) 8 A rat ran out in the yard,
E(2) 9 My brother started to talk about it,
E(3) 11 I told him to cut it out.
232 / William Labov

E(4) 12 (brother refuses)

E(5) 14 I twisted his arm
E(r) 18 He let me have it with the knife
E(7) 19 I started bleeding
E(8) 20 We ran to the doctor
E(9) 21 The doctor said, 'just that much more,
and you'd a been dead.'
Again, the forward sequencing is a coherent series of events,
but the causal sequence seems no more coherent when we ask
if E(r) is a likely outcome of E(5).
OES for Williams narrative, Example (3):

Or 4
E(l) 5 Mrs. Hatfield gave Hatfield money
6 and told him to get a bushel of peaches
E(2) 7 He went down to Cassidy's store
E(3) 16 Mrs. Hatfield come down
E(4) 17 She took her money away from Hatfield
E(5) 20 Cassidy said, 'That's another dollar . . . '
E(6) 21 Bill said, 'I'll fix you'
E(r) 24 He hit Cassidy with an axe
E(8) 26 He hit him again
E(9) 28 He run down through the woods
Here again the causal sequence fails when we address the ques-
tion 'How did that come about?' to E(r). The interpretation of
E(6), of course, depends on the seriousness of the offense
E(5). The narrator interprets E(5) as a joke. Accepting this,
we would interpret E(6) as a joking threat. If Hatfield had re-
turned with a bucket of water and thrown it on Cassidy, we
might understand that as an appropriate fulfillment of his
threat. But as it is, we are left with the same sense of shock
and wonder that Joanna Williams expresses: people behaved in
violent and incomprehensible ways in those days.
It would seem therefore that the construction of an objective
event sequence has not so far advanced our understanding of
the violent reactions in these narratives. The sense of
strangeness remains: these people did not behave as we ex-
pect people to behave. This effect is not due to the trans-
formation of the narrative through the insertion of interpreta-
tive or evaluative material, but seems inherent in the events
themselves. In Section 3, I use that event structure to move
to a higher level of abstraction, examining the sequencing of
speech acts and actions, to see if further comprehension is to
be gained at that level.
3. Action and reaction. The preceding sections have con-
firmed the general principle that there are no (necessary)
connections between utterances (Labov and Fanshel 1977).
 Speech Actions and Reactions in Personal Narrative / 233

Though there are tying relations between sentences--anaphoric,

elliptic--the coherence of discourse is not established at this
level but at a more abstract level of representation. Ultimately,
the cohesion of the three narratives that we are examining does
not depend on the sequence of narrative clauses but on the se-
quences of speech acts and actions that the narrative presents.
The first step in the analysis of the reported actions is to
translate the quoted speech acts into their least abstract repre-
sentation at the level of action. All three narratives, like most
accounts of human interaction, center around REQUESTS FOR
ACTION and sequential responses to them. In the Shambaugh
and Schuster narratives, these are followed immediately by vio-
lent action; in the Williams narrative, there intervenes another
kind of speech act, an INSULT.
3.1 The rule of requests. Most of the apparatus that is
needed for the analysis to follow is contained in the basic rule
for requests for action formulated in Chapter 3 of Labov and
Fanshel (1977). l l It is a rule of interpretation that states the
conditions that lead the addressee to believe that a request for
action has been made. The rule is presented here in a some-
what simplified form, omitting details that are not relevant to
the analysis to follow.
Rule of Requests:
If A addresses to B an imperative specifying an action X
at a time T, and B believes that A believes that
(conditions based on needs and abilities)
<1> X should be done,
<2> B has the ability to do X,
(conditions based on rights and obligations)
<3> B has the obligation to do X, and
<4> A has the right to tell B to do X
then A is heard as making a valid request for action X.
If we now turn to the objective event sequence or OES of
Williams' narrative, given at the end of Section 3, we can be-
gin to establish the series of actions represented by the re-
ported objective events. It is not difficult to identify E(l)
Mrs. Hatfield gave Hat field some money and told him to
get a bushel of peaches.
as a request for action made by Mrs. Hatfield of Billy Hatfield.
E(l) implements condition < 2 >. We can infer that Mrs. Hatfield
is the head of the household and is therefore an authority on
the need for peaches < 1 >. It is also apparent that her son has
234 / William Labov

< 3 > the obligation and she has < 4 > the right to tell him to get
peaches. We can therefore write the first of a series of actions
as an indexed A series at the higher level of abstraction than
the event sequence:
A(l)Mrs. Hatfield makes a request for action of
Billy Hatfield to buy a bushel of peaches.
Events (2) and (3) are Hatfield's and Mrs. Hatfield1 s descent
to Cassidy's house. They are simple locomotions that are essen-
tial to show that the participants in the interaction were in con-
tact and the audience present. No audience is indicated for
E ( l ) , but Cassidy and the children are witnesses to E(4), Mrs.
Cassidy taking back the money. This action can be considered
the cancellation of the request A(l) since one of the condi-
tions--Hatfield's ability to buy peaches—is removed. The re-
mark E(5) of Cassidy is an insult that can be reconstructed
using the principle of conditional relevance of negation:
That's one more dollar you won't spend for drink because
your mother took the dollar away.
It presupposes
You would have spent that dollar for drink if your
mother had not taken the dollar away.
This leads by inference to the general proposition, 'You are
the kind of person that spends every dollar that comes into
his hands for drink'.
Rules for threats and promises (Searle 1969) allow us to
identify E(6) as a threat, and the reportable actions E(r,8) do
not need deeper analysis to be identified as killing or man-
slaughter. 1 2 We then have derived the sequence of reported
actions:
A(l) Mrs. Hatfield makes a request of action of Billy Hat-
field to buy a bushel of peaches.
A(2) Mrs. Hatfield cancels the request by taking back the
money.
A(3) Cassidy insults Hatfield as an irresponsible alcoholic.
A(4) Hatfield threatens Cassidy.
A(5) Hatfield kills Cassidy.
This reduction of the narrative text to actions does not yet
illuminate the fundamental problem of the passage from A(3) to
A (5). But given this more abstract characterization of the
events, we can turn to the rule of requests for a further
understanding of what is reported to have taken place.
By A(l) it is established as shared knowledge that Mrs.
Hatfield has the right to request Billy Hatfield to go to buy
 Speech Actions and Reactions in Personal Narrative / 235

peaches, and he has the obligation to do so. It is further evi-

dent that she is the head of the household, since by A(2) it
appears that he has no money of his own. He is then in the
relationship of dependent member of the household to head of
the household, though he is a grown man. That a grown son
should have no regular income and that his mother should have
to scrape together a few dollars for food is a normal situation
in Appalachia. The important thing to note is that it is an
asymmetrical status well understood by others in the commun-
ity.
In the most probable interpretation of the reported event,
the action of taking back the money A (2) represents a cancel-
ling of the request A ( l ) . But the reason given for this by
the narrator and by the insult of A(3) is that even when Hat-
field has the money he does not have the ability to carry out
that request: that he is not a dependable person. The re-
ported action A (2) reduces his status from dependent member
of the household to something lower: a no-account person.
Furthermore, this act is performed publicly. If we continue
to accept provisionally the objective event sequence as an ac-
count of events that did occur, we are in a position to clarify
the causal sequence of actions involved.
The act of cancelling the request has a social meaning at a
higher level of abstraction. It is a CHALLENGE: an asser-
tion of a condition that, if true, lowers the esteem of a person
in a status or removes him or her from that status (Labov and
Fanshel 1977). The sequential response to a challenge is a
DEFENSE and often a COUNTER-CHALLENGE. But Billy Hat-
field apparently had no options open to him for defense or
counter-challenge, and the result is a profound and predictable
state of rage.
The fact that this rage found a violent outlet against Martin
Cassidy is not predictable. Nor can we say under what condi-
tions violent rage will result in such violent action. But it
seems from an analysis of the conditions governing the request
and withdrawal of the request that the controlling dynamic of
this situation is one of social status and challenge to social
status. Whatever verbal sequence was available for Billy to
deal with his mother had come to an end, and the route to
violent action was then taken.
I have less information on the social and cultural background
of the participants in the Williams story than in the other two
cases. The Shambaugh narrative deals with an encounter be-
tween strangers in a bar. The pattern of actions is easy to
establish with the help of the Rule of Requests and the aux-
iliary Rule for Indirect Requests, as given in Labov and Fan-
shel (1977: Chapter 3).
236 / William Labov

Rule for Indirect Requests:

If A makes to B a Request for Information or an assertion
to B about
<a> the existential status of an action X to be performed
by B,
<b> the consequences of performing X,
<c> the time T that X might be performed by B, or
<d> any of conditions for a valid request for X given in
the Rule of Requests
and all other conditions are in effect, then A is heard as
making a valid request of B for the action X.
The substance of event E(2) can then be understood as a re-
quest for action made indirectly.
(he) kep' givin1 me a bunch o' junk about I was sittin'
with his woman.
In Example (1), the Norwegian sailor has explained to Sham-
baugh that he is sitting with his woman. It can be easily in-
ferred that he believes that Shambaugh should not be sitting
with his woman but sitting somewhere else. Under subsection
< d >, this is an indirect assertion of one of the conditions of
the Rule of Requests, < 1 > the need for an action to correct
this situation. It is evident < 2 > that Shambaugh has the
ability to do so. The crucial questions under dispute concern
< 3 > the obligation of Shambaugh to move and < 4 > the sailor' s
right to make the request.
There are ways and means for putting off or refusing re-
quests in an accountable manner, which recognize the rights
and obligations of the other, and I deal with these later in the
Rule for Putting Off Requests. But Shambaugh does not re-
port himself using any of these means. He does not even dis-
pute the Norwegian sailor's rights. Instead, he refuses to hear
the sailor's talk as a valid request and says that he does not
want to fool with him. How can we account for this mismatch
of social perception between the Norwegian sailor and Sham-
baugh?
The situation in the Buenos Aires bar is not unique. I have
found a number of parallel narratives from Philadelphia, from
Scotland, and elsewhere where working-class people meet in
bars, at dances, at celebrations of various kinds. Most of the
narratives about the fights that break out give only a partial
view of the overall situation, and I did not understand it my-
self until it was explained to me by Joe Dignall, a 23-year-old
man from Liverpool. As a natural ethnographer, he was able
to lay out for me the sequence of events that often follow when
single men walk into a pub and ' t r y to cop off a few birds' , 1 3
 Speech Actions and Reactions in Personal Narrative / 237

A lot of fellas, if they're with a gang, they let their

birds sit with their
1
mates, while he stands at the bar with
his mates, talkin about things. And you could go up,
start chattin1 this bird up, an1 next thing--y'know, you're
none the wiser. An' she's edgin' yer on, on, you're a nice
fella, you've got a few bob. Great! And--you're chattin'
it up there, you're buyin' her a few shorts . . . Nex'
thing, eh, a fella comin1 there over there, 'Eh ay lads . . .
what are ya doin?' Well YOU don't know he's goin' with
her, so you tell HIM to push off. Nex' thing he's got his
friends--his mates on to you, an' uh . . . you're in lumber!
You've either got to run, or fight!
This, then, is the situation of mismatched information that
Shambaugh found himself in, though even in his account given
years after, he does not see it that way. He had evidently sat
down with his friends at a table where there was a girl who
had been with the Norwegian sailor. But to Shambaugh, the
Norwegian sailor had no standing at all. Shambaugh refused
him the right to enter into an argument. There was no verbal
sequence open to the Norwegian sailor: a violent reaction
followed.
Shambaugh's response can be characterized as an unaccounted
refusal. Goffman has outlined the sequences of actions in-
volved here as in Figure l.1Cf
Figure 1.

'COMPLIANCE •THANKS —•MINIMIZATION

7*
REQUEST

<
+ ACCOUNTING —•ACCEPTANCE
\

- ACCOUNTING—*• HUFF

This schema indicates that a request can be followed by com-

pliance, which is normally followed by expressions of thanks on
the part of the requester and an expression of minimization on
the part of the complier. A request can also be followed by a
refusal. This can then be followed by an accounting as to why
the request cannot be complied with (and can then be classed
as a way of putting off the request). There are then two fur-
ther routes. The requester can accept the refusal and the se-
quence comes to an end. Or he can by various mechanisms put
the request again, setting up a recursive sequence. If, on the
238 / William Labov

other hand, the refusal is not followed by an accounting, there

is no further verbal sequence. In polite society, the termi-
nation of verbal exchange can be called a 'huff. In many cir-
cumstances, this leads to a situation where individuals or fami-
lies do not talk to each other. In others, it leads to the kinds
of violent reactions we have been looking at.
We can then rewrite the event sequence of the Shambaugh
narrative as follows:
A(l) The Norwegian sailor makes a request for action of
Shambaugh: to move away from his woman.
A(2) Shambaugh refuses without an accounting.
A(3) The Norwegian sailor cuts Shambaugh's throat.
Shambaugh's refusal to recognize the right of the Norwegian
sailor to make the request led naturally to his failure to give
an accounting. He followed a sequence that did not leave
room for any further verbal exchange. The moral that he
drew for himself from the event is that the next time he shoves
someone, he will stand up and hit them. Others who know
these situations have drawn the opposite conclusion: that the
proper thing to do when someone says you are sitting with
their woman is to excuse yourself, get up, and move.
The Norwegian sailor cut Shambaugh's throat, but he missed
the jugular vein. One of Shambaugh's shipmates hit the Nor-
wegian on the head with a chair and he died. When Shambaugh
got out of the hospital, they gave him the knife; he still has it
with him in Cleveland.
The narrative of Schuster is also centered about a request
for action. In E(2), his brother started to talk about the rat,
and in E(3), Schuster tells him to 'cut it out'. This is re-
ported as a direct imperative. If it was heard as a valid re-
quest, then Schuster's brother must have believed that Schus-
ter believed that the four conditions for this rule held true:
< 1 > that there was a NEED: that he should stop talking about
the rat so as not to upset their mother; < 2 > that his brother
had the ABILITY to stop talking; < 3> that Schuster had the
RIGHT to tell him to stop; < 4 > that his brother had the obli-
gation to do what Schuster told him.
It should be underlined that this does not mean that Schus-
ter's brother himself believed that these conditions held, but
only that in order to hear the remark 'cut it out' as a valid
request for action (and not a joke or a suggestion), it is
essential that he believed that Schuster believed this. Once a
valid request is recognized, the consequences for further se-
quencing are well defined, as we have seen, and quite differ-
ent from the situation that prevails if the other person is seen
as joking or less than serious.
Given his recognition of the validity of the request, Schus-
ter's brother had a number of options for rejecting it. These
 Speech Actions and Reactions in Personal Narrative / 239

are sketched out in the following rule from Chapter 3 of Labov

and Fanshel (1977).
Rule for Putting Off Requests:
If A has made a valid request for an action X of B and B
addresses to A:
<a> a positive assertion or request for information about
the existential status of X,
<b> a request for information or negative assertion about
the time T, or
<c> a request for information or negative assertion about
any of the four conditions of the rule of requests,
then B is heard as refusing the request until the informa-
tion is supplied or the negative assertion is contradicted.
Schuster's brother's response is not given directly, and in
E(4), the refusal is indicated in parentheses. Schuster says
something that can be interpreted not as a report of an act
but as an excuse for it:
(2)12 'Course kids, you know, he don't hafta listen to me.
It is not uncommon in family narratives for someone to substi-
tute an excuse for an action ('Kids will be kids') for the re-
port of the action itself. I originally thought that this is what
Schuster had done here. But on further consideration of the
central problem of this narrative, that of accounting for the
violent reaction of his brother, I came to believe that line 12
contains an indirect quotation that can be reconstructed as
the direct quotation:
'I don't hafta listen to you.'
This would follow option <c> of the Rule for Putting Off Re-
quests, and in that option, selects the negative assertion about
the third condition: his obligation to do what Schuster says.
In denying that obligation, he refuses the request.
Labov and Fanshel point out that there is a great difference
in the interactive consequences of putting off or refusing a
request by reference to < 1 > needs or < 2 > abilities, on the one
hand, and < 3 > obligations and < 4 > rights, on the other.
References to needs and abilities are mitigating. If his brother
had denied the need to stop talking, saying 'Oh that won't
bother Mom' , we can imagine that Schuster's response would
have been quite different from what it was. But the act of
refusal by denying an obligation to listen produced a violent
response on Schuster's part.
240 / William Labov

The central actions in Schuster's narrative can then be sum-

marized:
A(l) Schuster makes a request of action of his brother:
to stop talking about the rat.
A(2) His brother refuses by denying his obligation to do
what Schuster said.
A(3) Schuster tries to enforce his request by physical
force: he twists his brother's arm.
A(4) His brother stabs Schuster.
So far, we have advanced in our understanding of the situ-
ation by seeing that his brother's refusal was made in an
aggravating form. This might explain for us the violence
that Schuster used. But it does not explain the terrible in-
crease in the level of violence on the part of his brother.
It is an important and yet curious fact that no one who
hears the story thinks of Schuster's brother as a bad person.
'Uncontrolled' is the most common adjective that is applied to
him. Some social or emotional force had driven him out of
control. We think immediately of the folk theory that great
grief often produces violent reactions: that we do not expect
someone who has suffered the loss of a father to behave
rationally.
It is evident that something was said that made Schuster
angry with his brother, and that his brother was even angrier
with him. What is the source of this violent anger? If we re-
flect on the conditions for the rule of requests, it clearly has
to do with the rights and obligations involved. If Schuster's
brother believed that Schuster believed he had the obligation to
obey him, and he rejected that claim, it is probable that he be-
lieved that his brother was assuming rights he did not have.
The inference that I have drawn here rests on my own personal
interpretation of the reported events, the result of many years
of familiarity with the narrative. In this interpretation, the
substance of what Schuster's brother said to him goes beyond
the indirect quotation of line 12:
'I don't have to listen to you. You can't take my father's
place and you never will.'
It seems to me that the violence of his reaction can only be
explained by his belief that Schuster was unjustly assuming
his father's place a few days after his father had died.
4. The defense of rights and the struggle for status. Re-
viewing the analysis of actions of the three narratives, two
common themes emerge. First, violent reactions are found when
the sequence of speech acts leads in a direction where speech
stops. Secondly, each of these situations is associated with a
dispute over the social status of the participants.
 Speech Actions and Reactions in Personal Narrative / 241

This consideration leads me to renew the emphasis put by

Labov and Fanshel on the role of social rights and obligations
in the dynamics of the Rule of Requests. As one might ex-
pect, the philosophical literature on speech acts deals with
needs and abilities and rarely touches on these dimensions of
social relations. In imagining conversations, one rarely
imagines the kinds of social situations that discourse is em-
bedded in. In the analysis of these three narratives, I have
tried to show, among other things, the importance of an under-
standing of that social context.
Table 1 is an analysis of the origins of violent conflicts in
narratives of personal experience.
Table 1. Origins of violent conflict given in 60 narratives
of personal experience.
Age of protagonist
Pre-
adolescent Adolescent Adult
Bothering 5 2
Play 8 4
Defense against aggression
of persons 4 1 1
of property 3 1 1
against unjust accusation 1
Status (gang, nongang) 4 1
Unknown 4 1
Women who were insulted or
who claimed to have been 3 2 2
Defense of rights
to a seat 1
to a cigarette 1
to the right of way 1
to walking space 2
to a piece of cake 1
to women 2 4
Total 24 23 13
These are not drawn from a random sample of a closed popula-
tion of narratives, but from 300 narratives that I have drawn
from a wide range of interviews for the study of narrative
structure. The causes of conflict are classified on the verti-
cal dimension, and the 60 narratives are broken down into
three columns by the age of the protagonist at the time of the
action.
Table 1 shows a striking differentiation by age. For pre-
adolescent narrators, there is a heavy concentration in narra-
tives where someone was 'bothering* another, where play led
to fighting and defense against physical aggression. Ado-
lescent protagonists show a much wider range of origins of
242 / William Labov

conflict, including those just mentioned, but with a heavier

concentration on gang status and a number of cases where the
origins were unknown. There are six cases of fights that
broke out over the defense of rights: the right to a certain
seat, to the loan of a cigarette, to the right of way on the
sidewalk, to being with a certain woman. Narratives of adult
conflicts show an even heavier concentration in this area: 7
of the 12 adult narratives of violent conflict deal with the de-
fense of rights, and 4 of these over the rights to women.
Looking more closely at this struggle over rights, it appears
that it is not the right itself that is at issue, but the general
status that the right pertains to. Following the general argu-
ment of Labov and Fanshel (1977), I would say that the funda-
mental cohesion of discourse is not at the level of speech act
but at the more abstract level of interaction where status and
role are negotiated. The arrays of actions outlined here are
therefore not the level of analysis that advances our under-
standing of the relation of violent reactions to speech acts, but
a more general characterization that Goffman has called MOVES.
In my own conception, a MOVE is an interaction that alters
or threatens to alter the relative social positions of the inter-
actants. These include challenges, defenses, retreats, counter-
challenges, supports, and reinforcements (Labov and Fanshel
1977). A CHALLENGE is an assertion which, if true, lowers
the esteem of someone in a certain status: asserting that some-
one is not a fit mother, or an effective employee, or a re-
sponsible partner in a game. Challenges may also be more
categorical, putting into question the right of a person to hold
a given status: of parent, adult, employee, or friend. SUP-
PORTS or SECOND DEFENSES do the contrary, raising the
esteem given to someone, reinforcing that person's right to
hold a given status. COUNTER-CHALLENGES and RETREATS
do more complicated work which ultimately covers the same
range of effects. REINFORCEMENTS preserve and maintain
the status quo.
Within this framework, the moves that led to violence in the
three narratives can be summarized in this way:
Moves of the Shambaugh narrative, Example (1):
M(l) The Norwegian sailor challenges Shambaugh's right
to sit with his woman.
M(2) Shambaugh makes a counter-challenge to the Nor-
wegian sailor's status as a person.
Moves of the Schuster narrative, Example (2):
M(l) Schuster challenges his brother's status as a
responsible member of the household.
M(2) His brother makes a counter-challenge to Schuster's
claim to take the role of his father.
 Speech Actions and Reactions in Personal Narrative / 243

Moves of the Williams narrative, Example (3):

M(l) Mrs. Hatfield reinforces the status of her son as a
dependent member of the household by sending him to
buy peaches.
M(2) Mrs. Hatfield challenges the status of her son as a
responsible adult by taking her money away.
M(3) Martin Cassidy challenges Billy Hatfield's status as a
responsible adult by rehearsing this fact in public.
Throughout this analysis, I have tried to keep in the fore-
front of attention the fact that we are dealing with reported
events, not the events themselves. The objective event se-
quence is a sequence of reported events; the arrays of actions
are reported actions; and the moves are reported moves. It
is therefore even more interesting to note that the moves ana-
lyzed here do not necessarily coincide with the presentation of
the self that emerges when we consider the evaluative material
and the subjective events added by the narrator. For the
three narratives studied here, the interpretation of events pre-
sented by the narrator, and the main point or focus of evalu-
ation are quite different from the series of moves that are de-
rived from this analysis. This observation suggests that we
may have bypassed some of the transformations of reality that
are the inevitable accompaniment of the narrative work.
The results of this further analysis are then three series of
moves that form the preconditions for violent reactions in the
narratives we have studied. We have seen that one route to
violence is a sequence of speech acts that comes to a termi-
nation: where there are no further verbal moves to be made.
One general principle that is well known to those who negoti-
ate violent and difficult situations is: keep talking. To do
this, negotiators have to have a good knowledge of the se-
quences that engage the other in verbal interaction, and the
ability to follow the patterns that lead to further talk rather
than those that lead to the termination of talk. The central
core of these techniques is the recognition that the other is a
responsible person in a structured social status. No matter
what violent or irrational behavior the other has shown, the
negotiator acts at all times as if the other knows the rights
and duties appropriate to that status. Once the recognition of
social status is withdrawn, the expectation of violent reactions
is greatly increased.
I am not suggesting that we can predict or control violent
reactions by this analysis. In no way would I claim to have
described the conditions that are sufficient or even necessary
for such violence. I have tried to reduce the strangeness and
inhuman quality of this violent behavior by showing its relation
to the general principles of social structure. I have also tried
to show that in the reports of violent conflict, we can hope to
find some comprehension of the conflict itself. By following
244 / William Labov

this line of analysis further, we may reach the point where our
analysis of discourse will be useful for those who have to deal
professionally with destructive and antisocial violence. At the
least, I hope to have brought Billy Hatfield, Jacob Schuster's
brother, and the Norwegian sailor within the range of our
human understanding, so that they no longer appear to us
as strange and terrifying creatures, but rather as people who
acted as we ourselves might act, if we too had been suddenly
deprived of our rightful place in the social world.
NOTES

1. The question which first raised the analytical questions

discussed here was posed by Michel Fournel, who raised the
issue of the relationship between the analysis of speech acts
in Labov and Fanshel (1977) and the ongoing analysis of narra-
tive that I have been developing for the past several years. I
am much indebted to him for that insight. I gratefully
acknowledge here my indebtedness to Erving Goffman, whose
contributions to my thinking are noted at several points in the
text. Teresa Labov has been an invaluable companion in these
explorations: her insights and corrections are to be found
throughout. I am particularly indebted to her for any concept
of social structure that I may have acquired over the years.
2. The philosophical and formal literature sometimes uses the
term SPEECH ACT in a more limited way, specifying a limited
number of performative categories or a subset of utterances
that appear to have a particular set of properties like requests
or promises. In the use of Labov and Fanshel (1977), all
utterances are analyzable as acts, and frequently as hier-
archically organized or parallel sets of acts. As will be evi-
dent, I use the term in that latter sense, referring to that
abstract level of analysis where a verbal action is categorized
as an action.
3. The question is often raised as to the relation of narra-
tives given in interviews to narratives told in unmonitored con-
versation. In an initial interview, the speaker has an ideal
audience: a listener who is interested in everything that is
said and who rarely interrupts. We therefore tend to get fully
formed narratives, without interruptions from others who may
have shared some of the same experience or have other points
of view. Some adjustments of reference are made: some peo-
ple are identified who would not have to be identified among
intimates, other identifications are left out because they are
irrelevant. But studies of unmonitored conversations show
many examples of such fully formed narratives, told without
interruption from beginning to end: narratives that can be
retold to general audiences with near-perfect understanding".
Narratives given in interviews are subsets of the whole set of
personal narratives. As Goffman has pointed out (personal
communication), narratives given in interviews tend to be about
 Speech Actions and Reactions in Personal Narrative / 245

those experiences that are best suited for interviews: events

of the most general interest that can be understood by anyone.
This subset is obviously well suited to the present inquiry into
the general conditions that relate speech acts and violent
actions.
4. The word shive [Jiva] is derived from the Hebrew word
for 'seven'. 'Sitting shive' refers to the Orthodox Jewish
mourning practice of commemorating a member of the nuclear
family for seven days. Family members stay at home without
doing any work and receive visits from friends and members
of the extended family.
5. See Schiffrin (1981) for the role of the historical present
as an organizing feature of narrative.
6. Many of these expressions take the form of feed-back or
back-channel signals when they do occur in response to narra-
tives. It is not their positive role in facilitating communication
that is at issue here, but simply their appropriateness as
opposed to expressions of surprise.
7. I was able to observe directly this kind of competition
among events one evening in West Philadelphia, when the nor-
mal conditions for reportability were shifted by a serious public
event. Black and white members of the local community had
gathered to observe a crisis in the relations of the police with
MOVE, a local group with its own life-style and ideology.
MOVE members were being pressed hard by police and other
authorities, and had appeared with submachine guns and rifles
on the front porch of their house. They were surrounded by
police cars and troops, and everyone expected that there would
be a bloody assault on the house at any moment. I was one of
four bystanders who were strangers to each other, all anxious
to serve as witnesses in case the police got out of hand. One
man started to tell a narrative of a run-in he had had with the
police after an accident. It was very long and repetitious, and
after five minutes one of the other men said in a very angry
tone of voice, 'Well what's the hang-up of the story? What's
the hang-up of the story?' The narrator said, 'I was just
getting to it', but the other had already turned his back and
Walked away.
8. Richard Bauman (1981) has called my attention to the
existence of such traditions, as in dog stories told in the
Southwest where the narrator is expected to lie. These stories
are part of a pattern of traditional storytelling which is quite
distinct from the kinds of personal narratives I am dealing with.
The narratives of personal experience discussed here are told
by ordinary people who have no reputation as storytellers, and
in many cases (e.g. Schuster) there is internal evidence that
the story has not been heard before by members of their own
family. Whereas the traditional storyteller takes ordinary events
and elaborates them to the status of reportable events, the
ordinary narrator begins with reportable events and tells them
in a simple and straightforward manner. The traditional story
246 / William Labov

is usually humorous; the ordinary story I am dealing with here

is serious. The traditional story is expected to be part fic-
tion; the ordinary story is expected to be true.
9. Note that Schuster uses two forms of embedding with zero
complementizer, (9) 'started talk' and (19) 'started bleed'.
Though not common, they are possible colloquial forms.
10. At first glance, it may seem that turning the head or
body is an objective physical act. But turning does not always
mean a 90 or 180 degree turn, and there is no minimal re-
orientation that corresponds to the expression, 'So I turned to
him and said . . . ' In many narratives, such expressions amplify
the activity attributed to the narrator without risk of contra-
diction .
11. In many ways, the Labov and Fanshel rules are parallel
to the rules formulated in Gordon and Lakoff (1975). The most
important difference is the absence of any reference to right
and obligation in the latter .
12. The legal metaphor used here to distinguish objective
from subjective events reminds us that the interpretation of the
acts involved here is indeed problematic: a violent physical
action can be represented legally as an assault, an attempted
murder, manslaughter, premeditated murder, and so on.
13. 'Bird' is the regular Liverpool term for 'young woman';
'mates' are close friends of the same sex; to 'cop off a few
birds' is to meet up with and spend the evening with; to be
'in lumber' is to be in serious trouble.
14. The schema presented here is my adaptation of one pre-
sented to a class by Goffman.
REFERENCES

Bauman, Richard. 1981. 'Any man who keeps more'n one

hound'll lie to you': Dog trading and storytelling at Canton,
Texas. In: 'And other neighborly names': Social process
and cultural image in Texas folklore. Edited by Richard Bau-
man and Roger D. Abrahams. Austin: University of Texas
Press. 79-103.
Goffman, Erving. 1976. Replies and responses. Language in
Society 5.257-313. Reprinted in: Forms of talk. Phila-
delphia: University of Pennsylvania Press, 1981.
Goffman, Erving. 1974. Frame analysis. New York: Harper
and Row.
Gordon, David, and George Lakoff. 1975. Conversational
postulates. In: Syntax and semantics, Vol. 3: Speech
acts. Edited by Peter Cole and Jerry Morgan. New York:
Academic Press. 83-106.
Labov, William. 1981. Field methods of the project on linguis-
tic change and variation. Working Papers in Sociolinguistics,
No. 81. Austin, Texas: Southwest Educational Development
Laboratory.
 Speech Actions and Reactions in Personal Narrative / 247

Labov, William. 1972. Sociolinguistic patterns. Philadelphia:

University of Pennsylvania Press.
Labov, William, and David Fanshel. 1977. Therapeutic dis-
course. New York: Academic Press.
Labov, William, and Joshua Waletzky. 1967. Narrative analy-
sis: Oral versions of personal experience. In: Essays on
the verbal and visual arts. Edited by June Helm. Seattle:
University of Washington Press. 12-44.
Schiffrin, Deborah. 1981. Tense variation in narrative. Lg.
57.1:45-62.
Searle, John R. 1969. Speech acts. London and New York:
Cambridge University Press.
 IDEAL READERS AND REAL READERS
Charles J. Fillmore
University of California, Berkeley

1. For the past year and a half I have been working on a

research project which investigates the ways in which school
children interact with standardized tests of reading comprehen-
sion. The children in the study are skilled and medium-skilled
readers chosen from third and fifth grade classes in two schools
in Berkeley and Oakland, California. The tests we have been
examining are selected from those currently given in American
schools to children at our subjects' grade level. 2
An example of the kind of material the team is working with
is the following passage, taken from the Metropolitan Achieve-
ment Tests3 and intended to be administered to third-grade
students.
The carpenter was astonished that such a weird, weak-
looking creature as Nasrudin was applying for a job.
'Okay, I'll give you a chance,' said the doubtful carpen-
ter finally. 'Take this ax and chop as much lumber as
you can.' At dusk Nasrudin returned.
'How many trees have you felled?' questioned the car-
penter.
'All the timber in the forest,' Nasrudin replied.
Shocked, the carpenter glanced out his window. There
were no trees standing on the hillside. Nasrudin had
destroyed the entire forest. 'Where did you learn to chop
lumber?' asked the astonished carpenter.
'In the Sahara Desert,' answered Nasrudin.
'That's ridiculous!' shrieked the carpenter. 'There aren't
any trees in the desert!'
'There aren't any, NOW,' said Nasrudin calmly.
The passage is followed, of course, by a series of test ques-
tions. Item (1) gives the first question.

248
 Ideal Readers and Real Readers / 249

(1) The carpenter told Nasrudin to

(a) look for another job
(b) cut down as many trees as he could
(c) go back to the Sahara Desert
(d) plant as many seeds as he could
The children confronting this question, presented as it is
immediately after the reading passage, are expected to under-
stand that they are not here being asked to continue the narra-
tive. That is, they have to sense that the story they have
just read has been finished, as far as they are concerned, and
that they are now being asked to show how well they under-
stood it. Should they by mistake construe their task as one
of advancing the narrative beyond the point where Nasrudin
made the boastful claim about creating the Sahara Desert, they
might find it quite reasonable that the carpenter should advise
Nasrudin to look for another job (since he was no longer needed
here), tell him to go back to the Sahara Desert (as a kind of
'get-out-of-my-life' remark), or indeed, order him to plant as
many seeds as he could (to make sure that something could get
growing on the bared hillsides). The test-takers must first
keep in mind the test-taking maxim, that if two answers appear
to be equally good, both are probably wrong, but they must
then realize that they are probably not being asked to advance
the narrative. What they must remember is that in Nasrudin's
probationary period, he had been given the ax and told to
'chop as much lumber' as he could. They had to figure out that
'chop lumber' is our author's unusual way of saying 'cut down
trees', and they must sense that it was the early conversation
between the carpenter and Nasrudin with which the question is
concerned.
The second question is shown in (2).
(2) How long did it take Nasrudin to complete the job?
(a) one day
(b) three days
(c) thirty days
(d) three years
In order to answer this question correctly, the children are
required to realize that in the sentence 'At dusk, Nasrudin
returned', the phrase 'at dusk' refers to the dusk of that same
day, and they must also realize that there is nothing in the
story that could back up any answer with the number 'three'
or 'thirty' in it. Those children who are uncommonly sensitive
to language will wonder what it might mean to 'complete the
job' under these circumstances, since the only task Nasrudin
had been given was to 'chop as much lumber' as he could.
The third question is stated in (3).
250 / Charles J . Fillmore

(3) Nasrudin suggested that there were no trees in the

Sahara Desert because
(a) trees can't grow in the desert
(b) no one had ever planted any there
(c) they had been destroyed by fire
(d) he had chopped them all down
The answer is that Nasrudin had chopped them all down.
This was 'suggested', to use the question's word, by Nasrudin's
answer, 'There aren't any, NOW, said after Nasrudin had ex-
plained that the Sahara Desert was the place where he had
'learned' to 'chop lumber'.
The fourth question is given in (4).
(4) After Nasrudin finished work, he
(a) left for the Sahara Desert
(b) told the carpenter what he had done
(c) applied for a new job
(d) yelled at the carpenter
The expected answer is that Nasrudin told the carpenter
what he had done. The ordinary scene a reader might con-
struct based on that description, however, is probably a bit
different from what we saw in the story, so a certain amount
of construing is necessary. The carpenter, it will be recalled,
asked Nasrudin, 'How many trees have you felled?', to which
Nasrudin replied, 'All the timber in the forest'. This utterance,
an elliptical answer to a question which presupposed an under-
standing of what he had done and speaks only to the question
of how much he had done, has to be construed as an instance
of Nasrudin telling the carpenter what he had done. The
tempting possibility that the correct answer is 'yelled at the
carpenter' is presumably introduced to take advantage of the
printer's decision to put the word 'now' in capital letters, in
the sentence, 'There aren't any, NOW. The capitalized word
suggests shouting, so what is being tested with this foil is the
child's ability to notice that what followed 'There aren't any,
NOW in the text is 'said Nasrudin calmly'.
The fifth question appears in (5).
(5) The carpenter had not expected that Nasrudin
(a) had ever seen the Sahara Desert
(b) really needed a job
(c) would be so rude
(d) could do the job so quickly
In this item the pluperfect form, 'had not expected', plays
an important role. The sentence has to be situated in the text
at some time point where it serves an explaining role. The
text reveals that something was unexpected in the place where
 Ideal Readers and Real Readers / 251

it shows the carpenter surprised. Being 'shocked' is an ex-

treme form of being surprised. The text describes the carpen-
ter as 'shocked', through a grammatical device whose function
may not be transparent to most third grade readers, and backs
this up by showing that the carpenter spoke from then on only
in sentences ending in exclamation points. He had been shocked
when he learned--and hence he had not expected that it had
been possible--that at the end of the first day on the job Nas-
rudin had conquered the entire forest.
2. It is of interest to our group how young readers con-
struct an understanding of reading passages of the kind we
have just seen, and how well that understanding can be ap-
pealed to in finding 'best answers' to test questions about
those passages.
There are two intended goals in our research, and a third
that we have taken on against our will. Our first goal is to
analyze reading-test items, both the passages and the test
questions, in such a way as to be able to isolate and describe
the kinds of background knowledge and the kinds of interpret-
ing and integrating skills which a reader must bring to the
passages in order to get out of them what their creators in-
tended. The second goal is to find out, by interviewing our
young subjects, whether they have that knowledge and those
skills. We begin by analyzing tests and devising a system of
annotations for them which can serve three purposes: it repre-
sents our view of the comprehension process of someone who
understands the passage with no difficulty, it provides the
material for our choice of the interview probes, and it gives us
a framework or checklist against which we can evaluate the
children's performance with the texts. The second part of our
work involves close observation of the children's experiences
with the texts, with free retellings, interviews, and metacogni-
tive probes, looking for the presence or absence of the kinds
of knowledge and skills which seem to us to be necessary for
understanding them.
The third and unwelcome task which fell to us is due to the
nature of the corpus we had chosen. Oddly enough, in view of
the time and expense that goes into the construction of reading
test items, we are dealing in this research with seriously flawed
texts, texts which frequently require of their readers an un-
common degree of tolerance and cooperation. The testing in-
dustry, we have come to realize, has created a new genre for
English written language, a genre whose characteristics are
determined by very unnatural requirements of lexical choice,
grammatical structuring, and synonym alterations, these dic-
tated, I presume, by the intention to test knowledge of par-
ticular vocabulary items, the need to produce something which
fits accepted readability formulas, which avoids gender or
ethnic stereotyping, and which satisfies copyright laws with
respect to the percentage of material that needs to be modified
252 / Charles J . Fillmore

in the case of passages taken from the trade literature. This

third goal of ours, then, is to understand some of the conse-
quences of the development of this special genre, to show why
it introduces complications that interfere with the ability of
these tests to measure what they are designed to measure, and
to see how well young children are able to master this particu-
lar genre through their repeated experiences with it.1*
3. We need, in our work, to be able to compare real read-
ers in their experience of these texts with the kind of reader
who gets out of them everything that is needed. For this
comparison we had to develop a particular kind of abstraction,
something we call the 'Ideal Reader'. It is this invented Ideal
Reader who sees the connections, creates the expectations,
performs the inferences, and asks the questions which our
annotation is designed to represent. It is with this Ideal
Reader that our flesh-and-blood readers are to be compared.
My purpose in this paper is to explain the nature of this
idealization and to show how we think we can decide on its
characteristics.
Our notion of the Ideal reader is localized to given texts and
to given interpretations of those texts. That is, we do not
speak of 'an ideal reader' in the abstract, but of 'the Ideal
Reader' of a given text on a given interpretation. We see be-
fore us a written text; we determine what we take to be a
'correct' or somehow necessary interpretation of it; and we
then project, from the text and the interpretation, to an in-
vented Ideal Reader: that individual who has exactly what it
takes to get from the text to the interpretation via the usual
principles of compositional semantics, schema building, inferenc-
ing, goal and plan detection, and so on. (I should point out
that the notion is a technical one, designating an abstraction
that is of limited utility. It is not necessarily a desirable
thing to be an Ideal Reader in our sense.) 5
Once we have decided on the characteristics of an Ideal
Reader (for a given text on a given interpretation), we can
then ask certain important empirical questions about the text
and its living readers. If, for example, we find ourselves
convinced that there are in the world no real readers who
match certain essential characteristics of our invented Ideal
Reader, then we can believe that we have on our hands a use-
less text. It is not written for anybody. If we find among
our young subjects that the only real readers who match the
conditions we have specified for the Ideal Reader are those
children who have had experiences not shared by the others,
especially if those experiences are distributed among the chil-
dren in patterns that follow social, economic, or ethnic differ-
ences, then we have learned something about the fairness of
the item or the representativeness of the test results as these
are affected by the item. With items that we believe are
reasonable and fair, we can examine in detail the nature of
 Ideal Readers and Real Readers / 253

the gaps between particular real readers and our projected

Ideal Reader, and ask how such gaps can be filled. Discover-
ing such gaps is a matter of reading failure diagnosis; insti-
tuting measures for bridging the gaps by providing children
with new experiences, facts, or skills, or by helping them be-
come aware of the gaps, is a matter of improving children's
reading abilities.
4. Simplifying a little, the Ideal Reader is someone who
knows, at each point in a text, everything that the text pre-
supposes at that point, and who does not know, but is pre-
pared to receive and understand, what the text introduces at
that point. Real readers, then, can differ from the text's
Ideal Reader in two directions. With respect to any given
point in the text, they may be underqualified, in that they do
not know what the text assumes they know at that point, or
they may be overqualified, in that they already know what the
text introduces. 6
(A passage in a linguistic textbook which uses the term 'con-
trast' without explanation but which defines, explains, and
exemplifies the notion 'neutralization', is addressed to a reader
who already understands 'contrast' in its technical linguistic
sense, but who does not yet know how linguists use the term
'neutralization'. The Ideal Reader for this text, then, is in
part described as someone who knows, at this point, about
contrast, but who does not know about neutralization. In this
respect, at least, such a text appears to be perfectly reason-
able. There are, we can be sure, any number of readers who
meet perfectly this pair of conditions. A text in which the
opposite choice had been made, however—one which explained
'contrast' but presupposed knowledge of 'neutralization'--would
be a pedagogically defective text, since only readers with very
unusual educational histories would be likely to satisfy just
those conditions.)
5. Linguists will have noticed that I have been talking about
readers and written texts, whereas it is much more common in
linguistic circles to talk about hearers and spoken language.
Our idealization is possible for written language more than for
spoken language because of the fact that written texts are more
characteristically monologic and closed: in a word, they tend
to be 'composed'. By contrast, we find in conversational lan-
guage situations in which the interlocutors need to negotiate a
common background as the conversation progresses, we find
texts in which the linguistic part of an interaction is insufficient
for constructing anything that could be called the intended
interpretation, and we find, in dialogue between people com-
municating at cross purposes, texts that lack any discernible
point or structure.
254 / Charles J . Fillmore

6. While I have said that the Ideal Reader is defined with

respect to a text that is composed--that is, to a text that is
fixed and in some sense complete--I do not intend to imply that
composed written texts can be interpreted statically. The inter-
preter's experience always has a clear dynamic aspect, to which
our work has to pay close attention. We need to show, for
example, that a text can create expectations in the reader's
mind at one point which it then satisfies or subverts at a later
point. The recognition of structure, development, or point; of
suspense, surprise, or closure; of the interruption and resump-
tion of a 'thread 1 , etc. , all make up part of a reader's experi-
ence with a written text.
It i s , in fact, just this dynamic aspect of the reading experi-
ence which is the most important part of our analysis, and the
most difficult part of our interview process with the children.
What I mean by that is that in the interview process, we have
had to be particularly sensitive to the difference between ask-
ing questions which will reveal the dynamics of the text-
understanding process as it occurs naturally, and questions
which themselves guide and advance that process.
We are pleased if we find a child, who, on reading the
seventh sentence in a text, spontaneously says, 'Oh, now I
see what's going on!', because we have just had revealed to
us the workings of the comprehension process in that child.
We have learned not to be pleased when a child says something
like that in response to one of our interviewers' questions.
Our job is to track the child's coherence-creating process, not
to shape it for him.
In order to present the dynamic aspect of a reader's experi-
ence with a text, we have developed a method of text analysis
which takes the text one segment at a time, asking ourselves
at each point in this unrolling of the text something like, 'Hav-
ing read this far, what would it have figured out, or be
puzzled by, or be expecting?' In order to make our analytic
task reasonably finite, we have found it useful to keep the
analyst's point of view distinct from that of both the Ideal
Reader and the real readers. In particular, we, the analysts,
have the advantage of knowing the whole text, knowing the
point of the text, knowing how everything comes out in the
end. We might know that the point of a particular passage is
that it introduces a surprise. For example, if we know that
the point of a particular line in a story is to reveal, for the
first time, that the characters we have been reading about live
in a tree, and if we feel sure that the reader is expected at
this point to be surprised, that assumption imposes specific
requirements on our annotation of earlier parts of the text.
We have to make the Ideal Reader assume, throughout some
earlier part of the text, that the people in the story do not
live in trees, and at the same time we have to make sure that
this assumption is known not to be explicitly bound to the
actual material of the text. That is, we need an annotation
 Ideal Readers and Real Readers / 255

which will show, at the point when the surprising information

is introduced, that the reader is supposed to feel surprised
but not cheated. If we ourselves, in preparing the annotation
for the text, had to examine it only one segment at a time,
ignorant at each point of what was going to come up next, we
would have no basis for including or excluding any part of
the probably boundless number of things that ordinary readers
would assume in the world of the unrolling text. That is,
knowing the nature of the surprise that is coming up gives us
a reason to ignore many aspects of the normal reader's en-
visionment of what's going on. 7
7. The analyst sees the text as a whole. The Ideal Reader
and the real reader see it one segment at a time. It is neces-
sary for u s , then, if we wish to explore the real readers' ex-
perience of the text, to present it to them one segment at a
time. The general method we use works like this: we show
the readers the first segment, then ask our first batch of
questions; then we show them the second segment, asking them
our next batch of questions; and so on. At each point, we
make sure that the previously exposed part of the text is still
available for scanning, rereading, consultation, etc.
The mechanics of this kind of presenting and interviewing
proved to be interestingly difficult. Since some of these diffi-
culties shed light on the process of text understanding, I be-
lieve they are worth discussing here.
At first our method was to decide on a particular segmentation
of a text, type it out with each segment on a separate line, and
have the subjects slide a piece of cardboard down over the text
one line at a time during the interviewing process. This method
introduced two major difficulties. First, as the card got lower
on the page, the exposed piece of text at the top ended up
looking more like verse than prose, and that, we felt, could af-
fect some people's interpretations of what they were reading.
Second, the method did not make it obvious to the subject how
much of the text was left, and we felt that subjects might use
different strategies for interpreting a sentence if they thought
it was the closing sentence in a text than if they thought more
was coming. This uncertainty made the experience quite unlike
normal reading; in normal reading, we almost always know how
close we are to the end, and that knowledge plays a large part
in shaping our expectations and putting our interpretative
faculties to work.
Our current method is more expensive and troublesome, but
it has eliminated both of the difficulties presented by the slid-
ing card method. We type the text on a sheet of paper, in
the normal way, double-spaced. If the text has been given n
segmentations, we make n+1 xerographed copies of it, and con-
struct a booklet. Page 1 of the booklet has the entire text
blocked out with a marking pen. Page 2 has everything
blocked out but the first segment. Page 3 has everything
256 / Charles J . Fillmore

blocked out but the first two segments. And so on. (Having
found that some children have trouble finding quickly the
place where they left off, we have begun using a small red
dot at the beginning of the increment to help them.) With
this method, then, as the text gets exposed, it looks like
ordinary printed prose, and it is very clear to the subject
how much of the passage is left before the end is reached.
(The passages we use all fit on one page.)
The interview works something like this. The child who
turns the first page sees
Once upon a time xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
etc.
exposed at the top of the second page. Our interviewer asks
something like, 'What can you tell me about this passage so
far?' The subject answers, 'Well, it's going to be a story.
Most likely a fairy tale.' 'How do you know that?' 'You only
say "Once upon a time" when you're telling a fairy tale.' 'Do
you have any guesses about what we're going to read when we
get to see more of the story?' 'Well, maybe something about a
poor old lady who lived in the woods, or maybe a rich king
with a beautiful daughter. I don't know.'
The child then turns the page to.expose the next increment,
and sees
Once upon a time there was a rich king xxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
etc.
The interviewer says, 'Say, you were right, weren't you?
The sentence isn't finished yet, is it? Do you want to stick
to your guess about the beautiful daughter?' 'Yeah.' The
child turns the page and sees
Once upon a time there was a rich king who had three
sons. xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
etc.
And so on through the text.
The new method of presentation solved some of our problems
in exploring the reading experiences of real readers, but there
remain a number of very serious problems connected with the
piecemeal presentation, no matter how it is adapted. One diffi-
culty with our method comes from the fact that the talk pro-
duced by the interviewer is itself a text which the children
have the right to believe has a point and a direction. If, in
connection with the text about the people who lived in trees,
we were to ask a question like, 'Do you think these people live
in houses like everyone else?', in order to be able to predict
 Ideal Readers and Real Readers / 257

whether our young subjects were going to be surprised when

they found out the truth, the children would quickly catch on
to our purposes. If we tried to balance this by asking dozens
of questions about normal readers' default assumptions ('Do you
think they grow hair on their teeth?', 'Do you think they sleep
at night?'), our questions would be intolerably distracting.
A second problem produced by the interviewer's own text
shows up when we are exploring the reasons for the choice of
particular test question answers. With some young children,
the question, 'Why did you choose this answer?' suggests
powerfully that the answer was wrong.
One source of difficulty, then, resides in the character of
the conversation with the interviewer. Another is the pacing.
A paragraph that takes less than a minute to read in normal
circumstances can take half an hour to go through in the
segment-by-segment presentation. The interview brings so
many things into the subjects' consciousness that the simple
thread that is our text can get completely lost. Using the
method with adult readers on adult texts, we learned that
passages that are humorous when read all at once are not
humorous when given out piece by piece. I am not merely
saying that the passages do not seem funny; sometimes their
humorous intent is not even discerned. Furthermore, if a text
takes a digression and then returns to the main theme, the re-
turn to the theme can seem very striking. In natural fast
reading, by contrast, the digression itself would hardly be
noticed.
With young readers, the factor of interest works against us
both ways. If the story is interesting, the reader wants to
get to the end to see how things turn out, and becomes im-
patient with all these boring questions. If the story is un-
interesting, the reader wants to get the whole experience over
with as soon as possible, and becomes impatient with all these
boring questions. A lot depends, in short, on the warmth and
charm of the interviewers, and on whatever rewards children
might feel in knowing that an adult is paying very close atten-
tion to their words and thoughts.
The upshot is that while we are leading children through our
kind of micro-analysis of a text, we may indeed be learning a
lot about whether they are prepared to understand the text we
are examining--which is, after all, our purpose—but the method
itself might be preventing them from actually understanding it.
A control is obviously needed, and for our control we have a
second group of children who are exposed to the text in a
different way. These children read the passages, answer the
test questions, retell what they have read in their own words,
and only then submit to the piecemeal presentation. The ques-
tions for this group are somewhat differently formulated: 'What
were you thinking when you read this?', 'Do you remember what
the next sentence is?', and so on.
258 / Charles J . Fillmore

8. As we see it, the main dynamic aspect of the reading

experience is that of constructing and revising an Envisionment
of the 'world of the text', some coherent 'image' or understand-
ing of the states of affairs that exist in the set of possible
worlds compatible with the language of the text. We do not
intend the word Envisionment, which we have borrowed from
John Seely Brown (personal communication), to suggest too
strongly the visual aspects of a text world, but at the same
time, we recognize that in the kinds of texts we are dealing
with, the visual aspects do, in fact, predominate.
We have found it useful to distinguish various confidence
levels of material in the Envisionment, according to whether
such material is explicitly justified by the linguistic material of
the text, whether it came into being by inferences which the
text is seen as clearly inviting, or whether they represent
interpretations which result from schematizations brought to
the text to situate its events in common experience, but which
do not follow necessarily from anything the text has provided.
In our discussions of real readers, we need still another level
of Envisionment: ways in which the world of the text has
been shaped by the idiosyncratic experiences and imaginings
of individual readers.
To illustrate the theory of levels of Envisionment, we can
consider the following three-line text, borrowed from Marcelo
Dascal (personal communication).
The princess ate some jam.
The queen slapped her.
The princess cried.
At what we call the E° level of Envisionment, there are only
those states of affairs that have to hold in the world of the
text for the individual sentences to be separately true. In our
case, the Envisionment at E^ has three disjoint parts: some-
body who is a princess eats some jam; somebody who is a queen
slaps a female being; and somebody who is a princess cries.
At the second level in the Envisionment, which we call El, it
is assumed that we are dealing with a cohesive text, rather
than with three independent statements, and so the princess
in Sentence 1 is the same as the princess in Sentence 3, and is
the her, the one who is slapped, in Sentence 2. The queen in
Sentence 2 is furthermore taken to be the princess's mother.
We have here the workings of a kind of Parsimony Principle
in text comprehension. If, in constructing an understanding
of the first sentence, we had to instantiate a Royal Family
schema, we find it parsimonious to use that same schema in-
stantiation for identifying the queen, too. Hence the queen
and the princess are mother and daughter. If at the beginning
of the second sentence ('The queen slapped her.') we have two
people 'on stage', the princess and her mother the queen, then
the her of this sentence must refer to the princess. The
 Ideal Readers and Real Readers / 259

Parsimony Principle is a text-interpretation maxim that says

something like: Don't bring more people or props into the text
world than are needed to make the text cohere.
In the example just given, the difference between EO and El
amounted to a difference between individual sentences and sen-
tence sequences. It was when we put the sentences together
that we began needing interpretations at the El level. The
difference between E^ and El can also be illustrated with single
sentences.
The favored kinds of sentences used in psycholinguistic ex-
periments typically have only an EO interpretation. In a sen-
tence like 'The actress sold a fish to the carpenter', the kind
of sentence you might be expected to learn or shadow or associ-
ate something with in a psycholinguistic experiment, we find
that there is nothing we know about either actresses or car-
penters that we can call on to motivate the scene of the one
selling a fish to the other. If we compare that with a sen-
tence like 'The cobbler sold a pair of boots to the mountain
climber', we find that the latter invites more interpretative
work. The Parsimony Principle would induce us to make the
hypotheses that the cobbler sold boots that he had made in his
shop, and that the mountain climber was buying boots that he
would use in mountain climbing. That i s , we tend to assume
that it is as a cobbler and as a mountain climber that the two
participants were engaged in this particular commercial event.
These assumptions are made at the El level, rather than at
the EO level, because we could, without contradiction, immedi-
ately find out that these hypotheses were completely wrong.
It could turn out that the cobbler was sitting in for his wife,
a shoe store clerk, while she was at the dentist's, and that the
person described here as the mountain climber was buying the
boots to give to his mother for Mother's Day.
The sort of 'default' interpretations that we assign to the
Envisionment at the El level remain tentative for a very brief
time. Knowing that we are dealing with a composed text, we
somehow believe that if we were not meant to give this inter-
pretation, the author owes us an explanation very soon. If
the text does not very quickly turn us away from the El
interpretation, the normal reader quickly converts it to an EO
confidence level. 8
At the third level of the Envisionment, which we call E2, we
'situate' elements of the passage in terms of our knowledge of
goals and institutions and folk theories of human nature. With
the story of the princess and the jam, we make sense of what
is going on by assuming that the queen's act of slapping the
princess was in punishment for the princess's having eaten
the jam, and we assume that the princess's tears are in re-
sponse to the queen's slap,3 showing pain, remorse, or shame.
The next higher level, E , contains particular real readers'
embellishments of the text world, the filling in of details not
motivated by the text itself. For our story we might find, in
260 / Charles J . Fillmore

particular readers' experiences of it, the assumption that the

queen had wanted the jam for herself and was very selfish, or
maybe that the princess had recently been made to promise to
stay on a diet until she found a husband. In our annotation
for the Ideal Reader, we obviously do not need such a level;
but in our representations of the subjects' interpretations of
our texts, we definitely do.
My co-workers and I have no stake in the proposed number
of levels, and we will not be surprised to find out that the
phenomena we are concerned with do not at all lend themselves
to a description in terms of discrete ranked tentativity levels.
What we are sure of--and this is the main practical purpose of
the levels theory--is that a child who is not able to do the kind
of inferencing or 'reading between the lines' that is necessary
for filling out material at what we are calling El and E2 is not
a good reader, nor is the child who fills in lots of E3 material
and immediately assumes that it was in the text. A child who
reads very slowly, figures out the meanings of the sentences
in the text one at a time but never puts them together into a
cohesive whole, is a poor reader in the first sense. Such a
reader cannot go beyond E0. Poor in a different way would
be the reader who, in reading our story about the princess
with the sweet tooth, would unhesitatingly answer 'Strawberry'
to the question, 'What kind of jam did the princess eat?1 Such
readers overtrust their own E3 embellishments of a text.
9. The lexical and grammatical material of the unfolding text
creates or identifies for the reader the conceptual tools needed
for constructing the Envisionment. The main kind of tool I
have in mind is the construct (or set of constructs) variously
known as 'scripts', 'frames', 'schemata', 'folk theories', 'cog-
nitive models', and the like. These constructs are intended
to represent the knowledge structures with which our experi-
ences with the world are held together. Sometimes in a given
portion of the text such tools are merely activated, merely
made available for later use. At other times they are actively
used in constructing and maintaining the Envisionment. In our
annotation we make a great deal of use of labeled schemata,
with which we appeal, informally, to structured knowledge that
the Ideal Reader is believed to have and which real readers
may or may not have, may or may not be able to use, or may
or may not know when to use.
Two applications of such schemata have already been illus-
trated: the Royal Family schema, which allowed us to see in-
stantly how the princess and the queen were related to each
other, and the schemata we used in bringing coherence to the
sentence about the cobbler and the mountain climber. A
slightly more complicated one-sentence example is: The de-
fendant had forged the will. Somebody called the defendant is
seen as a person who plays a particular role in a Criminal
Trial schema, that of the Accused; the object called the will in
 Ideal Readers and Real Readers / 261

our example is the Legal Instrument in an Estate Inheritance

schema; and the act of forging is the central act in a Forgery
schema. The character of each of these schemata could be
spelled out in considerable detail. The Parsimony Principle
would invite us, at perhaps an E2 level (the level at which we
seek to bring in outside explanations) to imagine that the act
of committing the forgery was one of the acts for which the
defendant was currently involved in the judicial process, and
to imagine that the will that got forged had the defendant
named as a beneficiary.
The schemata just mentioned have all been brought into the
Envisionment by virtue of particular lexical items that are keyed
to particular roles or steps in their associated schemata. In
the comprehension process, the interpreter activates the schema
connected with such lexical items and works at building an En-
visionment out of knowledge derived from these schemata. This
is done partly by combining primary schemata into larger as-
semblies or networks, sometimes, as we have seen, by bringing
in schemata not explicitly indicated by material in the text but
needed for holding the other schemata together.
There are various ways in which schemata can be linked to-
gether. Some are linked in semantic memory by what can be
called Knowledge Links (K-links), connections between schemata
provided by general knowledge, independently of any informa-
tion provided by the actual present text. For our forgery sen-
tence, K-links bring to play our knowledge that all of the
schemata lexically introduced by that sentence somehow fit into
a larger schema that could be called Judicial Process, and make
us aware that a defendant in the criminal trial has been accused
of a crime, and that forgery is a crime.
The text itself, which tells us about some 'world', links to-
gether instantiations of schemata by anchoring them to each
other in that world. These can be called Text Links (T-links).
An E^ T-link in our forgery sentence establishes, through the
grammar of the sentence, that the person who is the defendant
in a criminal trial is also the perpetrator in an act of forgery,
and that the product of that act of forgery was (or has been
put forth as being) a legal instrument in a matter of estate
inheritance.
The schemata that we have been considering so far are
schemata that operate in what with many others we have come
to call the Content domain. We distinguish three domains for
the Ideal Reader's speculations, puzzlements, and conclusions:
the domain of 'Content', by which we have in mind the proper-
ties and events in the world of the text; the domain of 'Text',
with its schemata of grammatical structure and text structure;
and the domain of 'Genre', where we have in mind those struc-
tures of expectation that come with knowing that one is dealing
with a folk tale, a detective story, an obituary, a reading test,
or the like.
262 / Charles J . Fillmore

Within the Text domain, our forgery sentence has, in addi-

tion to the grammatical structurings which aided us in provid-
ing the T-links, a number of features which mark it as 'text-
internal'. We recognize that a sentence like The defendant
had forged the will is not likely to be the first, and certainly
not likely to be the only sentence in a text. This judgment
we base on the use of the definite article (the defendant, the
will) and the use of the pluperfect (had forged). The text-
schematizing necessarily associated with them shows that there
must be a presupposed temporal reference point in the part of
the larger text where this sentence can occur (the point on
which the pluperfect is semantically anchored) and a pre-
established setting within which the descriptions the defendant
and the will are uniquely identifying.
Genre schemata arise from structured expectations created by
familiarity with particular genres. If we read in a folk tale
that the king's two older sons have both failed to slay the
dragon, we are filled with hope when we learn that the king's
third and youngest son has set out to try his hand. Were the
story to end with the third son being slain by the dragon, we
would feel that we had just been exposed to a new and cynical
derivative genre, not that we were wrong in forming the expec-
tations we had formed. When we read questions in a reading
test, we know that we are not being asked to figure out a
clever way to finish the passage we have just been reading,
but rather that we're being asked what we remember, or what
we can now figure out, about what the passage told us.
10. There is one more device that we need for our complete
annotation, and that is a representation of Point of View. It
has played a relatively small role in the work we have done so
far, so I will say little about it. For some texts, especially
texts involving histories, descriptions of objects or terrains,
and narratives, the text world not only has the properties it
has in some objectively describable way, but its properties are
presented to us from a particular point of view: there is an
observer, an experiencer, or a camera's eye, and perhaps al-
ways a temporal reference point.
I illustrate the notion merely by showing that some sequences
of sentences in a text can be seen to cohere only when they
are taken as sharing a single point of view. The sequence
He was coming up the steps. There was a broad smile
on his face.
is coherent in the sense I have in mind, since the position from
which his ascending the stairs can be described as 'coming' is
also the position from which it would be natural to see his smile.
Analogously, a sequence like
 Ideal Readers and Real Readers / 263

He was going up the steps. There was a wad of bubble-

gum on the seat on his pants.
is also coherent, given a constant viewpoint. By contrast, a
sequence like
??He was coming up the steps. There was a wad of
bubble-gum on the seat on his pants.
seem quite bizarre if taken as a visual description, and the
sequence
He was going up the steps. There was a broad smile on
his face.
seems to require us to imagine a side view. The recognition
of Point of View figures in any complete account of lexical se-
mantics (with respect, for example, to such words as come and
go), the semantic role of grammatical categories (involving, for
example, tense and definiteness), and the cooccurrence or se-
quencing of sentences in a text.
11. The descriptive framework we have ended up with in-
cludes : a streamlined record of the grammatical parsing of the
sentence; the schemata of Content, Text, and Genre which are
introduced by lexical and grammatical form, in the first in-
stance, and from the interpreters repertory of schematizing
devices, in the second instance; the questions and expectations
raised in the reader's mind concerning the still unrevealed por-
tions of the text; the growing and changing Envisionment of
the world of the text, with changes in the levels of tentative-
ness as the text develops; and point of view.
Given such machinery and the goals of our project, we seek
to annotate texts in a way which represents the Ideal Reader's
processing of them and which provides the basis for the con-
struction of probing devices designed to assess a real reader's
experience of the text.
I would like to return to the passage I started out with, the
story about the uncommonly gifted woodcutter, to show some-
thing about how this annotation is used in our work.
The method requires an initial segmentation of the text,
since we are assuming, for theoretical purposes and contrary
to fact, that readers bring their full armament of interpreting
skills to each piece of text without peeking at the next one.
The segmentation we provide is, of course, arbitrary; since we
are not equipped to carry out the kind of research it would
take to find out what natural segmentations real readers would
make, we have to be arbitrary. In general, our segmentation
procedure amounts to taking one clause at a time.
264 / Charles J . Fillmore

In my demonstration I am going to do the first segment of

the Nasrudin story with some degree of completeness, and then
just briefly touch on a few other aspects of the passage. The
first segment is
The carpenter was astonished XXXXXXXXXXXXXX
The Ideal Reader will take note that this is the first segment
of the text, that it is not a complete sentence (since there is
no following punctuation), that it has a definite noun phrase
as its subject, a passive factive complement verb as its predi-
cating element, and the simple past as its tense.
The lexical item carpenter introduces a Carpentry schema,
which is merely another way of saying that when we understand
the noun carpenter we do so by knowing something about what
carpenters do. (It is a word for which one would be hard put
to distinguish semantic information about what the word means
and practical information about what carpenters do.) The per-
son designated as the carpenter is the practitioner in a work
schema, the other elements of which include the tools, the
materials, and some notion of the products of a carpenter's
work. Our schematic knowledge about carpentering could be
spelled out in great detail, with, for example, different types
of activities associated with particular tools; but since we, the
analysts, know how the story develops, we know that what we
need for our Ideal Reader is a mere skeleton of the whole
schema, with only certain aspects drawn out in detail, these
involving the carpenter's need for wood. From that we get the
K-links that connect wood with trees and the cutting and mill-
ing of lumber.
The next major lexical item in the segment is astonished.
This word, we allow ourselves to say, introduces an Astonish-
ment schema. To understand what it means to be astonished
is to have an outline knowledge of the kind of scenario that
could lead a person to have the experience which this word
describes. In this schema there is at least the experiencer of
the emotion, a perception on the experiencer's part which
triggers the emotion, the event which was responsible for that
perception, and a history or set of expectations against which
this causing event stood out in sharp conflict. (One is
astonished when one notices that something very unexpected
has occurred.)
There are no K-links between Carpentry and Astonishment.
There i s , of course, a T-link: it is the workman in an in-
stantiation of the Carpentry schema who is the experiencer in
an instantiation of the Astonishment schema.
In addition to the Content schemata of Carpentry and Aston-
ishment, we recognize in this segment certain schemata in the
Text domain. These include the grammatical knowledge that
the verb astonish is a factive complement verb, giving rise to
the expectation that what follows is going to be a description
 Ideal Readers and Real Readers / 265

or mention of the event which, in the world of the text,

caused the astonishment experience. When the next segment
is exposed, the Ideal Reader is prepared to integrate the in-
formation it provides into the Content domain of the text-world
as the cause of the carpenter's astonishment. More super-
ficially, knowledge of the kind of word astonish is primes the
Ideal Reader to expect that the very next segment will be at,
to, or that, followed by a nominal, verbal, or clausal descrip-
tion of the causing event.
The expectation that what follows is to be an account of the
causing event comes not simply from the form of our segment,
but is further informed by the knowledge that the segment is
not a complete sentence. There could be a text which read
The house collapsed. The carpenter was astonished.
where our segment was a complete sentence. But that is not
what we have here.
There are also Text-domain schemata associated with the
definite article and the simple past tense, and their activation
invites the Ideal Reader to hypothesize that what we have here
is the beginning of a particular type of narrative, one in which
characters and props can be introduced with definite noun
phrases or proper nouns, and in which a presupposed time
point can be presented without explanation. These schemata,
and the conditions in which they are introduced, invite the
Ideal Reader to activate a particular schema from the Genre
domain. There are kinds of third-person narrative in which
such in medias res textual features are common and proper.
(I am only pretending, of course, to have access to a theory
of genres. Since our collection of texts is made up mostly of
narratives, expository prose, personal letters, and simple
poems, there are not really many distinctions of genre that
we have had to worry about. If we are to choose from that
limited inventory of genres, we are already fairly safe, after
even this first short segment, to guess that we are in a narra-
tive.)
After exposure to this simple segment, The carpenter was
astonished . . . , the Ideal Reader has two Content schemata in-
stantiated, has T-links between them provided by the gram-
matical relations present in the segment, has expectations
about both the content and the form of what the next piece of
the sentence is likely to be, and has an active hypothesis about
what kind of a text it is dealing with. Questions the inter-
viewer might ask of a real reader, on the presentation of this
first segment, to see how closely this real reader matches the
accomplishments of the Ideal Reader, include the following:
Do you know what a carpenter does?
What does it mean to be astonished?
266 / Charles J . Fillmore

When we turn the page and look at the next part of the
paragraph, what do you think we're going to find out
about?
Can you guess what the next word is going to be?
That's a pretty good guess. Can you think of anything
else?
So much for the first segment. With third graders we tend
to postpone questions about the genre until at least one or two
complete sentences have been exposed.
In the next segment, Nasrudin is described as weak-looking.
A Physical Strength schema, or scale, has to be introduced
into the Ideal Reader's awareness, with the knowledge that
weak and strong are the two extremes, and the associated
knowledge that somebody who is strong can do more work and
heavier work than somebody who is weak. This, of course,
turns out to be related to the carpenter's surprise, since Nas-
rudin looked weak but was asking for work that required
strength. The second part of the word weak-looking raises in
the Ideal Reader's mind the question of whether somebody who
is described as weak-looking really is weak. A predication of
appearance naturally invites a question about reality. That
question, you will recall, gets answered very soon in the story.
We can clearly put the Parsimony Principle to work with the
first full sentence, which is: 'The carpenter was astonished
that such a weird, weak-looking creature as Nasrudin was
applying for a job.' Without the Parsimony Principle, we could
imagine the carpenter peeking in the door of a personnel office,
or an employment bureau, and seeing Nasrudin standing in line.
With it, we try to use the characters in our scene maximally for
filling out the introduced schemata. In the applying-for-a-job
scene, we make Nasrudin the applicant and the carpenter the
interviewer. In the work for which Nasrudin is applying, we
make Nasrudin the potential employee and the carpenter the
potential employer.
These T-linking assumptions can be wrong, so we will put
them in El and keep them there until we see whether or not
they get immediately corrected. Not only do they not get cor-
rected: the next sentences cannot be made intelligible unless
they are true. The next words of the carpenter are, 'Okay,
I'll give you a chance. Take this ax and chop as much lumber
as you can.' The relationship between interviewer and inter-
viewee is suddenly transformed into a relationship between em-
ployer and employee.
The text offers the Ideal Reader a number of places to 'read
between the lines'. The first sentence, we have seen, is
apparently about the carpenter's inner life: it reports that
he was astonished at what he saw. The very next sentence
is a record of the carpenter's words: 'Okay, I'll give you a
chance.' The first word, 'Okay', can count as a signal of
assent; but a signal of assent does not follow naturally a
 Ideal Readers and Real Readers / 267

statement about that person's mental experiences. The first

sentence has to be understood as introducing, by presupposi-
tion, the first move in the narrative. We then construe Nas-
rudin's 'applying for a job' as an act of asking the carpenter
something like 'Will you give me work?1 It is to that appeal,
not explicitly present in the text, to which the utterance,
'Okay, I'll give you a chance', is a response.
Similar cooperation is required of the reader in the transi-
tion between the following two sentences, the first one in
quotes:
'Take this ax and chop as much lumber as you can.'
At dusk Nasrudin returned.
To make this sequence cohere, the Ideal Reader has to build
into its picture of the story the information that the carpenter
held in his hands or otherwise indicated the ax, that he spoke
the quoted sentence to Nasrudin, that Nasrudin took the ax
and went off to the forest to cut down trees with it, and that
it was at dusk of the same day that Nasrudin returned to the
place where the carpenter was. The text tells us none of that,
but we have to believe it. The Parsimony Principle would have
us believe that the carpenter's order was obeyed, since the
author does not tell us that it was ignored or defied. We are,
in short, not free to believe that Nasrudin might have stared
at the ax all afternoon, giving up his ambitions as a carpen-
ter's helper and returning at dusk to Istanbul.
The patterns of inferencing in this story reach great com-
plexity, especially in the place where the reader figures out
that Nasrudin has claimed responsibility for creating the Sahara
Desert, and where the reader realizes in the domain of Genre
that our narrative is a joke.
The reader of my text will have figured out (with relief or
disappointment, I cannot guess which) that the few remaining
pages are not going to provide a complete account of the Nas-
rudin story.
12. I explained earlier that one of our project's unwelcome
problems was that of coping with flawed texts. The Ideal
Reader abstraction works out most satisfactorily and most
straightforwardly with well-constructed texts, texts in which
the author's plans are discernible and in which the author's
assumptions about the text's readers are reasonable. Almost
all of the texts we have examined in our work are seriously
flawed in one way or another.
Here is one, picked almost at random:
In 1877 a machine appeared which surprised many people.
Can you guess the name of this strange new machine? As
you spoke into the mouthpiece and turned the handle, a
tube covered with a thin piece of tin moved around. As
268 / Charles J . Fillmore

the tube moved a needle pressed deep lines into the tin.
As you turned the handle once more, the needle touched
against the same lines and played back your words. This
was the9 first phonograph. How different from the hi-fi of
today!
Not only do our third-grade subjects have difficulties with
this text; so do many very literate adults. Our subjects had
a much worse time of it, however. Although these children
knew the words recordplayer and stereo, not one of them knew
the word phonograph. Where the text tells us that the tube
moved, those adults who know what the antique roll-type phono-
graph looked like are able to picture a cylinder rotating. The
children knew about toothpaste tubes and video tubes, but that
knowledge did not help them at all. Not one child was able to
create anything remotely resembling the intended image. The
pictures they were induced to draw of the machines they
imagined ranged from tractors to microphones to Coca-Cola
machines.
One attitude we could take toward such a passage, is that it
is merely difficult, and that nobody has a right to complain
about finding difficult items in a test. Our interview protocols
on this item convince us, however, that it simply does not be-
long in a reading test. The one child (out of 30 subjects) who
figured out that the machine was a record-player could answer
very few of the questions correctly, largely because she be-
came confused by the word phonograph, which she said was
unfamiliar to her. (She pictured it as a record player until
she came upon the word phonograph, and then she became un-
certain.) By contrast, a boy who misread the word machine
as magician and who performed the wildest sort of mental
acrobatics in order to preserve that part of his Envisionment,
was able, by absolutely absurd reasoning plus a certain amount
of test-taking know-how, to choose mainly correct answers to
the test questions.
The Ideal Reader for such a test has to be, quite simply, a
person who happens to know what the oldest phonograph looked
like and who can be cooperative enough with the text, once the
information that it is a phonograph is finally provided, to real-
ize that the description of the machine is some author's poor
attempt to describe what the Ideal Reader already knows it
should look like. For a text like this, the concept of the Ideal
Reader is workable, but pointless. In the usual case, we as
readers have to know something about the real world in order
to build on that to construct an Envisionment of the world of
the current text; in a case like this, however, what we have
to know in order to understand the text exhausts what the
text tells u s . 1 0 That, in my mind, is a clear case of a bad
text, most assuredly a bad test item.
 Ideal Readers and Real Readers / 269

13. Forthcoming from the project on which I am reporting

are an analysis of the interview data from our subjects and a
critical study of the worst of the test items. Here I have
merely tried to give an informal account of the goals and
methods of the project and a survey of the problems one has
to face when trying to monitor closely the reading experience,
and to show the relevance to this enterprise of the Ideal
Reader abstraction.
NOTES

The work reported here has been supported by the National

Institute of Education under Grant No. G-790121 Rev. 1, 'Text
semantic analysis of reading comprehension tests*. Co-Pis are
C. J. Fillmore and Paul Kay; others on the project are reading
specialist Judith Langer and a team of Berkeley graduate stu-
dents from Anthropology, Education, and linguistics (Karen
Carroll, Linda Coleman, Katharine Kovacic, Thomas Larsen,
Mary Catherine O'Connor); the project has received good ad-
vice from Patrizia Violi, Mary Sue Ammon, and Haj Ross.
1. The schools in which we have been allowed to do our
work are Cragmont Elementary School in Berkeley and Sequoia
Elementary School in Oakland. The principals of the two
schools, Dr. Benton Ng and Mr. Alfred C. Valdix, have our
gratitude.
2. These have included CTBS Comprehensive Tests of Basic
Skills, Level 1, Form S, Complete Battery, Expanded Edition,
CTB/McGraw-Hill, I n c . , copyrights 1968, 1969, and 1973; and
the Metropolitan Achievement Tests: Reading Instructional
Tests, Form J I , Elementary, The Psychological Corporation,
Division of Harcourt, Brace, Jovanovich, copyright 1978; and
(not cited in this report) the California Achievement Tests,
the Stanford Achievement Tests, and the Gates-McGinnitie
Reading Tests.
3. Form JI, page 15.
4. One part of each interview is the genre question, which
generally takes some such form as 'Where do you think you
might read a passage like this? In a story book? In a letter
from a friend? In the Weekly Reader?1 One answer we re-
ceived, in connection with an intendedly humorous narrative
passage, was 'in a CTBS test'.
5. Research applications of the Ideal Reader concept must
make use of a large and unarticulatable common sense component,
We all know that there are situations in which texts communi-
cate new information by presupposition, and very many situ-
ations in which a text develops its argument by reminding its
readers of things they already know. The Ideal Reader who
has the subtlety we need it to have will be able to convert, by
Peircean abduction, information that is formally presupposed
into information intended to be derived from the text, and will
270 / Charles J . Fillmore

be able to recognize situations in which old information is be-

ing introduced mainly as a link in the chain of an argument or
a narrative.
6. It will seem, to many readers of this text, that my col-
leagues and I have a hopelessly naive view of the nature of
the reading experience. People find pleasure in reading texts
they only minimally understand, there are often rhythms and
patterns in texts that can only be detected on repeated re-
reading", readers can experience suspense and surprise in a
narrative even if they know, as in a thoroughly familiar text,
exactly how it's all going to end, and quite frequently texts
are designed precisely to invite idiosyncratic imaginative re-
sponses in their readers--a situation for which the notion of a
'correct' interpretation is thoroughly unwelcome. It should be
remembered that the simplistic view of the reading experience
which we adopt for our research purposes is aimed at a level
of understanding for which readers can be reasonably expected
to choose 'best' answers j to multiple-choice questions.
7. It is this criterion, in conjunction with the decision to
define the Ideal Reader as relativized to given texts, which
simultaneously makes our work do-able (by defining a stopping
place) and guarantees that the Ideal Reader offers only limited
predictability to 'good' readers. There is obviously no reason
whatever for a reader to create only those inferences and em-
bellishments which the current text will build upon.
8. The Parsimony Principle is found in essentially the form
suggested here in such works as Harvey Sacks (1972), 'On the
analyzability of stories by children,' in J. Gumperz and D.
Hymes ( e d s . ) , Directions in Sociolinguistics, Holt, Rinehart
and Winston, Inc.; and Yorick Wilks (1973), 'Preference se-
mantics,' in E. Keenan ( e d . ) , Formal Semantics of Natural Lan-
guage, Cambridge University Press, and (1973), 'Understand-
ing without proofs,' in Proceedings of the Third International
Joint Conference on Artificial Intelligence, Stanford Research
Institute, Menlo Park, California. The Wilks references were
brought to my attention by Robert Kirsner.
9. From Comprehensive Tests of Basic Skills, Form S,
Level 1, page 8. Reproduced by permission of the publisher,
CTB/McGraw-Hill, Del Monte Research Park, Monterey, Cali-
fornia 93940. Copyright © 1973 by McGraw-Hill, Inc. All
rights reserved. Printed in the United States of America.
10. The case has been slightly overstated. One does learn
in this passage that the first phonograph appeared in 1877.
 LITERATURE: WRITTEN AND ORAL
William Bright
University of California, Los Angeles

My title suggests a topic more appropriate for a large book

than for a half-hour talk, so I should say at the outset that
my main goal is to discuss some recent research, by Dell Hymes
and others, in which American Indian oral narratives are ana-
lyzed as having the form of verse or poetry; and I want to ex-
press a word of caution about some of that work. However, for
the sake of orientation, I would like first to mention briefly my
personal background in the field, and then to clear some termi-
nological underbrush by discussing the problems of defining
the crucial terms 'literature' and 'poetry 1 .
My own linguistic training was originally in the post-Bloom-
fieldian school of the 1940s and 1950s--even though it was, to
my good fortune, more post-Sapirian than post-Bloomfieldian,
and was specifically in the tradition of American Indian anthro-
pological linguistics. In the context of the present Georgetown
University Round Table on Discourse, I am thus aware of be-
ing in a somewhat paradoxical position: on the one hand, I
can remember when descriptive linguistic analysis of American
Indian languages, starting from the phonetic level, rarely went
beyond the level of the word, much less to that of the sen-
tence. x On the other hand, it was emphatically borne in upon
the academic heirs of Boas, Kroeber, and Sapir that the col-
lection and analysis of texts were absolutely essential parts of
their linguistic field work; thus, for me to receive my Berkeley
doctorate in 1955, I not only had to file a grammar of Karok as
my 'official' dissertation, but simultaneously to submit two other
comparably weighty but 'unofficial' volumes: a Karok dictionary,
and a copious collection of Karok texts. The matched sets of
grammars, dictionaries, and texts which were thus produced
over several decades have quite regularly been issued in the
University of California publication series in linguistics, con-
tinuing the Boasian tradition manifested in publications of the

271
272 / William Bright

Smithsonian Institution and elsewhere. So the reality of lan-

guages as consisting of connected discourse remained part of
the consciousness of many American linguists for years; yet
all those volumes of native American texts on our library
shelves, though consulted by the occasional anthropologist or
folklorist, have tended to be neglected by linguists--and even
more so by scholars of literature. But now things are chang-
ing. Owing in particular to the efforts of Dell Hymes and of
Dennis Tedlock, the study of American Indian narrative is
having a revival, from both linguistic and literary viewpoints. 2
However, much work remains to be done, not the least of
which is that of terminological clarification. How should we de-
fine the differences between formal and informal language,
written and spoken language, literary and colloquial language
(cf. Tannen 1980)? How do we define the term 'literature' it-
self? Is it appropriate to speak of 'oral literature'? If so,
what are its distinctive characteristics? Within the framework
of 'literature', how can we define 'poetry'? And what about
'oral poetry': how is it to be distinguished from song or
chant on the one hand, and from prose on the other? I touch
on these problems only briefly, and then go on to a more spe-
cific question, raised specifically by the work of Hymes--
namely, what are the distinctive characteristics of Native
American oral narrative as poetry?
I would like first of all to remove from discussion the words
'formal' and 'informal'. I prefer to use these to refer to con-
trasting sociolinguistic registers--or as poles on a sociolinguis-
tic continuum--described by Ferguson (1959) under the rubric
of 'Diglossia'. 3 A more relevant pair of terms is 'written lan-
guage', i.e. writing, vs. 'spoken (or oral) language', i.e.
speech. These terms too are useful, as referring to media of
transmission, but still are not directly applicable to defining
'literature'. The point is that much written material--e.g. most
of the content of newspapers--is not generally considered to
be literature; yet other materials originally produced in oral
form--e.g. the Homeric epics--are universally considered literary
works, whether recited aloud or reproduced in printed form.1*
We are then confronted with the terrible question of how to
define literary language. 5 I merely suggest here that 'litera-
ture' refers, roughly, to that body of discourses or texts
which, within any society, is considered worthy of dissemina-
tion, transmission, and preservation in essentially constant form.
In our society, we typically associate literature with the written
medium; however, works originally composed in writing can of
course be performed orally, as when parents read aloud to chil-
dren, or when poets give public 'readings'. A further ques-
tion, however, is the appropriateness of the term 'oral litera-
ture'--a phrase which for some people probably still constitutes
an oxymoron: how can something consist of litterae. and yet be
oral? Still, the term has been widely used for literature which
is composed, transmitted, and performed orally; well-known
 Literature: Written and Oral / 273

examples are the Iliad and Odyssey in their original forms, and
the much longer epics of Ancient India--the Mahabharata and the
Ramayana- -as well as the Vedas and a whole huge corpus of
ancient Hindu literature. In modern times, the term 'oral liter-
ature' has been applied to the body of myths and legends exist-
ing in nonliterate societies of, for example, Native America and
Africa; witness book titles such as Ruth Finnegan's Oral Liter-
ature in Africa (1970). 'Oral literature' even covers the large
number of jokes, riddles, song texts, etc. which exist primarily
in the oral traditions of literate societies like our own: the
term 'folklore' i s , of course, often applied to such material,
but it suggests a different level of evaluation which I would
prefer to avoid. In any case, it is clear that texts which were
originally oral may be transcribed and transmitted in the written
medium; examples would include printed versions of the Iliad as
well as the contents of the Journal of American Folklore.
Finally, if I read such printed materials aloud to a child or a
friend, the text passes back into the oral medium. It must
be recognized, then, that the difference between speech and
writing is not necessarily basic to a definition of literature.
A secondary problem is that of possible distinctive qualities
in literature whose origin is oral. Finnegan argues persuasively
that a typical oral literature may differ in quantity, but not in
quality, from a typical written literature. Oral literatures have
developed a variety of genres from the epic to the love song
(1973:116); they display the same types of content as written
literatures, including intellectual perception and aesthetic ex-
pression (1973:118-124).
A question that arises here, however, is whether oral litera-
ture involves verbatim memorization, so that texts can be pre-
served in the unchanging form characteristic of written litera-
ture. Here we find some controversy: Jack Goody (1977:116-
120) has emphasized that many oral literatures show constancy
only in overall structure and in the recurrent use of formulaic
expressions, rather than in word-for-word repetition; yet other
writers have reported many cases in which high value has been
attached to exact memorization. A remarkable case is known
from Ancient India; once the Vedic hymns had been orally com-
posed (in strict meter), it was considered that their religious
effectiveness depended on their being transmitted without the
slightest change, and elaborate methods of teaching and memori-
zation were established to ensure this--culminating in the cele-
brated grammar of Panini. But since the early Sanskrit lin-
guistic texts date from the fifth to the late fourth centuries
B . C . , and since our earliest evidence of written Sanskrit dates
from the mid-third century B.C. (cf. Basham 1954:387-388,
394), we have reason to believe that PSnini's grammar itself--
which not only described but standardized a language--was
composed and initially transmitted without the use of writing,
i.e. in the oral medium alone. This belief is supported by the
274 / William Bright

form of sUtras or rules used by Panini and other early Hindu

scholars,6 in which memorization was facilitated by extraordinary
brevity. Even after writing was introduced to India, memori-
zation of sacred texts such as the Vedas by purely oral means
has continued down to modern times. For us Westerners, who
have leaned for so long on the crutch of writing, it is hard to
realize the capacities of the human memory! In a very differ-
ent context, Joel Sherzer (1980) reports that perfect memoriza-
tion of oral texts is still practiced by the Cuna Indians of
Panama. Our conclusion, then, must be that some oral litera-
ture can be and is transmitted verbatim, though much is not.
On the other side of the balance, we should recall that written
literature has not always been transmitted without variation.
Finnegan (1973:140) has, in fact, suggested that it is printing
rather than writing alone which is responsible for our Occiden-
tal ideas about the fixity of literary texts.
Yet if all the foregoing is accepted, we come to a still more
terrible question: How can we define 'poetry'? 7 A concept
which is traditional in our society, and still held by many indi-
viduals, is that 'poetry' refers exclusively to texts organized
in regular phonological patterns of meter, and often of rhyme
as well. By this definition, it is clear that many oral texts
have been composed in well-defined meters; examples include
the Iliad, the MahabhUrata, or Anglo-American folksongs of
recent centuries; but such texts were originally chanted or
sung, and we may wish to exclude from consideration oral
literature in which meter is imposed by a separate rhythmic
pattern. Of greater current relevance is the fact that, among
the literary texts of our own society--at least since the time of
Walt Whitman--there has been increasing recognition of 'free
verse', or poetry without well-defined metrical structure. At
the present time, most new poetry published in English lacks
recognizable meter. The question then arises of how it can
be distinguished from prose--along with the subsidiary problem
of defining the so-called 'prose poem'. Many poets don't seem
to worry much about how to define poetry: they simply know
a poem when they meet one. However, a rough definition that
appeals to me is this: a poem is a text in which linguistic
form--phonological, syntactic, and lexical--is organized in such
a way as to carry an aesthetic content which is at least as im-
portant, as regards the response of the receiver, as is the
cognitive content carried by the same text. Finnegan (1977:
89) says something like this when she states that, in poetry,
'style and structure are a kind of end in themselves...' In
traditional English poetry, both oral and written, a large part
of the aesthetic content was carried by well-known phonologi-
cal patterns of meter and rhyme. In more recent English
poetry, phonology still plays an important part, insofar as
sound is still used for aesthetic effect; but grammatical and
lexical8 structures are also exploited extensively for poetic
ends.
 Literature: Written and Oral / 275

Given the recognition of nonmetrical poetry in our own

modern society, can we recognize similar poetry in older or
more traditional cultures? In fact, it has long been realized
that parts of the Hebrew Bible--the so-called 'poetic' books,
such as the Psalms--are examples of nonmetrical poetry, pre-
sumably of oral composition; linguistic features other than
meter, e.g. syntactic parallelism, carry much of the aesthetic
content. By contrast, the historical books of the Bible lack
these special linguistic features, for the most part, and are
thus identified as prose. But what about the works of oral
literature which have been transcribed in preliterate societies
of our own century, by linguists and anthropologists? Can a
distinction between prose and poetry be recognized in, for
instance, American Indian materials?
In many older writings on American Indian literature (e.g.
Day 1951), one finds the implicit viewpoint that song texts
are poetry--but that everything else, such as myths, are
prose. This view perhaps derived from the traditional notion,
in English literature, that poetry must be metrical; and a musi-
cal performance would, of course, associate a song text with a
particular meter. By contrast, since American Indian myth
texts were normally not sung, they were classified as prose.
More recently, this view has been eclipsed, and modern col-
lections of American Indian literature translated into English,
such as Jerome Rothenberg's Shaking the pumpkin (1972),
contain numerous examples of myths which are presented typo-
graphically as poems--i.e. in 'lines' of verse. However, as
Finnegan warns us (1977:25):
. . . written literary poetry is normally typographically
defined . . . [but] obviously this particular rule will not,
by definition, work for oral poetry. One is thus forced
to look for other, apparently more 'intrinsic' character-
istics . . .
We need to ask, then, whether a poetic structure exists in the
original text, or whether it has been imposed by the English
translator. In fact, although the translators have often been
skilled poets, they have just as often been totally ignorant of
the native languages concerned; their procedure has simply
been to take literal English translations published by linguists
and anthropologists, and to rewrite them in more poetic form.
In such cases, we have no assurance that the native-language
texts are in any way recognizable as poetry rather than prose.
A long-standing concept of 'poetry' (as suggested, for exam-
ple, by de Groot 1946) is that it can be defined minimally as
'discourse organized in lines'—i.e. strings defined not arbi-
trarily, by the width of a page, but structurally; such a view
was also stated by Dell Hymes in 1960. A step toward demon-
strating the existence of such lines in American Indian narra-
tives was taken by Tedlock (1972) in his translations from
276 / William Bright

Zuni. Here we have a translator who knows the original lan-

guage, and who has scrupulously tape-recorded the expressive
features of pitch, loudness, rhythm, timbre, and--above all--
of pause as used in the oral performances of Zuni storytellers.
Thus, although Tedlock does not explicitly point out many lin-
guistic features which would identify Zuni narratives as poetry,
his organization into lines on the basis of pause strongly points
to a poetic structure in the Zuni originals. 9
Another approach, however, has been taken by Hymes (1976,
1977, 1980a,b), who has focused not on features of live perform-
ances, but rather on patterns that can be observed in published
texts--namely, the ways in which vocabulary, word formation,1 0
syntax, and semantics are used to create literary structures.
In his own research, Hymes has shown that Chinookan texts,
transcribed and published years ago, can be divided into
verses—defined not by meter or rhyme, but by other types of
structural features. As he has written (1976:153-156),
Verse having a defining phonic numerical regularity is
typically in view in discussion of 'meter' . . . Chinookan
oral narrative is at quite the opposite pole: it has a
characteristic grammatico-semantic repetition within a frame
as its base . . . One might introduce the term 'measure' for
verse that answers to the second pole . . . The Shakes-
pearean sonnet is metrical; Louis Simpson's 'Deserted Boy'
[a Chinookan narrative] is measured.
In seeking a basis for the analysis of measured verse, Hymes
has put great emphasis on the use of sentence-initial particles,
translatable into English as 'and', 'so', 'then' etc. With this
concept of the verse as basis, Hymes finds it possible to recog-
nize other structurally defined units, both smaller--such as the
line, defined typically in terms of its unity as a grammatical
predication--and larger, such as the scene and act, often de-
finable in terms of actors present or shifts of locale. Building
on the work of both Tedlock and Hymes, I myself have attempted
(Bright 1979, 1980a,b,c) to identify structures of measured
verse in the myths of the Karok tribe of California, and to
produce English translations in a corresponding poetic form.
Studying a tape-recorded text both with the approach of Ted-
lock--focusing on the expressive features of performance--and
with the approach of Hymes--identifying verses, etc. in terms
of linguistic structure--I find (Bright 1979) that the two ap-
proaches coincide 90 percent of the time in their identification
of basic units. X1
Hymes is very positive in his identification of American Indian
narratives as 'verse' or as 'poetry'. 1 2 Yet he admits that the
recognition of this poetic structure cannot be carried out in
any mechanical way. For example, although sentence-initial
particles provide the primary cue for the recognition of verse
 Literature: Written and Oral / 277

patterning in languages as widespread as Chinookan, Karok,

and Tonkawa, Hymes notes (1977:439-440):
Once such patterning has been discovered in cases with
such markers, it can be discerned in cases without them.
The Clackamas [Chinookan] narratives of Mrs. Victoria
Howard do not regularly make use of initial particles . . .
To determine the organization of her narratives, one has
to recognize and abstract features that co-occur [emphasis
added] with the use of initial particle pairs in the [Wish-
ram Chinookan] narratives of Louis Simpson . . . The dis-
covery of such pattern is not arbitrary, because it is
governed by the coherence and articulation of the particu-
lar narrative, by a rhetorical pattern that pervades
Chinookan texts.
Hymes has shown great skill in carrying out such analysis;
and other researchers have begun to identify line and verse
structures, formally marked not necessarily by initial particles,
but by other phenomena such as grammatical andx 3semantic
parallelism, in other Native American literatures. Yet the
more subtle are the devices which mark verse structure, the
more cautious the researcher must be--especially when dealing
with dead and moribund languages, where it is impossible to
validate one's analysis with members of a living and creative
speech community. The delicacy of the task is recognized by
Hymes, but at times I feel that he relies more on his own
intuition as an English-language poet than on objective charac-
teristics of his Native American data. Thus, in justifying his
verse analysis of a Wasco Chinookan text, as told by Hiram
Smith, in which initial markers are scarce, he writes (1980a:
77):
In sum, pervasiveness of the meaning dimension of rhetori-
cal patterning asks for such an analysis, [Hiram Smith's
own] English retelling provides for i t , and the concord r e -
lations in the Wasco text seem to require i t .
Here the statement that 'the English retelling provides for it'
is what disturbs me, even though the English is that of Hiram
Smith himself: we risk falling into the error against which we
were so often warned by Boas, and more recently by Hymes
himself--that of imposing English-based categories on linguistic
data from other cultures.
The possibility remains open, in fact, that some Native
American cultures--or culture areas--simply told their narra-
tives as prose--or, to put it more properly, that they lacked
a distinction between prose and poetry in their literary dis-
course. An area that concerns me in this regard is Southern
California. When I look at the Cahuilla texts published by
Hansjakob Seiler (1970), or at the delightful Diegueno text
278 / William Bright

collected by Leanne Hinton under the title 'Coyote baptizes the

chickens' (1978), I find neither initial particles nor grammatico-
lexical parallelism; I find, in fact, nothing like the clearly
recognizable structures of measured verse which we see in the
Pacific Northwest and in Northern California. To be sure, I
may simply be missing something; other scholars should cer-
tainly examine the Southern Californian materials. But lin-
guists have already in recent years tended to postulate too
many poorly founded universals: let us not assume that the
poetry/prose distinction must exist everywhere.
Other questions that arise are these. Consider a culture
like that of the Karok, where traditional narratives show abun-
dant characteristics of measured verse. But if Karok poetry
(apart from song texts) is what we find in myths, what are the
characteristics of Karok prose? Does a distinction in fact exist,
in the literature of the Karok (or the Chinookans), between
prose and poetry? Or should we say, as was perhaps the case
in some early Indo-European societies, that all literature was
poetic, and that prose was used only for nonliterary discourse?
Data now available on Karok are inadequate to answer these
questions. ll * But many other native American speech communi-
ties continue in full function, and deserve study in terms of
all aspects of Hymes' 'ethnography of communication'. Tedlock
can probably tell us for Zuni, or Sherzer for Cuna, a great
deal about the actual formal distinctions between different types
of poetic vs. nonpoetic and literary vs. nonliterary discourses.
In fact, we recently have a tantalizing report from Anthony
Woodbury concerning the Central Yup'ik Eskimo of Alaska: he
suggests that 'lines' and 'verses' are characteristic of all Cen-
tral Yup'ik discourse, and that his identification of these units
is readily verifiable by native speakers--but that 'lines' and
'verses' are used differently in distinct types of discourse,
ranging from 'poetry' to ordinary conversation. I look forward
eagerly to learning more about Woodbury's findings.
In the meanwhile, returning to the topic of oral literature, I
wish to add my voice to those of Tedlock, Hymes, and others
who have emphasized that Native American oral narratives must
be taken seriously as literature. We should learn, for a broad
range of Native American societies, how to differentiate prose
and poetry--and possibly other genres; we should learn the
defining characteristics of each genre; we should learn the
social function of each; and we should attempt to understand
the nature of written literature as it develops in Native Ameri-
can languages. Students of these topics have much to gain,
not only in increased appreciation of the richness of literary
traditions in the Western Hemisphere, but also in improved
comprehension of the nature of literary discourse among human
societies in general.
 Literature: Written and Oral / 279

NOTES
Thanks for suggestions and encouragement to Linda Arvanites,
M. B. Emeneau, Paul Friedrich, Dell Hymes, and Ken Lincoln.
1. One widely admired 'grammar' of the 1940s barely went
past what we might now call the morphophonemics. Against
that background, it is encouraging to realize that a conference
devoted exclusively to American Indian syntax has been held
in Calgary this spring.
2. Indeed, the lead article in a 1977 issue of that bastion
of literary scholarship, the Publications of the Modern Language
Association, was Jarold Ramsey's study of an Oregon Indian
myth.
3. To be sure, Ferguson uses the terms 'H(igh)' and 'L(ow)'
instead of 'formal' and 'informal'; but these labels invite con-
fusion with the social dialects of higher vs. lower social classes
or castes. Furthermore, although formal varieties, in countries
like India, are often written, they need not be--as when they
are used in impromptu political speeches or sermons; and in the
same connection, they may lack any of the aesthetic quality
which would prompt us to label them as 'literary utterances'.
By contrast, 'informal varieties' are not usually written, but
sometimes they are--as, increasingly, in some types of popular
fiction and comic books--which might or might not be considered
forms of 'literature'.
4. The possible confusion that results from use of these
terms is illustrated when Ochs (1979) refers to 'writing' as
planned v s . 'speech' as unplanned; or when Chafe (1979) sees
'writing' as integrated and detached, but 'speech' as fragmented
and involved. Without denying the validity of these labels, I
would suggest that they apply not strictly to the media of
'writing' and 'speech' as such, but more properly to the con-
trast between literary and colloquial language.
5. As Ruth Finnegan (1973:118) has said, 'The whole area
of "What is literature?" is of course a controversial and unend-
ing one.'
6. Thus the famous last sutra of Panini is a a, the interpre-
tation of which is: 'The long low vowel [a:] has, as its short
counterpart, the raised vowel [a]' (Renou 1954:144).
7. Finnegan (1977:24), with the length of an entire book at
her disposal, says, 'I cannot here enter into deep discussion of
the question "What is poetry?"'
8. On a cross-cultural basis, Finnegan raises another im-
portant point about prose vs. poetry: 'the local classification
of a piece as "poetry" [is] in one sense . . . the most important
[factor], but it is by no means simple. For one thing, the
relatively neat formal differentiation we make in our own cul-
ture between poetry and prose is not recognized everywhere
. . . ' (1977:25). 'It emerges, then, that any differentiation of
"poetry" from "prose", or indeed of "poetry" as a specific
literary product or activity, can only be approximate . . . the
280 / William Bright

whole delimitation of what is to count as "poetry" necessarily

depends not on one strictly verbal definition but on a series
of factors to do with style, form, setting and local classifica-
tion, not all of which are likely to coincide1 (27).
9. He states, indeed, that 'prose has no real existence out-
side the written page' (1972: xix).
10. Regarding the poetic 'line' defined merely in terms of
pause, Hymes comments (1977:453-454): 'there remains the
problem of differentiating pause that is motivated, that
heightens the organization of lines, from pause that is in-
herent in the spoken medium . . . one cannot be content with
a purely definitional victory for the claim of the pervasiveness
of poetry as lines . . . Pausing may itself be culturally shaped,
but if it is, one needs evidence beyond the fact of its occur-
rence.'
11. This gives me confidence that occasional ambiguities of
one approach can be resolved by reference to the other: thus
linguistic sequences which have two possible grammatical inter-
pretations can be disambiguated by reference to phenomena of
pitch and pause; and conversely, accidental hesitations which
create 'false' pauses in performance can be recognized because
they are interruptions of normal sentence structure.
12. As he has recently stated (Hymes 1980b: 34-35): 'the
study of Native American languages has yet to take its mater-
ials seriously enough . . . It now appears that we have misled
ourselves as to the myths and tales we have thought long
known. We have allowed to stand a perpetuation of the cardi-
nal sin, the distortion of another cultural reality through impo-
sition of categories of our own. We have thought that Native
American myths and tales are prose and have printed them as
such . . . All the collections that are now in print must be re-
done. They do not show the structure of the texts they repre-
sent . . . Hidden within the margin-to-margin lines are poems,
waiting to be seen for the first time.'
13. See Karttunen and Lockhart (1979) for Classical Nahuatl,
Tedlock (1979) for the colonial Quiche of the Popol Vuh, Nor-
man (1980) for modern Quiche, and Sherzer (1980) for Cuna.
14. During my major period of field work on Karok, in 1949
and 1950, the texts which I transcribed were mainly limited to
narratives--because I was interested in the myth literature, be-
cause 'stories' were an easy type of text to elicit, and because,
except for a few days at the end of my work, I had no tape-
recorder. So I never obtained any conversational texts. I
did transcribe--from dictation, not from tape--some nonnarra-
tive, ethnographic texts (Bright 1957:282-301), and they show
a structure of sentence-initial particles similar to that found in
narratives. Does this mean that I was actually getting ethno-
graphic poetry? Probably not; the situation in which a Karok
speaker dictated texts to me, word by word, was analogous to
the native situation in which stories were told to children for
piece-by-piece repetition, and so the style of narrative may
 Literature: Written and Oral / 281

have artificially been extended to descriptions of salmon-fishing

and sweathouses. To be sure, elderly speakers of Karok are
still living, and it may still be possible to tape-record more
natural samples of Karok discourse. However, no functioning
Karok speech community exists now, and so conditions are not
ideal for resolving these matters.
REFERENCES
Basham, A. L. 1954. The wonder that was India. New York:
Grove.
Bright, William. 1957. The Karok language. University of
California Publications in Linguistics, 13. Berkeley and Los
Angeles.
Bright, William. 1979. A Karok myth in 'measured verse':
The translation of a performance. Journal of California and
Great Basin Anthropology 1.117-123.
Bright, William. 1980a. Coyote's journey. American Indian
Culture and Research Journal (UCLA) 4.21-48.
Bright, William. 1980b. Coyote gives salmon and acorns to
humans. In: Coyote Stories II. Edited by Martha Kendall.
IJAL-NATS Monograph 6. Chicago: University of Chicago
Press. 46-52.
Bright, William. 1980c. Poetic structure in oral narrative.
In: Spoken and written language. Edited by Deborah
Tannen. Norwood, N . J . : Ablex.
Chafe, Wallace L. 1979. Integration and involvement in
speaking, writing, and oral literature. In: Spoken and
written language. Edited by Deborah Tannen. Norwood,
N . J . : Ablex.
Day, A. Grove, ed. 1951. The sky clears. New York:
Macmillan. [Reprinted, Lincoln: University of Nebraska
Press, 1964.]
de Groot, A. Willem. 1946. Algemene versleer. The Hague:
Servire.
Ferguson, Charles A. 1959. Diglossia. Word 15.325-340.
Finnegan, Ruth. 1970. Oral literature in Africa. Oxford:
Clarendon.
Finnegan, Ruth. 1973. Literacy v s . non-literacy: The
great divide? Some comments on the significance of 'litera-
ture' in non-literate cultures. In: Modes of thought:
Essays on thinking in Western and non-Western societies.
Edited by Robin Horton and Ruth Finnegan. London:
Faber and Faber. 112-144.
Finnegan, Ruth. 1977. Oral poetry: Its nature, significance,
and social context. Cambridge: University Press.
Goody, Jack. 1977. The domestication of the savage mind.
Cambridge: University Press.
Hinton, Leanne. 1978. Coyote baptizes the chickens (La
Huerta Dieguefio). In: Coyote stories. Edited by William
282 / William Bright

Bright. IJAL-NATS Monograph 1. Chicago: University of

Chicago Press. 117-120.
Hymes, Dell. 1960. Review of: Ob-Ugric metrics, by Robert
Austerlitz. Anthropos 55.574-576.
Hymes, Dell. 1976. Louis Simpson's 'The deserted boy.'
Poetics 5.119-155.
Hymes, Dell, 1977. Discovering oral performance and measured
verse in American Indian narrative. New Literary History
8.431-457.
Hymes, Dell. 1980a. Verse analysis of a Wasco text: Hiram
Smith's 'At'unaqa'. UAL 46.65-77.
Hymes, Dell. 1980b. Tonkawa poetics: John Rush Buffalo's
'Coyote and Eagle's Daughter'. In: On linguistic anthro-
pology: Essays in honor of Harry Hoijer, 1979. Edited by
Jacques Maquet. Malibu: Undena. 33-87.
Karttunen, Frances, and James Lockhart. 1979. The struc-
ture of Nahuatl poetry as seen in its variants. In: Estudios
de Cultura Nahuatl, Mexico City.
Norman, William M. 1980. Grammatical parallelism in Quiche
ritual language. In: Proceedings of the 6th Annual Meeting,
Berkeley Linguistic Society. 387-399.
Ochs, Elinor. 1979. Planned and unplanned discourse. In:
Discourse and syntax. Edited by Talmy Givon. New York:
Academic Press. 51-80.
Ramsey, Jarold W. 1977. The wife who goes out like a man,
comes back as a hero: The art of two Oregon Indian narra-
tives. PMLA 92.1:9-18.
Renou, Louis. 1954. La grammaire de Panini, fasc. 3. Paris:
Klincksieck.
Rothenberg, Jerome, ed. 1972. Shaking the pumpkin: Tra-
ditional poetry of the Indian North Americans. Garden City,
N.Y.: Doubled ay.
Seiler, Hansjakob. 1970. Cahuilla texts, with an introduction.
Language Science Monographs, 6. Bloomington: Indiana Uni-
versity.
Sherzer, Joel. 1980. Tellings, retellings, and tellings within
tellings: The structuring and organization of narrative in
Cuna Indian discourse. Paper presented at the conference
on 'Oralitti: Cultura, letteratura, discorso', Centro Inter-
nazionale di Semiotica e Linguistica, Urbino, Italy.
Tannen, Deborah. 1980. Spoken /written language and the
oral/literate continuum. In: Proceedings of the 6th Annual
Meeting, Berkeley Linguistics Society. 207-218.
Tedlock, Dennis. 1972. Finding the center: Narrative poetry
of the Zuni Indians. New York: Dial. [2nd ed. , Lincoln:
University of Nebraska Press, 1978.]
Tedlock, Dennis, 1979. Las formas del verso quiche, de la
paleograffa a la grabadora. Paper presented to the Primer
Congreso sobre el Popol Vuh, Santa Cruz del Quiche,
Guatemala. To be published in its Actas (Guatemala City:
Piedra Santa.)
 Literature: Written and Oral / 283

Woodbury, Anthony C. 1980. Rhetorical structure in Central

Yup'ik Eskimo narrative. Paper presented at the LSA Annual
Meeting, San Antonio. [Abstract in Meeting handbook, p .
79.]
 MEANING, RHETORICAL STRUCTURE,
AND DISCOURSE ORGANIZATION IN MYTH

Sally McLendon
Smithsonian Institution and
Hunter College of City University of New York

The role of texts in language. Linguists have increasingly

come to appreciate the centrality of discourse and texts in lan-
guage and communication. In a recent book examining the
organization of discourse in English, Halliday and Hasan have
gone so far as to propose that 'A text . . . can be thought of
as the basic unit of meaning in language' (1976:25) or as 'the
basic semantic unit of linguistic interaction' (1976:295).
The characterization of a text as the basic semantic unit of
linguistic interaction suggests that the text is the form taken
by the particular meaning(s) communicated in a linguistic inter-
action, and that the ability to produce and comprehend texts
requires a shared understanding on the part of speakers of
how meanings are to be associated with texts (or how meanings
are to be created by texts) in a language. Halliday and Hasan
(1976:299) assume that 'what creates text is the TEXTUAL or
text-forming component of the linguistic system of which cohe-
sion is one part'. Unfortunately, they do not specify what
this text-forming component consists of besides the phenomenon
of cohesion which is the topic of their book, and their study
focuses on written texts to the virtual exclusion of oral texts.
The rhetorical structure of texts. It has been clear for a
long time that a text involves the simultaneous manipulation of
several formal systems of language: the sound system, the
lexical system, and the grammatical system. Linguists have
largely left to literary critics, however, the investigation of
what seems appropriately called the rhetorical system, which
may well be the most important aspect of the organization of
form in a text to convey meaning (and is probably to be

284
 Meaning, Structure, and Organization in Myth / 285

identified with what Halliday and Hasan refer to as the 'text-

forming component'). The rhetorical system establishes in a
text relations among sounds, among lexical items, among gram-
matical patterns, and among discourse units, which structure
a discourse and signal its organization and meaning.
Traditionally, the most commonly studied rhetorical devices
in addition to figures of speech have been those which involve
repetition of some sort and which establish cohesion, such as
alliteration and rhyme in the sound system and parallelism in
the grammatical system. The rhetorical system of a language,
more importantly, however, establishes relations of disjunction
as well as continuity or cohesion between elements in a text.
Repetition is only the most obvious device for establishing
continuity. Continuity can also be achieved through the sorts
of cohesive devices discussed by Halliday and Hasan, like pro-
nominalization, anaphora, and lexical implicature, as well as by
the lack of clitics and particles which have disjunctive discourse
function, the suppression of pauses and junctures at constituent
boundaries, and other grammatical and prosodic phenomena.
Disjunction can be signaled by pauses, junctures, phrase,
clause, and sentence boundaries, the use of substantives rather
than pronouns, the use of clitics and particles signaling larger
narrative units, and the presentation of material as an inde-
pendent sentence rather than as a dependent clause or phrase.
The cohesion in English which Halliday and Hasan discuss in-
volves devices for establishing continuity, but patterns of
continuity and cohesion take on significance only by contrast
with breaks in that continuity, signaled by disjunction of some
sort. Both cohesion and disjunction are equally important in
the shaping of a text.
Written forms of oral texts. In written discourse, rhetorical
organization is reflected by punctuation and indentation, while
in oral discourse, the actual oral delivery of the text, the
pauses, junctures, and pitch changes, combine with patterns
of grammatical and lexical cohesion and disjunction to organize
a text rhetorically into a coherent whole. This organization is
often missed or obscured when oral material i s , as it is often
too aptly put, 'reduced to writing'. An example of the tradi-
tional scholarly presentation of an oral text is given in
Figure 1. In part, this is probably because we have all been
trained to attend primarily to the phonology of words, although
we also know that sentences have characteristic intonational
patterns associated with them. However, we also lack easy to
use and read conventions for representing in writing the
rhetorical function of intonational phenomena.
Dell Hymes (1976, 1977, 1980a,b, 1981) has recently published
several stimulating attempts to recover the rhetorical organiza-
tion of Chinookan texts recorded by Sapir and Jacobs before
the easy availability of tape recorders which offer an alterna-
tive to the traditional text presentation in Figure 1. By close
 Figure 1. KU«N&'LA-B£J-CIKE AND HIS BROTHER, CHI-Mfew, AT K U - S A « - D A - N 6 « Y 6 W [Told by Ralph Holder, May, 1973. 00

to
1. yu xa na«pho-le, chi«Mewqay, ku-nu'la-bu'cikeqay, 1. They were living there, Wolf and Old man 0)

ma'du'xaciMak, ku'sa'-da-nccyow. Coyote, him and his brother, at ku'sa«-da«no'yow.

2. rni'n ^ikkiliday xa cni«Mew mi • pal ya-?o

qa*mukk*le; ba* khi qa-lal me«rkille, ma-^ay
2. Well, been long like that Wolf had a
toothache, then he lay there sick, he never
I
3. *[mil] ka«te* xa ku*nu«la- eat. 3. Then (Nevertheless) Old Man Coyote go
yo'-qa^qo^yal kakkllle, ma-'Jay to Big Valley to get some food; from there he
hi«p; bayawa k^i tu«nu, lami, mi-n f&'m ka.1- bring home them digger squirrel; from there he
k^i-dikkille. 4. bayawa kaluhun xa k^i ma-^ay bring home field mice and gopher. 4. When he
hi« kalk^i-di^ba* ca•rdu•bakin, qo*diy du-bakin, come home that food that he bring he clean ...

*Mr. Holder inserted when transcribing.

 Meaning, Structure, and Organization in Myth / 287

attention to the placement of discourse particles and clitics in

the text, and the structure and content of the sentences they
mark off, he attempts to reflect systematically what was proba-
bly the rhetorical structure intended by the narrators, utiliz-
ing the written conventions developed largely for poetry in
western European languages. These in fact reflect rhetorical
structure visually through placement in terms of lines and
verses on the page. William Bright (1979a,b, 1980) has re-
presented Karok texts similarly, but using the prosodic fea-
tures available on tape recordings of these texts, as has Joel
Sherzer, I understand. Anthony Woodbury (in press) has used
a line and verse format which reflects rhetorical structure indi-
cated in prosodic as well as syntactic features for a whole
series of Inuit oral narratives which he has converted to written
form for the Alaska Native Language Center.
Hymes has suggested that the success of the approach indi-
cates that at least some Native American narrative is poetry
rather than prose. It is not clear that it is useful to impose
such a distinction on the literature of societies that do not
traditionally make such a distinction. It also seems likely that
the communicative value of a line and verse presentation of
narrative material has more to do with the responses of audi-
ences literate in European languages to the format of poetry
than necessarily with the nature of the narrative itself. As
Hymes (personal communication) has said, 'it slows the eye
and hence feeds the mind.' 1
For the past five years, I have been experimenting with the
same use of written conventions to reflect the rhetorical struc-
ture and discourse organization signaled in the actual oral
delivery of texts in Eastern Pomo, a Hokan language of native
North America, spoken in northern California.
In 1973, I was fortunate to have the opportunity to begin
research with a remarkably talented speaker of Eastern Pomo,
Ralph Holder of Upper Lake, California. Mr. Holder was a
monolingual speaker of Eastern Pomo until he went to school
around the age of seven. From the age of six on, he lived
primarily with elderly grandparents who had been adults be-
fore serious white settlement began in this area. They appar-
ently preferred to use Eastern Pomo in all communicative con-
texts until their deaths when Mr. Holder was 16. Mr. Holder's
grandfather, Jim Bateman, was a master myth teller. He is, in
fact, mentioned by several Eastern Pomo in the turn of the cen-
tury fieldnotes of A. L. Kroeber as a person to be consulted
about myths and traditional precontact ways. Mr. Holder spent
his childhood and youth listening to his grandfather recount
both myths and his own experiences of childhood and youth,
a great many of which Mr. Holder has recorded in Eastern
Pomo with me over the past nine years (McLendon 1977a,b,
1978).
Continuing work with the 14 long, complex, and rich tradi-
tional narratives called ma-ru- 'myth' which Mr. Holder recorded
288 / Sally McLendon

in Eastern Pomo has revealed that a close analysis of pauses,

pitch, rise and fall of voice, relative speed or deliberateness
of delivery and the location of breath taking, in conjunction
with a consideration of the form and content of the utterances
associated with these intonational features, permit the nonsub-
jective recognition of both the rhetorical structure and the dis-
course organization of a text. This organization, once recog-
nized, can then be reflected visually in a line format such as
Hymes has developed.
This new presentational format has the additional advantage
of making it possible to use the narrator's original words in
both the native language and the translation versions of the
text. When the rhetorical structure is reflected visually, the
original words seem to take on heightened dramatic potential
and meaning in ways that seem highly appropriate from the
point of view of English literary standards. Thus, this line
format has the additional advantage of eliminating, or at least
making less pressing, the need (or temptation) to edit and
retell the English translation of a Native American myth in
order to make it more accessible to an English-speaking audi-
ence (which increasingly even the Eastern Pomo, themselves,
are). One can then simultaneously provide both a more
scholarly version and a more immediately available one, using
such a presentational format.
Features of the oral delivery of Eastern Pomo myths which
are useful in recognizing their rhetorical structure. In myths,
Eastern Pomo sentences must be in the hearsay evidential mode.
This means that the inflected verb of the main or matrix clause
must be suffixed with the hearsay suffix - • le, and that the
hearsay clitic xa must be postposed to the first constituent of
that clause (cf. McLendon 1975, 1979). Thus, in Sentence 1
in Figure 1, the main clause is:

yu xa n
perfective they dwell plurally-they say
say
The inflected verb is na-p^b'le 'dwell plurally', and the hear-
say clitic xa follows the initial constituent, yu.
In myths, Eastern Pomo sentences consist of one or more
phonological phrases. Phonological phrases are characterizable
as having one primary word stress associated with a significant
rise in pitch. Each phonological phrase has a pitch norm
throughout which reflects (1) the relative location of the phrase
within a sentence, (2) its degree and type of syntactic related-
ness to the surrounding phrases, and (3) the communicative
importance within the sentence of the semantic content of that
phrase.
Eastern Pomo sentences begin on a higher pitch than they
end. When a sentence is complex in syntactic structure and
 Meaning, Structure, and Organization in Myth / 289

includes several phonological phrases, successive phrases drift

down in pitch, rather like the phenomenon known as downstep
in many West African languages. However, when a sentence
contains one or more phonological phrases which are syntactic-
ally in apposition, the pitch of these phrases stays at the same
level.
Boundaries between phonological phrases and, ipso facto, be-
tween sentences, are signaled by perceptible pause or its lack
(suppression seems a more accurate term), plus changes in the
pitch of the phonological phrase on either side of the pause.
Nonsuppressed pauses are of two main types: (1) one in
which the breath is held back by audible glottal stricture, and
(2) one in which breath is audibly exhaled at the end of a
phrase and/or audibly inhaled at the beginning of the next.
Glottal stricture seems to draw attention to the cohesion in-
tended by the speaker between the two phonological phrases
it separates, while inhalation and exhalation mark a greater
disjunction. As one might expect, suppression of pause seems
to signal greater cohesion intended between the two phrases
linked by it, and so far has mainly been observed when the
two phrases are syntactically clauses in apposition. Not all
apposed clauses are linked by suppressed pauses, however.
Sentence boundaries in Eastern Porno myths, thus, are sig-
naled by a combination of syntactic and prosodic phenomena.
The recognition of sentence boundaries is indispensable to a
correct understanding of rhetorical structure but is far from a
mechanical process. A myth consists observationally of a suc-
cession of fully independent main clauses (having the hearsay
clitic xa plus the -'le suffix), separated by other material of
varying degrees of dependency, ranging from formally subordi-
nated to potentially independent (but marked as apposed by
the absence of the hearsay clitic xa). These structures are
grouped prosodically into phonological phrases by patterns of
pause and pitch change. One recognizes sentence boundaries
as much by what follows as by what precedes. A sentence
consistently begins on a higher pitch than that on which the
preceding sentence ended, and the beginnings of sentences are
marked syntactically by characteristic features. Discourse
particles are sentence initial, and adverbial particles occur
either sentence initially or as the second constituent of a sen-
tence.
The generalizations presented here about Eastern Porno sen-
tences describe the regularities which are observable when one
breaks a text into sentences in this way. Since these regu-
larities are the basis for recognizing sentence boundaries, the
process is unavoidably circular. The correctness of the sen-
tence boundaries determined in this way is supported by the
semantic cohesion which holds within each sentence and which
is greater than that which holds between adjacent sentences.
290 / Sally McLendon

The rhetorical organization of a portion of an Eastern Pomo

myth. The beginning portion of the Eastern Pomo text pre-
sented in Figure 1 in traditional run-on prose format is re-
presented in Figure 2. The prosodic structure of the actual
delivery is visually represented by presenting phonological
phrases as separate lines, except when two phrases are linked
by suppressed pause preceded by a fall in pitch.
The type of pause and pitch changes are indicated by the
following mnemonic abbreviations printed above the line.
EX Exhalation
IN Inhalation
CLD LIPS Closed Lips
GSTR Glottal Stricture
P Rise in pitch accompanying one primary stress
in a phrase
1/2P Flattened pitch rise
4- Drop in voice and pitch
-V
Steady pitch without change
<r Drop in pitch with suppression of pause
Traditional English punctuation--commas, periods, quotation
marks, etc.--are used to signal the sentential organization re-
flected by the interaction of syntax and prosody.
Within a sentence, successive phonological phrases closed by
EX and introduced by IN are, after the initial phrase in a sen-
tence, placed one directly below the other, as in Sentence 1 in
Figure 2. Phrases separated by glottal stricture are indented,
as in Sentence 2 in Figure 2. The second of two phrases sepa-
rated by suppressed pause which is preceded by a steady pitch
(-1-*-) is dropped one line below the first, but at the same point
at which the preceding phrase ends. Phrases with suppressed
pause plus a falling pitch are written on the same line, sepa-
rated by a comma.
It is easiest to follow the discussion if one has heard the
text. This is unfortunately not possible in a written format.
Without the oral performance, it is particularly important to
read through the text which follows using the line divisions,
diacritics, and punctuation to imagine what the text sounds
like.
The visual presentation of the text in terms of lines makes
the prosodic features easier to hear, even if one is unfamiliar
with the language. More importantly, it also makes clearer the
patterns of cohesion and disjunction which organize the material
rhetorically.
For example, in the first sentence of Figure 2, which sets
the stage, as first sentences always do in well-formed perform-
ances of Eastern Pomo myths, the protagonists are named, their
kin relations specified, and their residence at a real, known
site established. Each of these three types of information is
given in a distinct phonological phrase (the naming of the
 ta

Figure 2. 5'

1/2P +BX CLD LIPS

1. yu xa na-pho-lo, 1. They were living,
PERF they-say dwell-PL

1/2P +" P ^ ^ +
c"i-Mewqay, ku-nu-la-bik-cikeqay, Wolf and Old Man Coyote,
Wolf-and Coyote-Old Man-and

1/2IN P +£X
0)
ma • du•xaciMaA, him and his brother.
his-own-younger-brother-with a
1/2IN P +EX
ku'5a*-da-no-yow.
(village s i t e on Scott's Creek)

(higher pitch)
IN P ^ EX CLD LIPS
2. mi-n 'ikkiliday xa 2. Well, been long like that
like-that was-HAB-SR they-say

IN P -*GSTR
chi-Mew
Wolf

P EX
mi-pal ya^'o qa-muk^-le; had a toothache;
3p-PAT-MASC tooth vise-like-hold
JJV P v" 1/2? ^ +i
10
ba- k h i q a - l a l rae-rkille, ma-'ay qa-wa-lakNiy. then he lay there sick, he never eat.
then 3p-CL sick lay-HAB food eat-not
(higher pitch) ' }5
IN P _^
3. *[mll] ka-te- xa 1/2P •*• 3. Then (Nevertheless) ^-
[that previously beside they-say ku-nu-la-bu-cikehe?ml-p P i Old Man Coyote ,-
mentioned action] Coyote- Old Man-AG yo--qa-qo-yal kakkllle, go to Big . QJ
South-Valley-towards run-habitually Valley ~
P +BX
ma*'ay hi-p; to get some &
food collect food; ^
(higher pitch) 3
IN t p ? CL
bayawa k h i qu-Mar p 4-BX from there he O
there-from 3p-CL ground-squirrel kalk h i-dlkkille; bring home them digger squirrel; 3
homewards-bring-carrying-HAB
(no
IN) ^ ^ PAUSE GSTR
bayawa k h i : from there he
there-from 3p-CL
P ^EX
tu-nu,
fieldmouse
IN
lajni mi-n ''a-m P +EX CLD LJPS . bring home field mice and gopher,
gopher like-that things kalkhi-dlkkllle.
homewards-bring-carrying-HAB

(higher pitch)
IN P +EX
4. bayawa kaluhun xa k n i 4. When he come home
there-from homewards-go-CoR t h e y - s a y 3p-CL

IN P _^ GSTR
ma-'ay hi- kalk n i-di?ba. that food that he bring
food 3p-CoR homewards-bring-that
 P ^ + P tEX
ca-rdu-baton, qo-diy du-bakin,
a
he clean that, fix it good, 5
clean-make-CoR good-make-CoR

IN P GSTR (0
ma-tolqakilin xa khi then he cooks that
cook-cause-HAB-CoR they-say 3p-CL

ma<duxacal si-xakh-le. and gives It to his brother.

his-own-younger-brother-PAT give

(higher pitch)
IN GSTR
5. ba- xa ialki-ya-lkille: 5. Then he refused:
then they-say refuse-do-HAB 0)

IN P a
"ma•?ay ku•hu 'I don't feel like eating
food eat-to-not wi t&. P GSTR
lp-PAT feel ya-?6he? (0
tooth-specific
5.
qa-muqa-mu'lin. because my tooth ache. to
vise-like-hold-extended-CoR
o
(higher pitch)
IN P +Hf
ba- wi ma-boya." That (my jaw) is swell up.'
that lp-PAT swell-up

P +CLD LIPS GSTR

ni»n xa khi Ne-le, that's what he say.
that they-say 3p-CL say

P ^
•*(xa:) chi>Mewhe7ml-p.
Wolf-AG
(BREATH) 10
jr
(higher pitch)
IN ^ P +" Pf \
6. ba- xa yu ke*hel qa-wa-lkllle, (hX) ?in ka-ya ku-temkllle; 6. Then he eat alone, then they sleep;
then they-say-PERF alone eat-HAB CoR after sleep-PL-HAB

CLD LIPS GSTR

('SO
2
P +"P +CLD U P o
xa-?asa k h i yup h a k a k k l l l e , ba-ka k h i y e - h e - l l e . Next morning he run down (to Big Valley),
davm-because-SR 3p-CL again run-HAB that-only 3p-CL do that's all he been doing. 3
(higher pitch)
a
o
IN p •*
7. mi-n ka-nkh k h i ye-he-liday xa P •*• 7. Been that way for a long time
l i k e - t h a t long-time 3p-CL do-SR they-say mi-p ba- p i>qot h ke-le p and he had suspicion
he-AG that suspect ma-du-xacal, of his brother,
his-own-younger-brother-PAT

GSTR
5 h i,

c h i-n 'in mi-p ba. what he been doing

what CoR he-AG that

(higher pitch)
IN P P GSTR
ma-?ay qa-wa-lak^y me-rhe? never e a t , laying down
food eat-not l a y - s p e c i f i c

P +ET
qa-was ma'bo-. h i s jaw swell up.
jaw swell-up

(higher pitch)
IN P ^
8. "ha-' ba- ku-taba^e," 8. 'I'm gonna find o u t , '
I t h a t search-for-SOBJ

P f +EX CLD LIPS

ni-n xa k h i balk'le. that's what he's thinking to himself.
that they-say 3p-CL think-REFL
 a
3
5*
to

a
3
a
(higher pitch)
IS P \- P _^
O
-5
9. ba- xa k h i, xa-'aqan xa W P EX 9. Next morning he (Q
then they-say 3p-CL dawn-SR they-say 3p-CL du-weMi early in the morning 0)
early-in-the-morning 3

wadu-ke-le ^ starts to go
go-begin yo-qa-qbyal p lEX to Big Valley
South-Valley towards ma-'ay hi-p. to get some more
food collect food.

*Mr. Holder inserted when transcribing.

**Mr. Holder deleted when transcribing.

KEY: PL = plural; SR = switch-reference; CoR = co-reference; HAB = habitual; AG = agent; PAT = patient; p = person; MASC = masculine; CL = clitic; SUBJ
subjunctive; REFL = reflexive; PERF = perfective [Granunatical abbreviations]
to
IN = inhale; EX = exhale; GSTR = glottal stricture [Intonational abbreviations]
VI
296 / Sally McLendon

protagonists is actually in two phonological phrases linked by

suppressed pause). The termination of the phrase presenting
each of these three types of information ends with the same
falling intonation and exhalation. The second and third types
of information, the specification of kin relation and the identifi-
cation of residence, are introduced by partial inhalation. A
prosodic parallelism is thus established which reflects both the
equal weight assigned these three pieces of information and
their discreteness. It also underscores the syntactic parallel-
ism established by the fact that they are all three in apposition
to the main clause, yu xa na«p"6'ie.
Pause phenomena simultaneously separate and connect syntactic
material. The suppressed pause in line 2 of the first sentence
between chi-Mew-qay 'Wolf-and' and ku-nii-la-bw cike-qay 'Coy-
ote-Old Man-and', while separating them, also signals the
greater cohesion between the naming of the two protagonists in
comparison with the specification of kin ties and residence. The
exhalation following ku- nti' la-bh' cike-qay, by contrast, functions
disjunctively to signal the distinction between the naming of the
protagonists and the specification of their kin relations, but in
the larger context of the succession of phrases closed by ex-
halation, also creates cohesion. In the second sentence, the
pause with glottal stricture separating line 2 c^i'Mew 'Wolf from
the rest of its clause both establishes a disjunction which fore-
grounds 'Wolf and signals the grammatical and semantic cohe-
sion of 'Wolf with what follows.
In Eastern Porno myths, significant discourse units larger than
sentences are distinguished by the use of a small number of
sentence initial particles and phrases, very much like what
Hymes (1976, 1977, 1980a,b) has drawn attention to in Chonookan
and other Native American languages. There i s , in fact, an im-
portant disjunction between the first and second sentence in
this text which is signaled syntactically by the initial phrase in
Sentence 2: mi* n ^ikkiliday . . . 'Well, been long like that . . . '
At least two levels of discourse unit seem to be distinguished
by these elements. Smaller units, analogous to paragraphs, can
be marked off by bh'... 'Then . . . ' . Discourse sequences simi-
lar to episodes are marked off by mi- n plus a verb suffixed
with -iday, as in Sentence 2. These episodes show a consistent,
clear-cut, thematic cohesion, as in Sentences 2 through 6 in
this text. They can also include sequences distinguished by
ba* , as in Sentences 5 and 6.
As Hymes (1980a: 10-11) has also pointed out, it seems to
facilitate the appreciation of the text by a nonnative speaker if
the thematic cohesion within an episode is reflected by an epi-
sode title in the written version of the text. Figure 3 presents
the same portion of an Eastern Porno myth with lines represent-
ing phonological phrases separated by pauses^ (as described
earlier) and English titles added to suggest the focus of each
episode. Traditional English punctuation is used to signal
sentential organization, as in Figure 2.
 Meaning, Structure, and Organization in Myth / 297

In an earlier Georgetown University Round Table paper

(McLendon 1977a), I proposed that one must understand the
cultural presuppositions which a narrative presumes if one is
to appreciate fully the drama, point, and organization of a
narrative. Here the two named protagonists are specified to
be older and younger brother, and associated with two mammal
predators whose range of prey overlaps and who therefore are
potential competitors for the same food supply. They are
predators who also competed with the Eastern Porno in the pre-
contact period for some of the same prey--small mammals, fish
and fish eggs, birds and wildfowl and their eggs. Thus the
sibling relationship is associated with potentially competitive
animals who are similar in size, type, and habits.
In the second episode of this myth, which begins with Sen-
tence 2, Wolf is described as lying passively, sick with a tooth-
ache and not eating--in fact, refusing the quite delicious food
that his brother, Old Man Coyote, is described as going to con-
siderable trouble to acquire and prepare for him.
To refuse food is an unusual and deviant thing to do in
Eastern Porno society. It is impolite unless one is too sick to
eat, or ritually prepared for doctoring, gambling, or ceremonial
activity (all of which involve contact with supernatural forces).
When one is ritually prepared, one must abstain from eating a
number of foods, particularly meat and fat. If one refuses
food and is not sick, then members of the society usually
assume the refusal is motivated by exigencies of ritual prepared-
ness, and often become suspicious that the ritual preparedness
might be intended to bring them harm. The acceptance or re-
fusal of food is thus a highly charged issue in the traditional
society presupposed by the myths. Wolf, it will later be re-
vealed, is actually catching deer magically in a fish net but not
sharing his catch with Coyote Old Man (a reprehensible thing
to do) and seems likely, in fact, to be ritually prepared.
In this second episode, succinct though it seems, the narrator
takes the time to establish the desirability of the food being
offered by Coyote, and the care and attractiveness with which
it is being prepared, presumably to emphasize and highlight
Wolfs refusal to eat, since Wolfs refusal to eat the food which
Coyote offers is the point of this episode. Each of the five
sentences in the episode makes a distinct and indispensable
contribution to making this point through a distinct manipulation
of the rhetorical system.
Sentence 2 prepares for Wolfs refusal, and establishes Wolf
as of dominant interest in this episode, by naming Wolf with a
substantive, in a separate phonological phrase at the beginning
of the sentence, set off by the glottal stricture pause, rather
than referring to him with a pronoun or kinship term. Wolf's
passivity is emphasized by the fact that the noun c^i'Mew
'Wolf is the patient of the main clause verb qa-miik^-le. In
terms of informational content,
1
it would be equally acceptable
to make merkille 'lie (sick) the main verb, as in 'Wolf lay there
 Figure 3.
00

SETTING THE STAGE

CO
1. yu xa na^p^o-le, 1. They were living, 0)
c^i•Mewqay, ku•nu •la-bu•cikeqay, Wolf and Old Man Coyote,
ma-du'xaciMak, him and his brother,
ku•Sa--da•no•yow. at ku-sa*-da*no#yo.
o
WOLF'S TOOTHACHE

2. mi-n 'ikkiliday xa
jTa
2. Well, been long like that
chi-Mew Wolf
o
mi-pal ya-?6 qa-mukh-le; had a toothache;
ba* k^i qa-lal me-rkille, ma-^ay qa*walaknuy. then he lay there sick, he never eat.

3. *[mil] ka-te- xa 3. Then nevertheless

ku*nu-la-bu*cikeheTmi-p Old Man Coyote
yo'-qa-qo-yal kakkllle, go to Big Valley
ma* ?ay hi-p; to get some food;
bayawa kni qu'Mar from there he
kalkhi-dikkille; bring home them digger squirrel;
bayawa kni:: from there he
tu-nu,
lami mi-n ?a*ni bring home field mice and gopher.
kalk h i«dlkkllle.
4. bayawa kaluhun xa khn i 4. When he come home
ma-?ay hi- kalk i«dl?ba- that food that he bring
ca*rdu'bakin, qo«diy du«bakin, he clean that, fix it good,
ma-tolqakilin xa k n i then he cooks that
ma-duxacal si«xak n "le. and gives it to his brother.

5. ba« xa calki-ya-lkille: 5. Then he refused:

"ma-'ay ku«hu*ba?khuy f "I don't feel
wi ta. like eating
ya'ohe? because my tooth
qa-muqa-mu'lin.
ache.
ba« wi ma-boya." That (my jaw) is swell up."
ni«n xa khi Ne«le that's what he say,
**[xa:] c^ the Wolf.
 (0
0)
3
5*
in

6. bar xa yu Ice-hel qa-wa-lkllle, (hS) ?in ka-ya ku-£emkllle; 6. Then he eat alone, then they sleep;
h h h
xa--?asa k i yup a k a k k i l l e , ba-ka k i y e - h e - l l e . Next morning he run down, that's all he been doing.

OLD MAN COYOTE'S SUSPICIONS

7. mi-n ka-nk n k n i ye-he-liday xa 7. Been that way for a long time
mi-p ba* p^i-qot n ke-le and he had suspicion
ma*du*xacal,
what
on his brother,
o
p ba* what he been doing 0)
y q me«rhe? never eat, laying down 3
4a-was raa'bo-. his jaw swell up.

8. "ha- ba* ku-^aba'e, 8. 'I'm gonna find out,1

ni>n xa k n i ba that's what he's thinking to himself.

9. ba« xa k n i , xa«'aqan xa 9. Next morning he

du'weMi early in the morning
wadu-ke*le starts to go
yo'-qa-qoyal to Big Valley
to get some more food.

*Mr. Holder inserted when t r a n s c r i b i n g .

»*Mr. Holder deleted when t r a n s c r i b i n g . to
300 / Salty McLendon

sick, not eating, with a toothache'. However, this would have

required that Wolf be marked as the agent of the verb, and
would have displaced the emphasis from the toothache and Wolf's
passivity. The syntactic and prosodic organization of the in-
formation in Sentence 2 focuses on Wolf and his toothache, as
the result of which he lies sick and does not eat.
Sentence 3 juxtaposes and contrasts intensive activity on
Coyote Old Man's part with Wolf's passivity, compressing into a
single sentence Coyote Old Man's going hunting in Big Valley
(where he apparently has traps set), and his bringing home
tasty, tender small mammals to roast on the coals--digger
squirrels, field mice, and gophers.
Sentence 3 consists of three independent clauses: (1) Old
Man Coyote goes to Big Valley to get some food; (2) from there
he brings home digger squirrel; (3) from there he brings home
field mice and gopher. These clauses could have been presented
as independent sentences without a loss of information, but have
been combined into a single sentence presumably because
Coyote's hunting activities are not the topic of this episode, but
rather provide a background against which Wolf's subsequent
behavior will be all the more marked. To present these three
independent clauses as separate sentences would give them too
much rhetorical weight.
Similarly, ku- nu- la-bU'Cike-he^mi- p 'Coyote-Old Man-agent' is
not foregrounded in Sentence 3 with a glottal stricture pause as
Wolf was in Sentence 2. This is Wolfs episode, although Coyote
is doing all the running, and the different prosodic treatments
of the names of the two protagonists underscore this fact.
Sentence 4 shifts the focus to the food which Coyote brings
back. Greater semantic cohesion between Sentences 3 and 4
than between Sentences 2 and 3 is signaled by the parallelism
of bayawa 'from there' introducing three successive clauses.
Two of these are adposed to the main clause of Sentence 3,
while the third is the first constituent in the main clause of
Sentence 4. This intrasentence cohesion is reinforced by the
recurrence of the preverb kal 'homewards' preposed to the
verbs of all three clauses. However, the fact that the third
clause occurs in a separate sentence, and with its verb marked
as dependent (with the switch reference suffix - m ) , constitutes
a disjunction signaling a change of focus from the action of
collecting the game to the game itself, and its careful prepa-
ration for eating.
The third parallel clause is followed by the only nominalized
clause in this whole sample:

ma-9ay hi- kal-k h i-di- 9 ba- . . . 'that food that he bring . . . '
food 3pCo- homewards-bring-that
Ref.

This stands out both syntactically and prosodically as one must

assume it was meant to do, since it would have been equally
 Meaning, Structure, and Organization in Myth / 301

possible to convey the propositional content of this clause by a

clause embedded with the switch reference suffix -in, parallel
to the following three clauses on the next two lines, i . e . :
ma«9ay hi kal-k^i'din
food 3p Co- homewards-bring-Co Ref
Ref
giving the sequence:
From there he come home
he bring home food
he clean that, fix it good,
then he cooks that
and gives it to his brother.
Such a sequence would have left the focus on Coyote and his
activities, however. The use of a nominalized clause shifts the
focus to the food Coyote is bringing home and thereby builds a
transition to the next sentence, in which Wolf rejects this food,
which also heightens the impact of the rejection, keeping the
focus of attention firmly on Wolf and his behavior-
Sentence 5 presents Wolfs refusal of the carefully prepared
food, highlighted by a direct quote. The climax could be in-
tensified, although Mr. Holder did not chose to do so, by using
a dramatic voice for Wolf, one which would emphasize the humor-
ousness of the scene, since everyone knows that Wolf has a
small pestle in his mouth to simulate the appearance of a swollen
jaw caused by an abscessed tooth, which could make him talk
oddly.
Once again, it seems significant that only Wolf is given a
direct quote in this episode. In terms of information, it would
seem equally plausible to report a conversation between Wolf and
Coyote over the food. However, direct quotes always carry a
potential for heightened dramatic effect and limiting this poten-
tial to Wolf is another device for making Wolf dominant in this
episode. The fact that Coyote has direct quotes in the next
episode signals a shift in focus between these episodes.
Sentence 6 returns to Coyote again, compressing several inde-
pendent clauses into a single sentence. Once again, Coyote is
far more active than Wolf, but the compression of so many inde-
pendent clauses into a single sentence, with pairs of clauses
linked by suppressed pauses and with no explicit arguments ex-
pressed in the first two, deemphasizes this activity. Once
again, Coyote's activity is backgrounded, providing a coda to
the climax of Wolfs refusal, and a transition to the next episode
in which Coyote's suspicions are finally aroused. It also empha-
sizes Coyote's persistence in the praiseworthy and appropriate
behavior of hunting for his brother and sharing food with him,
even in the face of his brother's deviant and suspicious be-
havior .
302 / Sally McLendon

The next episode begins with Sentence 7, which opens with

the episode marker mi-n plus a verb suffixed with -iday.
Neither Coyote nor Wolf is named. It i s , in fact, not neces-
sary to re-identify the protagonists with nominals from episode
to episode, and this is the neutral, unmarked pattern. Wolf's
behavior is recapitulated with an interesting shift in emphasis.
The putative toothache is no longer mentioned. Instead, only
the features observable by Coyote are mentioned: the fact that
Wolf is always lying down, does not eat, and has a swollen jaw.
This episode is going to provide another explanation for this
cluster of observable traits, and the shift in the characteriza-
tion of Wolfs behavior prepares the listener for the new expla-
nation. Perhaps the most striking change from the preceding
episode, however, is that Coyote's activities are no longer com-
pressed several to a single sentence (compare Sentence 9 with
Sentence 3) and Coyote now has a direct quote.
Space does not permit a detailed examination of the interlock-
ing and overlapping patterns of cohesion and disjunction which
organize this entire text rhetorically. I hope, however, that
these few examples suggest how much meaning is conveyed by
this process, which a native speaker must be presumed to re-
spond to, at least subliminally, but which a written presenta-
tion of the traditional sort given in Figure 1 makes much more
obscure.
I think it also suggests that just as a text is a creative,
semantic unit and not merely the sum total of a string of sen-
tences, a sentence is also a creative construct, imposed by a
speaker on the syntactic and lexical resources of the language
by means of the rhetorical system, with a meaning at least
partially unique to its occurrence in that particular text.
In fact, a discourse involves a speaker imposing a point of
view and a frame of reference on events, thoughts, actions,
plots. Imposing a point of view involves choosing the signifi-
cant components of the experience, event, etc. to be described,
presenting them with sufficient communicative competence, back-
ground and contextualization to make it possible for one's
audience to share one's point of view. This process of selec-
tive emphasis and down-playing involves what the Prague School
(in the translation of Garvin 1964) has referred to as fore-
grounding and automatization. Sentences within a discourse are
creative constructs within which the speaker manipulates the
syntactic, lexical, and phonological resources of the language to
assign relative discourse prominence to the phenomena described.
This seems to be one of the means by which the individual
voices alluded to by Becker (this volume) are achieved.
I hope that these examples have also suggested the desirability
of basing syntactic description at least in part on the sorts of
sentences which naturally occur in discourse if one is ever to
understand fully syntactic structures and their role in language.
For example, in the portion of the text presented here, one can
see that apposition is one of the most frequently occurring
 Meaning, Structure, and Organization in Myth / 303

devices for combining clauses and phrases into sentences in

Eastern Pomo, even clauses which are morphologically also
marked as dependent, as in lines 3 and 4 of Sentence 4, which
translates 'He clean that, fix that good, cooks that ...* The
verb in each of these three clauses is suffixed with the switch
reference suffix -in, but the clauses are juxtaposed without an
intervening clitic xa, signalling that they are in apposition to
each other.
The single nominalized clause in this entire sample, which
also occurs in Sentence 4, stands out both syntactically and
prosodically, as discussed earlier. Nominalized clauses are
rare in Eastern Pomo and always seem to have a disjunctive,
topic-shifting rhetorical function, as in Sentence 4. Although
one can translate this nominalized clause with a relative clause
in English, it seems clearly to differ from English relative
clauses in function as well as structure. This difference would
be hard to recognize were one to study clause nominalization in
Eastern Pomo only through sentences elicited as translations of
English sentences with relative clauses. Such a technique, in
fact, usually elicits switch reference embedded adverbial
clauses rather than nominalized clauses, perhaps because the
topic-shifting function of nominalized clauses is hard to make
clear without a discourse context.
The rhetorical system of a language, then, is not simply a
matter of style or optional choices. It is the communicative
principle which organizes discourse, determining sentence
boundaries, paragraphs, and episodes. It is the means through
which speakers impose a point of view on what they say. When
systematically reflected in the written form of oral texts, it
makes transparent the organization of a discourse and the art
inherent in all literature, but rarely studied if the language is
unwritten.
NOTES
It is a pleasure to acknowledge that the research on which
this paper draws was supported at various periods by NIMH
grant R01 MH 22887, a Guggenheim Fellowship, a City Univer-
sity of New York Faculty Research Award, and a Fellowship
from the Smithsonian Institution. Versions of this paper were
also read at the Conference on American Indian Languages at
the 1980 American Anthropological Association Meetings in
Washington, B.C., the March meeting of the Albert Gallatin
Philological Society, and the 1981 Hokan-Penutian Conference
held at Sonoma State University, Rohnert Park, California; I am
grateful for the comments and discussions provided on these
occasions. This paper has also profited a great deal from dis-
cussions with Anthony Woodbury, as well as with William Bright
and Joel Sherzer. My greatest debt is to Dell Hymes, who
both through his own pioneering research and in conversation
301 / Sally McLendon

and correspondence, encouraged and urged me to attend to the

meaning in narrative form.
1. Erickson and Scollon in their papers in this volume draw
attention to the metrical organization of talk. The metrical
structure which Dell Hymes so convincingly finds may reflect
the fact that talk is metrical as much as that Native American
oral literature is poetry.
REFERENCES

Bright, William. 1979a. A Karok myth in 'measured verse':

The translation of a performance. Journal of California and
Great Basin Anthropology 1.117-123.
Bright, William. 1979b. Coyote gives salmon and acorns to
humans. In: Coyote Stories II. Edited by Martha Kendall.
International Journal of American Linguistics, Native Ameri-
can Texts Series, Monographs. Ann Arbor, Mich.: Uni-
versity Microfilms.
Bright, William. 1980. Coyote's journey. American Indian
Culture and Research Journal 4(1-2). 21-48.
Garvin, Paul. 1964. A Prague School reader on esthetics,
literary structure, and style. Washington, D.C.: George-
town University Press.
Halliday, M. A. K. , and Ruqaiya Hasan. 1976. Cohesion in
English. London: Longman.
Hymes, Dell. 1976. Louis Simpson's 'The deserted boy'.
Poetics 5.119-155.
Hymes, Dell. 1977. Discovering oral performance and mea-
sured verse in American Indian narrative. New Literary His-
tory 8.431-457.
Hymes, Dell. 1980a. Tonkawa poetics: John Rush Buffalo's
'Coyote and Eagle's daughter'. In: On linguistic anthro-
pology: Essays in honor of Harry Hoijer. Edited by Jacques
Maquet. Malibu, Calif.: Undena Publications for the UCLA
Department of Anthropology. 33-88.
Hymes, Dell. 1980b. Particle, pause and pattern in American
Indian narrative verse. American Indian Culture and Re-
search Journal 4(4).7-51.
Hymes, Dell. 1981. In vain I tried to tell you. Studies in
Native American Literature, 1. Philadelphia: University of
Pennsylvania Press.
McLendon, Sally. 1975. The Eastern Porno language. Uni-
versity of California Publications in Linguistics, 74. Berkeley:
University of California Press.
McLendon, Sally- 1977a. Cultural presuppositions and dis-
course analysis: Patterns of presupposition and assertion of
information in Eastern Porno and Russian narrative. In:
Georgetown University Round Table on Languages and Lin-
guistics 1977. Edited by Muriel Saville-Troike. Washington,
D.C.: Georgetown University Press. 153-190.
 Meaning, Structure, and Organization in Myth / 305

McLendon, Sally. 1977b. Bear kills her own daughter-in-law,

Deer. In: Northern Californian texts. Edited by Shirley
Silver and Victor Golla. International Journal of American
Linguistics, Native American Texts Series 3.26-65.
McLendon, Sally. 1978. Coyote and the ground squirrels.
In: Coyote stories. Edited by William Bright. International
Journal of American Linguistics, Native American Text Series,
Monograph 1.87-111. Ann Arbor, Mich.: University Micro-
films.
McLendon, Sally. 1979. Clitics, clauses, closure and dis-
course in Eastern Porno. In: Proceedings of the Fifth
Annual Meeting of the Berkeley Linguistics Society. 637-646.
Berkeley, Calif.: Berkeley Linguistics Society.
Tedlock, Dennis. 1971. On the translation of style in oral
narrative. Journal of American Folklore 84.114-133.
Tedlock, Dennis. 1972. Pueblo literature: Style and
verisimilitude. In: New perspectives on the Pueblos.
Edited by Alfonso Ortiz. Albuquerque: University of New
Mexico Press. 222-224.
Tedlock, Dennis. 1977. Towards an oral poetics. New Literary
History 8.507-519.
Tedlock, Dennis. 1978. Finding the center: Narrative poetry
of the Zuni Indians. From performances in the Zuni by
Andrew Peynetsa and Walter Sanchez. Lincoln: University
of Nebraska Press. New preface. (First published, New
York: Dial Press, 1972).
Tedlock, Dennis. 1979. Beyond logocentrism: Trace and
voice among the Quiche Maya. Boundary 2.8.321-333.
Woodbury, Anthony, (in press) Narratives, stories and tales
from Chevak, Alaska. Fairbanks: Alaska Native Language
Center, University of Alaska.
Woodbury, Anthony. (1980). Rhetorical structure in Central
Yup'ik Eskimo narrative. Paper presented at Winter Meetings
of the Linguistic Society of America, San Antonio, Texas.
 THE INTERPLAY OF STRUCTURE AND FUNCTION
IN KUNA NARRATIVE, OR:
HOW TO CRAB A SNAKE IN THE DARIEN

Joel Sherzer
University of Texas at Austin

This paper is intended as a contribution to a discourse-

centered approach to the study of the language-culture-society
.relationship. One aspect of the recent interest on the part of
several disciplines in the detailed and precise analysis of dis-
course is a focus on structure and style in such a way that it
seems difficult if not incorrect to make a distinction between
ordinary language on the one hand, and literary and poetic
language on the other. All discourse has features that have
characteristically and traditionally been considered to be liter-
ary, and analysis of poetic structure is often what discourse
analysis is all about. So rather than shove off the study of
metaphor, foregrounding, cohesion, line and verse structure,
dramatization, and grammatical aspects of style on literary
critics, we find that attention to such matters is basic to the
work of linguists, anthropologists, and folklorists.
In fact, as has recently been stressed in a series of papers
by Dell Hymes, analysis of the poetic organization of discourse,
especially narrative, is a logical continuation of the Boas,
Sapir, Whorf tradition in anthropology and linguistics. The
concern is not with the relationship between grammar, con-
ceived in a narrow abstract sense, and thought, as one limit-
ing and probably dead-ended interpretation of the Sapir-Whorf
hypothesis would have it, but rather with the poetic and
rhetorical organization of discourse as an expression and actuali-
zation of the intimate intersection of language and culture. As
but six of many quite distinct manifestations of this trend to-
ward consideration of poetic, literary aspects of and approaches
to discourse as central to the study of language and speech,
we can point to Bauman's (1977) focus on culturally and socially
situated performance as the locus of verbal art; Friedrich's

306
 Interplay of Structure and Function in Kuna Narrative / 307

(1979) studies of the symbolic and poetic potentialities of gram-

mar; Gumperz' (1971) studies of code-switching and contextuali-
zation, in which such poetic processes as foregrounding and
metaphor are shown to be at the heart of everyday communica-
tion; Hymes' (1977, in press) grammatical/cultural/rhetorical
analyses of North American Indian narrative, in which it is
argued that narrative is central to the creative expression of
what culture is all about; Labov's (1972b) analysis of the struc-
ture of the quite literary personal narratives that occur within
everyday conversational interaction; and the work of Tedlock
(1978) in the analysis and especially the translation of American
Indian performance style. A somewhat related approach is
Geertz' (1973) thick description of cultural texts, which, while
not texts in the sense that the term text is usually used by
linguists, are approached in a literary way.
In addition, my approach in this paper involves another im-
portant aspect of the analysis of discourse--attention to and
indeed focus on the intersection and interplay between struc-
ture and function. I examine in some detail a single narrative
in use among the Kuna Indians of Panama, a magical chant used
to grab a dangerous snake and raise it in the air. An investi-
gation of this chant from phonological details to overall narra-
tive organization reveals a constant and dynamic interplay and
intersection of structure and function. Attention to this inter-
play is crucial to an understanding of each of the devices in
the text, the meaning of the text as a whole, the role of the
text in the event in which it occurs, and the significance of
the text in Kuna culture and society more generally. It is not
my purpose here to t r y to clarify or elaborate on the different
meanings and uses that have been given to the concept of
function. My use of this concept is in keeping, in a general
sense, with the way it has been used by the various research-
ers who take a multifunctional approach to the study of lan-
guage u s e . x
The Kuna are a society of more than 25,000 agriculturalists,
most of whom inhabit a string of islands along the northeast
coast of Panama, known as San Bias. While there are literate
Kuna, their discourse is essentially oral. The Kuna have a
rich, complex, and varied system of language and speech, which
can be viewed in relation to Kuna politics, religion, curing,
magic, and puberty practices, but which can also be studied
in and for itself, in terms of textual, discourse, and literary
properties. It is important to stress that there is no single
feature which characterizes Kuna discourse as a whole, but
that there is rather a set or complex of such characteristic
features. Furthermore, it is not possible to use these fea-
tures to identify Kuna language and speech diagnostically as
oral rather than as written. While some of these features may
turn out to be more characteristic of oral than of written lan-
guage and speech, others are clearly found in written discourse.
And much more cross-cultural research is needed before it can
308 / Joel Sherzer

be determined if there is a set of features which uniquely

characterize oral language, speech, literature, and culture.
The text I discuss here is one of the many quite diverse exam-
ples of Kuna discourse.
I am going to set the scene for my discussion with a very
brief overview of the types of Kuna narrative. One useful way
to classify Kuna narratives is as either first person narratives
or third person narratives. First person narratives can in
turn be grouped into, on the one hand, new information nar-
ratives, characteristic of informal occasions and especially in-
formal conversations and greetings, and, on the other hand,
retellings, in which personal experiences are retold in a formal
and often ritual language and style, characteristic of the speak-
ing and chanting of the Kuna gathering house, the public,
political, and social meeting place. Third person narratives
are also quite common and appreciated among the Kuna. They
are the form used for the performance, in a ritual language,
of tribal traditions, myths, legends, history, and stories.
Public performances of these tribal traditions in third person
narrative form are typically used as models of good or bad
behavior by narrators. They are part of the exhortative
rhetoric of Kuna public politics and oratory. In a certain
sense, then, these third person narratives are understood to
be embedded within a first person form: rI exhort you to do
X.' In addition, one of the salient devices of third person and,
indeed, all narration among the Kuna is the incorporation of
direct quotations, including reported conversations, into the
narration. This is so common in Kuna speech that the actual
speaker is often not the main narrating voice of the narrative
he is telling. In these dramatizations, first person narratives
are embedded into third person narratives. This interplay of
first and third person narrative, including especially the em-
bedding of narrator within narrator, is a basic characteristic
of Kuna discourse (see Sherzer in press).
Another type of third person narrative is the magical ikar
'way' or 'text' which is the central and crucial act in Kuna
curing and magical events. Magical ikars are memorized chants
in the esoteric ritual language of the spirit world. 2 They are
performed by specialists, wisits 'knowers' of them, and ad-
dressed to particular representatives of the spirit world. Kuna
magic operates in the following way. The spirit addressees of
a magical ikar are convinced by means of the ikar that the per-
forming specialist is able to control them, that he knows their
language and every aspect of their essence and existence--the
location and nature of their origin, their present abode, their
physical and behavioral characteristics, and their names. After
the demonstration of this potential to control, the ikar describes
precisely what action the specialist wants to occur--for exam-
ple, the curing of a disease or the grabbing and raising of a
snake. Ideally, the spirits, upon hearing and understanding
the narrative, and because of hearing and understanding the
 Interplay of Structure and Function in Kuna Narrative / 309

narrative, do what is described in it. Kuna magical texts are

thus dramatic scripts for action performed by a live narrator
and played out by spirit actors. Because the spirit world
underlies and animates the real, actual physical world, what
occurs in the spirit world is subsequently played out in the
actual world. In this sense, the third person narratives of
Kuna magic can be understood to be embedded within a first
person 1 command of the form: 'I tell you, that i s , I command
you X. Magical ikars are performative in the sense that their
performance is essential to the successful completion of a magi-
cal action; saying is doing in that correct narration not only
describes an action or event, but actually accomplishes it by
causing it to occur. Kuna magic contains no hocus pocus or
abracadabra. ** Rather, it is based on a highly intelligible lan-
guage. This language is intelligible in two senses. First, al-
though it is not understandable to most human nonspedalists,
it is completely and necessarily understandable to the spirit
addressees. Second, it is analyzable, in the sense of linguis-
tic analysis, phoneme by phoneme, morpheme by morpheme,
word by word, and line by line.
I turn now to an exploration of one magical ikar, nakpe ikar
'the way of the snake', used to grab and raise a dangerous
snake on the performing specialist's arm, in terms of the inter-
play of structure and function. This chant is one of a set of
magical ikars known as kaeti, literally 'grabber'. Other ikars
in this set are used to attract bees and wasps and to grasp a
hot iron rod. Kaetis, like other magical ikars, are performed
in a variety of contexts. They are performed in the actual act
of control—for example, raising the snake. And they are per-
formed for practice, for learning and teaching, for pleasure,
and to remind the ever present and listening spirits that the
specialist has the potential to control them. The public, verbal
display of knowledge and potential power, to both the human
and spirit worlds, is as important as, if not more important
than, the physical act of grabbing the snake. The particular
performance of nakpe ikar that I discuss here was a practice
session by a specialist-knower.
Why snakes? The Darien jungle in which the Kuna walk,
farm, and hunt is the site of some of the most dangerous
snakes in the world. This is reason enough to be worried
about snakes and to want to be able to control them. In addi-
tion, according to Kuna belief, snake spirits are among the
most evil of spirits and, like other animal spirits, have the
potential to cause serious disease, without the actual, physical
animal necessarily biting or even coming into contact with the
victim. 5 It is no wonder that, in addition to nakpe ikar, there
is a complex of magical chants whose purpose is to calm and
control various snakes and snake spirits. Let us now examine
the text of nakpe ikar itself.
The performance of nakpe ikar that I recorded lasts eight
minutes. It opens with the setting of the scene: as the
310 / Joel Sherzer

6
specialist is working in his jungle farm, the snake appears,
The specialist [is] at the edge of his field.
The specialist is surveying his farm.
When the sun is halfway up in the sky.
The specialist is surveying his farm.
At the edge of his field.
The specialist is sharpening his little knife.
He is sharpening his little knife.
With a file.
When he finishes sharpening his little knife.
When he finishes sharpening his knife.
Then the specialist moves.
Then the specialist advances.
When the sun is halfway up in the sky.
The specialist is working with his little knife.
He is working with his little knife.
He is cutting small bushes.
He is clearing small bushes.
As he is cutting small bushes.
As he is clearing small bushes.
Machi oloaktikunappi nele [the snake's spirit name] 7 is present.
The snake is described.
Machi oloaktikunappi nele raises his chin.
His chin seems white.
Under the grass cuttings.
Machi oloaktikunappi nele sticks out the point of his tongue.
He sticks out the point of his tongue.
It looks like the dark blue of the koka plant dye.
The point of his tongue salivates.
Machi oloaktikunappi nele is present.
Indeed Machi oloaktikunappi nele is present.
In his abode under the grass cuttings.
The snake verbally challenges the specialist.
Machi oloaktikunappi nele calls.
'How well do you know the abode of my origin?'
Machi oloaktikunappi calls.
 Interplay of Structure and Function in Kuna Narrative / 311

And the specialist responds to the challenge.

The specialist counsels Machi oloaktikunappi.
'Indeed [I] know the abode of your origin.
Indeed [I] have come to play in the abode of your origin.
Indeed [I] have come to encircle the abode of your origin.'
He counsels Machi oloaktikunappi.
The snake prepares himself for the contest.
Machi oloaktikunappi nele.
He prepares his silver hooks.
He moves his silver hooks across his mouth.
He moves his silver hooks up and down.
Machi oloaktikunappi nele is present.
The specialist in turn prepares himself by applying special
medicines.
Indeed the specialist fortifies his purpa [soul].
Indeed he augments his purpa.
He gives nika [strength] to his hand.
He puts a lake of medicine on his hand.
The specialist competes with Machi oloaktikunappi nele.
He is counseling Machi oloaktikunappi nele.
The specialist is calling to Puna olotuktutili [name of bland
medicine].
He is calling to Puna olotuktutili.
He is calling to Oloputi nolomakke tule [name of medicine which
renders blow gun weak].
He is calling to Oloputi nupyasae tule [name of medicine which
causes blow gun to double u p ] .
Indeed the specialist is ready for Machi oloaktikunappi nele.
Indeed his purpa [soul] is augmented.
Indeed his purpa is strong.
He competes with Machi oloaktikunappi nele.
Then comes the verbal display of power. The specialist
shows that he knows the snake initimately by listing the
parts of the snake's body in a series of lines.
He is counseling Machi oloaktikunappi nele.
Indeed Machi oloaktikunappi nele is present.
312 / Joel Sherzer

'The specialist knows well your purpa [soul]'

The specialist is saying.
'He captures your purpa.'
The specialist is saying.
'Indeed how your lips were placed on.
The specialist knows well.'
The specialist is saying.
'How your chin was put in place.
How your lower chin was formed.
The specialist knows well.'
The specialist is saying.
Indeed the specialist is saying.
'How your pupils were formed.
The specialist knows well.'
The specialist is saying.
The specialist is saying.
'How the point of your tongue was put in place.
The specialist knows well.'
The specialist is saying.
He counsels Machi oloaktikunappi nele.
Indeed the specialist.
'How your golden arrow was put in place.
How your golden arrow was buried in.
The specialist knows well.1
The specialist is saying.
He counsels Machi oloaktikunappi nele.
Indeed he counsels Machi oloaktikunappi nele.
Indeed Machi oloaktikunappi nele.
'How your necktie was hung on.
The specialist knows well.'
The specialist is saying.
He counsels Machi oloaktikunappi nele.
Indeed the specialist.
'How the venom of your golden arrow was put in place.
The specialist knows well.'
The specialist is saying.
He counsels Machi oloaktikunappi nele.
 Interplay of Structure and Function in Kuna Narrative / 313

The specialist.
'How your flat head was formed.
The specialist knows well.'
The specialist is saying.
Indeed the specialist.
The specialist is saying.
'How your spinal cord was put in place.
How your spinal cord was made flexible.
The specialist knows well.'
Then there is a description of the desired magical action
itself, the raising of the snake.
Machi oloaktikunappi is under the grass cuttings.
The vine [euphemism for snake] is dragging [in horizontal
position].
The vine is turning over [in horizontal position].
The specialist is signalling toward his hand.
Toward his hand.
The vine has almost arrived, almost arrived.
He wags his golden blow gun.
Indeed on the specialist's hand.
The uttering of the performative formula:
"'Simply I raise you" I am saying.'
And the act is done.
He counsels Machi oloaktikunappi.
On his hand.
The vine is dragging [in hanging position].
The vine is turning over [in hanging position].
The snake admits defeat.
Machi oloaktikunappi calls.
'My specialist, [you] know well my purpa [soul]' he says.
Machi oloaktikunappi calls.
And he expresses his fear.
Indeed Machi oloaktikunappi is calling [in hanging position].
'My specialist, what will [you] do to me, will [you] kill me?'
Machi oloaktikunappi is calling [in hanging position].
311 / Joel Sherzer

Having won the contest and controlled the snake, the special-
ist shows himself to be friendly and compassionate.
Indeed the specialist counsels Machi oloaktikunappi.
'How indeed could [I] kill you? We have just become good
friends.
How indeed could [I] kill you?'
He counsels Machi oloaktikunappi.
This is the text of nakpe ikar. I turn now to an examina-
tion of the various devices which are used to structure this
text. Nakpe ikar is in a linguistic variety and style which is
shared by Kuna magical specialists and the spirit world and
differs from everyday colloquial Kuna along several dimensions.
The most salient and diagnostic phonological characteristic of
this magical language is that many vowels which are deleted in
everyday, colloquial speech are not deleted in the ritual chant-
ing of magical ikars. As a result, various consonantal assimi-
lation processes that automatically follow vowel deletion do not
occur in these chants (see Sherzer 1973). Thus palitak-
kekwichiye 'he is surveying' in nakpe ikar would be partayk-
wisye in colloquial Kuna; osamakkenaiye 'he is clearing' would
be osamaynaye; and sokekwichiye 'he is saying' would be
sokkwisye. Both the presence of these underlying vowels
(from a generative point of view) and the melodic patterning
of the chanting contribute to the phonological marking, in a
sociolinguistic sense, of this ritual magical variety and style.
They are also an important aspect of the esthetic, verbally
artistic quality of Kuna magical texts. In addition, melodic
shapes contribute to the marking of the poetic line structure
of the text.
With regard to morpho-syntactic structure, nakpe ikar, like
all magical chants, is marked by the use of a particular set of
nominal and verbal prefixes and suffixes. These forms have
several functions, which operate simultaneously. They are
part of the structural apparatus of the grammar of the magical
linguistic variety, serving as nominalizers, stem formatives,
and tense-aspect markers. They are sociolinguistic markers of
this particular linguistic variety, distinguishing it from other
Kuna linguistic varieties and styles. They contribute to the
esthetics of magical texts in three ways--they increase the
length of words, especially in terms of the number of morphemes
per word; they are ornamental embellishments; and they are
one of the devices used to mark poetic lines.
There are also morphemes which, while they occur in col-
loquial Kuna, have a greater frequency and a different and
wider range of meanings and functions in magical chants such
as nakpe ikar. An excellent example is the suffix -ye, which
is used in colloquial Kuna as an optative and emphatic with
verbs and an emphatic and vocative with nouns. It is also
 Interplay of Structure and Function in Kuna Narrative / 315

used as a quotative marker. It occurs with great frequency in

the language of magic, perhaps stressing the optative mood of
magical chants. But it is also a place filler, giving the per-
former time to remember the next line of these memorized
chants. -Ye can be viewed as a verbally artistic embellisher
as well; it is sometimes repeated two or three times. And,
since it often occurs at the ends of lines, it serves, along with
other devices, as a poetic line marker.
Morphological structure is directly involved in the magical
functioning of nakpe ikar. A set of four optional verbal suf-
fixes can be used in Kuna to specify the position of the sub-
ject of the verb, as either -kwichi 'standing', sii 'sitting',
-mai 'lying' or 'horizontal', or -nai 'hanging, as in the air'.
This optional grammatical category is crucial to the climactic
moment of nakpe ikar, the actual raising of the snake in the
air. In this section of the text, the snake is first described
as dragging and turning over in a -mai 'horizontal' position,
that i s , free on the ground. After the performative formula,
'"Simply I raise you" I am saying,' during which the snake is
raised in the air, it is again described as dragging and turn-
ing over, but this time in a -nai 'hanging' position. That is,
while the text never explicitly and specifically states that the
specialist has actually succeeded in grabbing and raising the
snake, the simple, economic shift in verbal suffixes, from -mai
'horizontal' to -nai 'hanging', on the same pair of verbs, drag-
ging and turning over, quite poetically and powerfully signals
that the snake is in the air, on the specialist's hand.
The process of moving from -mai to -nai involves the pro-
jection of a paradigm syntagmatically, the classic Jakobsonian
definition of poetry. And this occurs as the magical, powerful
climax of the text, addressed to the spirit of the snake itself,
and thus precisely convincing it that it has been controlled,
grabbed, and raised, and causing all of this to occur in actu-
ality. This is a true case of poetry in action. 8 The poetic-
magical potential of Kuna grammatical structure is actualized in
these crucial lines of nakpe ikar, 9in which grammar becomes
poetry and poetry becomes magic. A number of scholars have
recently argued that a most fruitful way to conceive of the
Sapir-Whorf hypothesis is not in terms of a matching up of
language and culture as abstract separate entities, but rather
as a dynamic, integrated actualization of language in culturally
meaningful and socially situated discourse (see especially Fried-
rich 1979 and Hymes in press.) The actualization of the poetic-
magical potential of the Kuna suffixes of position in nakpe ikar
provides an excellent example.
Another salient aspect of nakpe ikar, which must be ap-
proached in terms of the interplay of structure and function,
is lexicon. Like all magical chants, nakpe ikar uses the
vocabulary of the linguistic variety particular to the spirit
world. In fact, this vocabulary is the most diagnostic marker
of this linguistic variety. Many words in nakpe ikar are
316 / Joel Sherzer

entirely different from the corresponding words in colloquial

Kuna 1and have no meaning at all in colloquial Kuna. Thus,
'knife is esa in colloquial Kuna and ipetintuli in nakpe ikar;
'kill' is opurkwe in colloquial Kuna and kunnukke in nakpe
ikar; 'small' is pippikwa in colloquial Kuna and totokkwa in
nakpe ikar. In addition, there are words in nakpe ikar which
have a different meaning in colloquial Kuna, resulting in a
figurative, metaphorical effect. Thus, the fangs of the snake
are described as mananswelu 'silver hooks', olosiku 'golden
arrow', or oloputi 'golden blowgun'; and the stripe along the
snake's body is described as his mussue tukku 'necktie'.
While this comparison of colloquial Kuna and spirit Kuna seems
to reveal a metaphorical structure in the lexicon of nakpe ikar,
according to Kuna belief, personal, creative metaphor is not in-
volved here, although such metaphor is highly developed in
other styles of Kuna discourse, most notably in political oratory.
Rather, in the spirit world, snakes, like humans, have arrows,
blow guns, and neckties. This knowledge, and especially its
expression in the text of nakpe ikar, is an important aspect of
the specialist's verbal demonstration to the snake spirit that he
knows all there is to know about snakes, and especially, snake
spirits. In addition, the associated metaphorical effect, like all
the other poetic features in the chant, is pleasing to and ap-
preciated by the spirit world and plays a significant role in the
magical, controlling power of the text.
Related to the lexical structure of nakpe ikar and also essen-
tial to its magical power is the use of names. A crucial element
in controlling an object is knowing its spirit name. Constantly
labelling the snake Machi oloaktikunappi nele, its spirit name,
is another aspect of the specialist's demonstration to the spirit
that he has intimate knowledge of it and can thus control it.
It is interesting that at the climactic moment of the raising of
the snake, the snake is no longer labelled Machi oloaktikunappi
nele, but rather by the colloquial metaphorical euphemism kali
'vine'. This is an intriguing insertion of everyday language
into a magical text. The names of the medicines used by the
specialist to protect himself against the snake involve still an-
other type of verbal magical power. Like the snake's name,
they are the medicines' spirit names and knowledge of them is
essential to being able to control and thus use them. But in
addition, the text, by means of the creation of names, endows
the spirit medicines with properties which are encoded in these
names. Thus, oloputi nolomakke tule (literally: 'weak blowgun
medicine') renders the snake's fangs ineffective and oloputi
nupyasae tule (literally: 'double up blowgun medicine') takes
the strength out of the fangs. To describe an action in the
spirit language causes that action to occur; to name an object
causes the object to exist and to have the properties encoded
in the name.
Another feature of the vocabulary of nakpe ikar, which is
very characteristic of Kuna magical chants in particular and
 Interplay of Structure and Function in Kuna Narrative / 317

all Kuna discourse more generally, is a reflexive and meta-

communicative focus and orientation. Nakpe ikar is constantly
pointing inward to itself, situating itself as a communicative
event, and specifying what is happening within this event at
the very moment that it occurs. In particular, the text is
literally punctuated by the verbs uanae 'counsel' and soke
'say', as the performer-specialist insistently informs the snake's
spirit that he is counseling it and telling it to do certain things.
In addition to their intersecting referential and metacommunica-
tive function, these two verbs, uanae and soke, occurring as
the last word in a line, contribute to the formal marking of
poetic lines and structured groups of lines. And they are also
place fillers and holders, giving the performer time to think of
his next memorized line.
I turn now to a pervasive feature of the structural organiza-
tion of nakpe ikar and indeed of all Kuna magical chants, which
is also extremely common in both oral and written ritual and
poetic discourse around the world--syntactic and semantic
parallelism. There are various types of parallelism operating
in nakpe ikar.
There are certain crucial lines which are repeated identically,
or almost identically, 10 throughout the text, punctuating it by
marking the boundaries of sections within it. Examples are the
lines: Machi oloaktikunappi is present; the specialist counsels
the snake; the specialist is saying; the specialist knows well.
Adjacent lines are linked by several types of parallelism.
Two lines are identical, with the exception of the deletion of
a single word:
Machi oloaktikunappi nele sticks out the point of his tongue.
sticks out the point of his tongue.
Two lines differ in nonreferential morphemes (in addition to
the possible deletion of a word):
The specialist is sharpening (nuptulu-makke-kwichiye) his
little knife.
is sharpening (nuptulu-sae-kwichiye) his
little knife.
in which the verb stem formative -makke of the first line is
replaced by the verb stem formative -sae of the next line.
Two lines are identical except for the replacement of a single
word, the two words being slightly different in meaning and
within the same semantic field:
He is cutting small bushes.
He is clearing small bushes.
The specialist moves.
The specialist advances.
318 / Joel Sherzer

Another parallelistic pattern involves not single pairs of

lines, as in the preceding examples, but rather an entire set
of lines, a stanza-like frame which is repeated, each time with
a change in the word used to fill a particular slot. In the long
section of the text in which the specialist demonstrates his inti-
mate knowledge of the parts of the snake's body, the following
frame is repeated:
'How your [body part] was formed in its place.
The specialist knows well.'
The specialist is saying.
In this way, all of the body parts of the snake are listed.
It seems worth noting here, since one of the themes of this
Georgetown University Round Table is the differences between
oral and written discourse, and since Goody (1977) and others
have pointed to the list as a characteristic of written discourse,
that this example of the use of parallelism to perform orally a
list of items is but one of the many such cases in Kuna and
other nonliterate societies. In fact, one of the functions of
this kind of frame-parallelism in oral discourse seems1 1 to be
precisely the memorization and performance of l i s t s .
Parallelism thus serves a set of intersecting and overlapping
functions in nakpe ikar. It often involves the syntagmatic
projection of a paradigm or taxonomy (of body parts, medi-
cines, or movements). In addition to its poetic function, this
process of projecting taxonomies onto a fixed line, verse, or
stanza enables the generation of a long text or portion of text.
Length is an important aspect of the magical power of chants
like nakpe ikar. The more recalcitrant the snake, the longer
the specialist will make the text, precisely by generating more
lines by means of parallelistic structures. At the same time,
the performer's intimate knowledge of the nature of the spirit
world, especially its parts and taxonomic classification, is also
displayed by parallelistic structures and processes. And since
specialists must memorize these texts, parallelistic line, verse,
and stanza frames seem to provide mnemonic aids to memoriza-
tion. Finally, this extensive parallelism aids in actual perform-
ance, providing both time and procedures for moving from line
to line, narrative description to narrative description. It is
no wonder, given these various functions, that parallelism is
so pervasive in nakpe ikar.
The last aspect of the structure of nakpe ikar that I have
chosen to examine is the crucial interplay of first, second, and
third person within the narrative. Nakpe ikar, like all Kuna
magical chants, is a third person narrative. Although the
actual performer is the specialist who will raise a snake, the
text describes the spirit world in which both specialist and
snake are third persons. But at various points in the text,
the dialogue between the specialist spirit and the snake spirit
is quoted and the pronouns 'I' and 'you' are used. 'You' is
 Interplay of Structure and Function in Kuna Narrative / 319

used in many lines of quoted dialogue to refer to the specialist

spirit and the snake spirit, and 'I' is used in several lines of
quoted dialogue to refer to the snake spirit. However, 'I' is
used to refer to the specialist spirit in only one line, the cli-
mactic moment of the text, the performative formula within the
performative text, the actual moment of the grabbing and rais-
ing of the snake, the crucial line, '"Simply I raise you" I am
saying', in which the double quotation marks within single
quotation marks indicate that the actual specialist is quoting
the specialist spirit who is quoting himself. 12 In this climactic
moment of nakpe ikar, the grammatical category of person, like
other features in the structure of the text, serves several
functions, including especially poetic and magical ones. Once
again, grammar becomes poetry and poetry becomes magic.
In this paper I have explored the structure of a single Kuna
narrative, nakpe ikar, a magical chant used to grab and raise
a dangerous snake. At one level my approach has been analo-
gous to the linguistic analysis of a poem. And this is part of
my point in this exercise, since analysis of nakpe ikar requires
recognition of its poetic properties. 1 3 But at the same time,
examination of this magical narrative is intended as an example
of one kind of ethnographic approach to discourse. Attention
to the constant and dynamic interplay of structure and func-
tion in nakpe ikar reveals a complex web of relations within
Kuna language, culture, and society, which involves the stra-
tegic importance of snakes in Kuna culture; the relationship be-
tween humans and animals and plants; the use of language and
speech to display knowledge, respect, and control; the relation-
ship between the world of humans and the world of spirits,
mediated by the poetic-rhetoric of memorized oral chants; the
role of grammar, parallelism, metaphor, and narrative structure
in this poetic rhetoric; and the belief in the power and ability
of language to solve specific problems.

NOTES
I am most grateful for the comments of the following indi-
viduals on earlier versions of this paper: Richard Bauman,
Mac Chapin, James Howe, Dina Sherzer, and Anselmo Urrutia.
1. See, for example, Givon (1979), Grossman, San, and
Vance (1975), Halliday (1973), Hymes (1974), Jakobson (1960),
and Silverstein (1976). My use of the term function in rela-
tion to structure is not intended to indicate an adherence to
the way these notions have been related by certain schools of
social anthropology, especially British. Rather, I refer here
to an emerging focus on the interplay of structure and function
in recent research in the analysis of discourse.
2. In the performance of magical ikars for curing and dis-
ease prevention, slight variations of an essentially nonreferen-
tial nature are tolerated, involving very superficial aspects of
the phonology and morphology of noun and verb suffixation.
320 / Joel Sherzer

The magical texts performed for Kuna girls' puberty rites by

contrast are completely fixed. Not the slightest variation in
phonology or morphology is tolerated. The degree to which
Kuna puberty rites texts are fixed in form is reflected in a
personal experience, In 1970 I made a tape recording of a
puberty rites specialist teaching a text to several students.
Between 1970 and 1978 I never discussed this text with him.
In March of 1979, nine years after the original recording, I
brought him a transcription I had made of the text, in order
to translate it into ordinary colloquial Kuna, from which it
differs considerably. Since he does not read or write, he
asked me to read him the text. I did so line by line and he
translated each line into colloquial Kuna. Typically, I barely
began a line and he finished it, never missing a morpheme or
even a phoneme from my transcription. In a few cases where
I misread a tiny detail of my own writing, he corrected me.
The Kuna thus provide still another counterexample (for many
others see Finnegan 1977) to the view (see Goody 1977, Lord
1960, Ong 1977) that there is no pure, verbatim memorization
of fixed texts in nonliterate, oral societies.
3. I am indebted to Chapin (1981) for the comparison of
Kuna magical texts with dramatic scripts.
4. That is, the magical ikars described here. Another form
of discourse sometimes used in magic is the sekretto, a short
verbal charm which involves a considerable amount of non-
intelligible, nonanalyzable language.
5. Such quasi-metaphorical disease causation theories, in-
volving animal spirits, are found in other societies, for exam-
ple, the Ainu (see Ohnuki-Tierney 1977). For the details of
Kuna theories and practices in relation to disease and curing,
see Chapin (1981).
6. Nakpe ikar was performed by Pranki Pilos of Mulatuppu
on March 2, 1971. In this translated presentation, lines are
determined by melodic shape and pauses. In Pilos' perform-
ance, groups of lines (verses in Hymes' 1977 terminology) were
marked by a cough and an extra long pause. I have repre-
sented this structure here by leaving an extra space between
lines.
7. The ordinary Kuna word for this snake is tappet, a small
pit viper.
8. It is interesting to compare this case of poetry in action
with the sequence of increasingly powerful sounds (ritual in-
sults) reported by Labov for a group of Harlem Blacks (Labov
1972a:349):
Junior: I'll take your mother.
Rel: I took your mother.
Both the Kuna magical text and the Harlem sounding bring
about their results with what Labov terms the 'minimax' solu-
tion: 'striking semantic shifts with minimal changes of form.'
 Interplay of Structure and Function in Kuna Narrative / 321

Labov's comments after the oral presentation of my paper at

The Georgetown University Round Table called my attention to
the similarities between these otherwise very different forms
of discourse.
9. See Rosaldo (1975) and Tambiah (1968) for studies of the
role of poetry in the magic of other societies..
10. I include identical repetition as a type of parallelism
here, although other students of parallelism might not, because
of its role in the overall parallelistic patterning and structuring
of Kuna magical texts.
11. Erickson (this volume) examines another, quite different,
function of listing, in an American conversational setting.
12. I have used [I] within brackets in other lines within
the text where I felt that the English translation required it.
The brackets are intended to indicate that the original Kuna
version did not have the Kuna word ani T .
13. For a more general discussion of Kuna poetics, see
Sherzer (1977).
REFERENCES

Bauman, Richard. 1977. Verbal art as performance. Rowley,

Mass.: Newbury House.
Chapin, Mac. 1981. Medicine among the San Bias Kuna.
Ph.D. dissertation. University of Arizona.
Finnegan, Ruth. 1977. Oral poetry. Cambridge, England:
Cambridge University Press.
Friedrich, Paul. 1979. Language, context, and the imagi-
nation. Stanford, Calif.: Stanford University Press.
Geertz, Clifford. 1973. The interpretation of cultures. New
York: Basic Books.
Givon, Talmy, ed. 1979. Discourse and syntax. Syntax and
semantics, Vol. 12. New York: Academic Press.
Goody, Jack. 1977. The domestication of the savage mind.
Cambridge, England: Cambridge University Press.
Grossman, Robin E., L. James San, and Timothy J. Vance,
eds. 1975. Papers from the Parasession on Functionalist.
Chicago: Chicago Linguistic Society.
Gumperz, John. 1971. Language in social groups. Stanford,
Calif.: Stanford University Press.
Halliday, M.A.K. 1973. Explorations in the functions of-lan-
guage. London: Edward Arnold.
Hymes, Dell. 1974. Foundations in sodolinguistics: An
ethnographic approach. Philadelphia: University of
Pennsylvania Press.
Hymes, Dell. 1977. Discovering oral performance and mea-
sured verse in American Indian narrative. New Literary
History 8.3:431-457.
Hymes, Dell, (in press) Particle, pause and pattern in Ameri-
can Indian narrative verse. American Indian Culture and
Research Journal.
322 / Joel Sherzer

Jakobson, Roman. 1960. Concluding statement: Linguistics

and poetics. In: Style in language. Edited by Thomas A.
Sebeok. Cambridge, Mass.: MIT Press. 350-377.
Labov, William. 1972a. Rules for ritual insults. In: Lan-
guage in the inner city: Studies in the Black English ver-
nacular. By William Labov. Philadelphia: University of
Pennsylvania Press. 297-353.
Labov, William. 1972b. The transformation of experience in
narrative syntax. In: Language in the inner city: Studies
in the Black English vernacular. By William Labov. Phila-
delphia: University of Pennsylvania Press. 354-396.
Lord, Albert B. 1960. The singer of tales. Cambridge,
Mass.: Harvard University Press.
Ohnuki-Tierney, E. 1977. An octopus headache? A lamprey
boil? Multisensory perception of 'habitual illness' and world
view of the Ainu. Journal of Anthropological Research 33.
245-257.
Ong, Walter J. 1977. Interfaces of the word. Ithaca and
London: Cornell University Press.
Rosaldo, Michelle. 1975. It's all uphill: The creative meta-
phors of Ilongot magical spells. In: Sociocultural dimensions
of language use. Edited by Mary Sanches and Ben G.
Blount. New York: Academic Press. 117-203.
Sherzer, Joel. 1975. A problem in Cuna phonology. Journal
of the Linguistic Association of the Southwest 1.2:45-53.
Sherzer, Joel. 1977. Cuna Ikala: Literature in San Bias.
In: Verbal art as performance. By Richard Bauman. Row-
ley, Mass.: Newbury House. 133-150.
Sherzer, Joel, (in press) Tellings, retellings, and tellings
within tellings: The structuring and organization of narra-
tive in Kuna Indian discourse. Proceedings of conference
on 'oralita: cultura, letteratura, discorso,' Urbino, Italy,
1980.
Silverstein, Michael. 1976. Shifters, linguistic categories,
and cultural description. In: Meaning in anthropology.
Edited by Keith H. Basso and Hugh A. Selby. Albuquerque:
University of New Mexico Press. 11-55.
Tambiah, S. J. 1968. The magical power of words. Man 3.2:
175-208.
Tedlock, Dennis. 1978. Finding the center: Narrative
poetry of the Zuni Indians. Lincoln: University of Neb-
raska Press.
 THE LINGUISTIC BASES
OF COMMUNICATIVE COMPETENCE
John J. Gum perz
University of California, Berkeley,
and Princeton University

Let me begin with a reminiscence which may serve to intro-

duce the point I wish to make. This is the third paper on
conversation I have given at the Georgetown University Round
Table on Languages and Linguistics. My first on this topic,
which reflects an initial attempt to come to terms with the se-
mantic complexities of verbal interaction processes, was given
in this room in 1972, at a session which included Harold Gar-
finkel, Erving Goffman, and Harvey Sacks (Shuy 1972). Sacks
gave his well-known paper on puns and punning, which by
now has become a classic of conversational analysis. Yet if
audience response counts as a measure of success, his presen-
tation was far from successful.
The room was only partly filled and, in fact, several listeners
walked out muttering that they could not understand the
speaker. Sacks' unorthodox style of presentation seemed to
jar the audience of linguists. He introduced basic theoretical
notions by means of descriptive phrases such as: 'sequential
ordering', ' positioning of utterances', 'the interactional job
that utterances do', 'what they (i.e. speakers) need to do is
exhibit understanding', 'the use of the performance rule'. In
all of these expressions, verb constructions like 'ordering',
'positioning', 'understanding', and 'using (or for that matter
also violating) rules' are consciously employed to suggest that
reference is being made to acts that speakers perform by means
of their talk and not to givens of language usage.
Today such interactive approaches to speaking and the ex-
pressions Sacks introduced have become familiar. It almost
seems as if we are witnessing a change in paradigm. Whereas
nine years ago discourse was one of many possible sociolinguis-
tic topics, secondary perhaps in theoretical importance to

323
324 / John J . Cumperz

problems of variation, language and social stratification, bi-

lingualism and language planning, today it has become the main
topic of this three-day national conference, and basic issues of
discourse coherence which Sacks was among the first to raise
are now of central concern to sociolinguists and linguists alike.
The issues I want to discuss today are questions of linguistic
theory that arise from this new perspective and that to my
knowledge have not been seriously posed. What have we
learned over the last decades by applying micro-conversational
analyses to conversational data such as have only recently be-
come available for systematic study through innovations in
audio and video technology? What does the interactive approach
to communication, which sees communicating as the outcome of
exchanges involving more than one active participant, imply for
the way we look at linguistic data and for our theories of gram-
mar and meaning? What do conversational exchanges tell us
about the interplay of linguistic, sociocultural and contextual
presuppositions in interpretation?
A key concept we need to reconsider is the notion of com-
municative competence. The term is a familiar one, coined by
Dell Hymes to suggest that as linguists concerned with com-
munication in human groups we need to go beyond mere descrip-
tion of language usage patterns to concentrate on aspects of
shared knowledge and cognitive abilities which are every bit
as abstract and general as the knowledge that is glossed by
Chomsky's more narrowly defined notion of linguistic competence.
Among European social scientists the term has become familiar
through the writings of Jurgen Habermas, who argues that an
understanding of communication is basic to a more general
theory of social and political processes. He calls for a theory
of communicative competence that would specify what he terms
'the universal conditions of possible understandings'. But it is
far from clear exactly what facts of human interaction such a
theory must account for and how we can characterize the knowl-
edge speakers must have and the socioeconomic environments
that can create these conditions.
Habermas (1970) in his informal discussion relies on notions
of what he calls 'trouble free communication' and assumptions
about sharedness of code which recall Chomsky's ideally uni-
form communities, as if understanding depended on the exist-
ence of a unitary set of grammatical rules. Yet sociolinguistic
research during the last decade has demonstrated not only that
all existing human communities are diverse at all levels of lin-
guistic structure, but also that grammatical diversity, multi-
focality of linguistic symbols, and context dependence of inter-
pretive processes are essential components of the signalling re-
sources that members rely on to accomplish their goals in
everyday life (Gumperz in preparation).
Other empirical findings that by now have become generally
accepted are: that generalization about ongoing processes of
language change must build on empirical data on everyday
 The Linguistic Bases of Communicative Competence / 325

speech in a range of natural settings; that basic issues of lan-

guage acquisition can best be explained by reference to the be-
havioral facts of mother-child interaction; that the grammatical-
ity judgments which furnish the data for syntactic analysis de-
pend on speakers' ability to imagine a context in which the
sentence could occur; and that, as several speakers in this
Georgetown University Round Table have pointed out, discourse
consists of more than the sum of component utterances. The
theoretical linguists' insistence on maintaining a strict sepa-
ration between linguistic and extralinguistic phenomena has
thus become untenable in many key areas of linguistic research.
Yet even though these points are by now gaining acceptance,
and context and sociocultural presuppositions are beginning to
be brought into our explanatory models, our ideas of what is
rule governed about speaking and about how meaning is con-
veyed continue to be based on concepts deriving from sentence-
based grammatical analysis. We talk about language and cul-
ture, language and context, as if these were separate entities,
which stand outside the actual message and which, like the
ideas in the conduit metaphor (Reddy 1979), can be likened to
bounded concrete objects. Moreover, and more importantly
perhaps, our methods for analyzing contextual and social as-
pects of communication rest on procedures of taxonomic cate-
gorization and on statistical distribution counts which are quite
distinct from the introspective, interpretive methods of linguis-
tic analysis. Many sociolinguistic studies of communicative
competence, in fact, aim at little more than statements of regu-
larities that describe the occurrence of utterances or verbal
strategies isolated by traditional methods of linguistic analysis
in relation to types of speakers, audiences, settings, and situ-
ations. This leads to a highly particularistic notion of compe-
tence, which some psycholinguists claim has little relevance for
basic cognitive processes; they argue that lexical and syntactic
measures are the only valid indices of verbal skills.
I believe that we can avoid the difficulties that this raises by
integrating the sociolinguistic findings on variability with
Habermas's call for a theory of possible human understanding.
What is needed is an approach which can relate the specifics of
situated interpretation to the panhuman ability to engage others
in discourse. I propose therefore that we redefine communica-
tive competence as 'the knowledge of linguistic and related com-
municative conventions that speakers must have to initiate and
sustain conversational involvement'. Conversational involvement
is clearly a necessary precondition for understanding. Com-
munication always presupposes some sharing of signalling con-
ventions, but this does not mean that interlocutors must speak
a single language or dialect in the sense that linguists use the
term.
Code-switching studies over the last ten years have docu-
mented a variety of speech situations in societies throughout
the world where speakers build on the contrast between two
326 / John J . Gumperz

distinct grammatical systems to convey substantive information

that elsewhere in equivalent situations can be conveyed by the
grammatical devices of a single system. Moreover, participants
in a conversation need not agree on the specifics of what is
intended. People frequently walk away from an encounter feel-
ing that it has been highly successful only to find later that
they disagree on what was actually said. Studies of communi-
cative competence, therefore, must deal with linguistic signs
at a level of generality which transcends the bounds of lin-
guists' grammatical system and must concentrate on aspects of
meaning or interpretation more general than that of sentence
content. It is furthermore evident that the perceptual cues we
must process in conversational exchanges are different from
those that apply to decoding of isolated sentences.
The following example of a brief recorded exchange between
two secretaries in a small university office serves to organize
my discussion of relevant interpretive processes.
1. A: Are you going to be here for ten minutes?
2. B: Go ahead and take your break. Take longer if
you want.
3. A: I'll just be outside on the porch. Call me if you
need me.
4. B: OK. Don't worry.
Brief though it i s , this exchange nevertheless contains in it-
self much of the data we need to determine what participants
intended and how it was achieved. Setting aside for the mo-
ment our natural tendency to concentrate on the meanings of
component utterances, we note that B interprets A's opening
move as a request to stay in the office while she takes a break.
By her reply in line 3, A then confirms B's interpretation and
B's final 'OK. Don't worry' both reconfirms what was agreed
and concludes the exchange.
Given this evidence showing that both speakers have actively
participated and have proffered and agreed on (for the moment
at least) interpretations, we can proceed to employ the lin-
guist's interpretive, introspective methods of analysis to seek
hypotheses as to what knowledge participants rely on and what
signalling cues they perceive to accomplish what they do.
Note that B's interpretation is an indirect one which responds
to more than the referential meaning or the illocutionary force
of A's utterance. The inferential process here seems to have
some of the characteristics of Gricean implicature. That is, we
assume that B assumes A is cooperating, that her question must
therefore be relevant, and that since there is no immediately
available referent she searches her memory for some possible
context, i.e. some interpretive frame that would make sense.
But this begs the question how B arrives at the right infer-
ence. What is it about the situation that leads her to think A
is talking about taking a break? A common sociolinguistic
 The Linguistic Bases of Communicative Competence / 327

procedure in such cases is to attempt to formulate discourse

rules such as the following: 'If a secretary in an office
around break time asks a co-worker a question seeking infor-
mation about the co-worker's plans for the period usually al- 1
lotted for breaks, interpret it as a request to take her break.
Such rules are difficult to formulate and in any case are neither
sufficiently general to cover a wide enough range of situations
nor specific enough to predict responses. An alternative ap-
proach is to consider the pragmatics of questioning and to
argue that questioning is semantically related to requesting,
and that there are a number of contexts in which questions can
be interpreted as requests. While such semantic processes
clearly channel conversational inference, there is nothing in
this type of explanation that refers to taking a break.
Note that all the foregoing arguments rely on sentence-based
views of language which assume that the cues that conversa-
tionalists process are basically those covered in traditional
phonological, syntactic, and semantic analysis. I believe that
conversational inference relies on additional types of linguistic
signalling and that an understanding of how these signs work
to channel interpretation is basic to a theory of communicative
competence.
Some of these contextualization cues (Gumperz 1977, Cook-
Gumperz and Gumperz 1978, Gumperz and Tannen 1979) have
to do with what, following sociological work in conversational
analysis, have come to be called sequencing or turn-taking
processes. Sacks, Schegloff, and Jefferson (1974) argue that
speaker change is a basic conversational process and that
turn-taking mechanisms are organized about transition relevance
places, which determine when a next speaker can take the floor.
But they give no data on how such transition relevance places
are signalled. Conversations, unlike sentences, do not come
pre-chunked. Conversationalists must process verbal signs to
determine when to take turns without interfering with others'
rights. For example, B's utterance in line 2 comes in immedi-
ately at the end of A's sentence, while A in line 3 waits long
enough to allow B to produce not one but two sentences. How
is this negotiated?
Moreover, responses in this exchange, as well as in most
verbal encounters, are rhythmically organized in such a way
that--as Erickson and Scollon have pointed out in this George-
town University Round Table--moves follow each other at regu-
lar time intervals. This rhythmicity is important in maintaining
conversational involvement (Erickson and Shultz 1981). Halliday
(1967), in discussing problems of segmentation in longer pass-
ages, argues that language is chunked into semantically holistic
information units, but his discussion focuses largely on written
texts and on the role of syntax in chunking. Chafe (1980)
proposes the notion of idea unit to deal with related issues
and points to the role of tempo and pausing in segmentation.
These problems are treated in considerable detail in recent work
328 / John J . Cumperz

on intonation. Ladd's (1980) comprehensive review of the

relevant literature indicates that chunking cannot be described
in terms of a single phonetically determined set of signalling
cues. Chunking is an act of interpretation involving simul-
taneous processing of signs at several levels of signalling:
prosodic, phonological, syntactic, lexical and rhythmic, which,
like the process of phonemic categorization described by struc-
tural linguists, depends on learned conventions which differen-
tially highlight or ignore some cues at the expense of others.
Chunking or phrasing of speech, moreover, does more than
just signal transition relevance places; it also serves to indi-
cate relationships among items of information and to set off or
foreground others. A's utterance in line 1 could have been
split by a tone group boundary, while B's line 4 could have
been grouped together under a single clause contour.
Another process of key importance for conversational infer-
ence is the signalling of utterance prominence to indicate which
of several bits of information is to be highlighted or placed in
focus. In our English rhetorical tradition this is done partly
through syntax and lexical choice and partly through placement
of prosodic accent. Given a particular choice of words, we
have certain expectations about normal accent placement.
These can then be systematically violated to convey additional
information not overtly given in the message. If in line 1 'you'
had been accented rather than 'be here 1 , B might have thought
the question referred to whether she, as opposed to someone
else, was going to stay in the office. She might not have
recognized it as a request. As it is, the interpretation B
actually does make relies in large part on the fact that 'be
here' is emphasized by primary stress, 'ten minutes' carries
secondary stress, and the two phrases come under the same
contour. We assume that B perceived the question as focusing
on her being (i.e. remaining) for a period equivalent to that
normally associated with the morning break, and that this led
to her inference. Other rhetorical traditions rely on informa-
tion signalled through a different combination of signalling
channels or subsystems. What is important is that perception
of focus always relies on expectations of how these channels
cooccur, and these expectations are not dealt with in our usual
grammatical analyses, which tend to focus on one subsystem at
a time.
A final signalling cue of relevance here is the choice of dis-
course strategy. Note that A could have achieved her end by
simply asking 'Can I take my break?', in which case a simple
one-word or one-phrase answer like 'yes' or 'OK' would have
been sufficient to complete the exchange. But given her choice
of words, our experience with similar situations tells us that
more talk is expected. There is something of a formulaic nature
about exchanges such as these which affects our interpretations.
This discussion suggests that conversational inference is best
seen not as a simple unitary evaluation of intent but as involving
 The Linguistic Bases of Communicative Competence / 329

a complex series of judgments, including relational or con-

textual assessments on how items of information are to be inte-
grated into what we know and into the event at hand, as well
as assessments of content. This is a point that has been1 made
several times during this Round Table. Agar and Hobbs dis-
tinguish between global and local assessment. Iivia Polanyi,
in her comment on Coulthard's paper, suggested that we need
to distinguish between discourse and sentence level inferences.
Fillmore's (this volume) approach to reading also reflects a
similar perspective.
One can visualize the process as consisting of a series of
stages which are hierarchically ordered so that more general
relational assessment serves as part of the input to more spe-
cific ones. Perception of contextualization cues, moreover,
plays a role at every stage.
It is assumed that the initial assessment in an exchange con-
cerns the nature of the activity being proposed or performed.
This sets up expectations about what likely outcomes are, what
topics can be covered, what can be put in words and what
must be conveyed indirectly; and what counts as suitable styles
of speaking and thereby provides the motivation for entering
into the interaction in the first place. At the next lower level,
decisions are made about the more immediate communicative or
discourse tasks such as narrating, describing, requesting,
which together make up particular activities. Such discourse
tasks have some similarity to the linguists' speech act, but
they differ in that they typically consist of more than one
utterance and in that they are described in terms of primary
semantic relationships that tie together component utterances,
rather than in terms of illocutionary force.
Note that whereas activities are often culturally or situation-
ally specific, discourse tasks are universals of human inter-
action. An understanding of how relational signs function to
signal these tasks can provide basic insights into how interpre-
tations are agreed upon and altered in the course of an inter-
action by differentially foregrounding, subordinating, and
associating various information carrying items. If conversa-
tional involvement is to be maintained, higher level relational
signs must be shared although participants may disagree on the
meaning of words and idioms. On the other hand, however
participants may agree on what sentences mean in isolation,
yet when relational signals differ, conversational cooperation is
likely to break down. Cross-cultural analysis of how discourse
tasks are signalled--that is, how focusing, phrasing, corefer-
entiality and other aspects of cohesion are signalled—can form
the basis for empirical investigations of pan human features of
communicative competence.
It must be emphasized that verbal strategies for negotiating
conversational interpretations are for the most part indirect.
Information is not overtly expressed in surface content, but
330 / John J . Gumperz

must be inferred on the basis of tacit presuppositions acquired

through previous interactive experience. Indirect signalling
mechanisms differ from lexicalized signs in that, like nonverbal
signals, they are inherently ambiguous. Any single utterance
is always subject to multiple interpretations. One decides on
what interpretation to accept by examining what Austin has
called 'uptake', that is, the conversational process through
which lines of reasoning are developed or altered.
Given the nature of the signalling system, participants, in
order to be able to develop their arguments, are constantly
required to test and display the tacit knowledge on which they
rely to make inferences in the first place. Wherever conversa-
tional cooperation is maintained over time, that i s , wherever we
find evidence that conversationalists actively react to and work
with each others' responses to establish cohesive themes, we
can assume at least some sharing of tacit contextualization
strategies.
Failure to achieve this type of cooperation, on the other
hand, may in some cases, although certainly by no means in
all, indicate undetected differences in signalling systems. In
the midst of an exchange, when conversationalists are faced
with the need to respond in time and have little opportunity
to reflect, such difficulties tend to go undiagnosed. The fact
that they exist must be discovered through post-hoc empirical
analysis. It is here that the new audiovisual technologies
which for the first time in human history enable us to freeze
and preserve for systematic study samples of naturalistic ex-
changes can provide truly novel insights into the workings of
communicative processes.
Our recent empirical studies in ethnically diverse urban
settings indicate that miscommunications attributable to unde-
tected systematic differences in signalling conventions occur
considerably more frequently than casual observations would
lead one to suspect. A possible linguistic reason for this is
that contextualization conventions are distributed along areal
networks which do not necessarily coincide with language or
dialect boundaries as established through historical reconstruc-
tion or typological comparison of grammatical categories. Such
conventions are created through prolonged interactive experi-
ence in family, friendship, occupational, or similar networks of
relationships. Typically, they affect the signalling of contextual
and interutterance relationships through formulaic expressions,
phrasing or chunking, focusing, anaphora, deixis, or other
grammatical cohesive mechanisms. Once established through
practice, they come to serve as communicative resources which
channel inferences along particular lines. Knowledge of how
they work becomes a precondition for active participation in
verbal encounters. The knowledge is of a kind which cannot
easily be acquired through reading or formal classroom instruc-
tion. Personal contact in situations which allow for maximum
feedback is necessary.
 The Linguistic Bases of Communicative Competence / 331

Potential language learners thus face a real dilemma. They

must establish long-lasting intensive person-to-person contacts
in order to learn, yet their very lack of the necessary strate-
gies makes it difficult for them to establish such contacts. In
real life situations, learning of discourse strategies is most
successful when outside conditions exist which force interlocu-
tors to disregard breakdowns and stay in contact or to give
the learner the benefit of the doubt. This is the case in
mother and child interaction or in apprenticeship situations at
work. But conditions in modern urban societies are hardly
favorable to informal experiential learning. Here contact with
others of different background is often the norm in public af-
fairs, while friendship circles are limited by similarity of back-
ground. Public situations, moreover, most frequently revolve
around evaluation of ability or intent to cooperate and, given
the nature of the tensions of urban life, rarely provide the
conditions where breakdowns can be disregarded. The result
is that the ability to achieve one's goals--that i s , to get things
done in face-to-face public settings—is often a matter of shared
background. Outsiders who enter the urban scene may learn a
new language or dialect well at the level of sentence grammar
or lexicon, and this knowledge is sufficient for the instrumental
contacts with outsiders that fill up much of the working day.
But situations of persuasion, where what is evaluated is the
ability to explain, describe, or narrate, are often difficult to
manage. Here breakdowns tend to lead to mutual stereotyping
and pejorative evaluations.
To be sure, not all problems of interethnic contact are com-
municative in nature. Economic factors, differences in goals
and aspirations, as well as other historical and cultural factors,
may be at issue. But we have reason to suspect that a sig-
nificant number of breakdowns may be due to inferences based
on undetected differences in contextualization strategies, which
are after all the symbolic tip of the iceberg reflecting the
forces of history. The existence of communicative differences
must, of course, be demonstrated. It cannot be presupposed
or inferred from grammars or the usual ethnographic descrip-
tions. Here conversational analysis becomes a diagnostic tool
to determine whether the linguistic prerequisites of possible
communication exist.
How do we go about documenting the functioning of con-
textualization strategies? One way to accomplish this is to
concentrate on naturally occurring events such as court pro-
ceedings, job interviews, medical diagnoses, and committee
meetings, where discourse strategies play a key role in the
evaluation of performance. Let me present some data from
transcribed testimony of a Navy hearing held in connection
with a perjury trial. The accused was a Navy physician born
in the Philippines, who had been indicted for perjury in con-
nection with statements he had made concerning a burn injury
he had treated.
332 / John J . Gumperz

A principal goal of the hearing was to document his profes-

sional qualifications. He had spent many years in the United
States and speaks English well. The questioning deals with
his training in burn treatment.
Q.I. Any other sources of burns that you've observed?
A. 2. Occasionally from gasoline and kerosene burn because
as far as the
3. situation there, most of the houses don't have any
natural gas or electric
4. stove as here. They use kerosene instead as a means
of fuel for cooking.
5. The reason why I'm saying this also is because this
hospital where I
6. had my training is a government hospital, so most of
the patients that
7. go there are the poverty stricken patients unlike you
going to a medi-
8. cal center, it's usually the middle class who go where
you don't have
9. this problem.
In the preceding parts of his testimony dealing with his
training in the Philippines, the witness has repeatedly compared
conditions there with those in the United States. His argument
in lines 2-4 rests on such a contrast, and from the way he be-
gins the second part of his answer, starting in line 5, one
would expect a similar comparison of 'there' with 'here'. But
the content of his sentences does not seem to bear out these
expectations. If one examines what he says, starting with:
'most of the patients who go there are the poverty stricken
patients unlike you going to a medical c e n t e r . . . ' , one is unsure
what is being compared: poverty stricken patients with middle
class patients, or medical centers in the United States with
government hospitals in the Philippines?
Participants in an interaction, as well as most casual observ-
ers, are likely to see such conflicts as reflecting on the wit-
ness's credibility, but our experience with similar types of
interethnic situations leads us to suspect that in situations
like this, where expectations signalled at one level of generality
are not born out by lower level signalling processes, systematic
processing difficulties ultimately attributable to grammatical
presuppositions may be at work.
Note that the passage is too long to be processed as a whole.
A reader will have to rely on syntactic and prosodic knowledge
to sound it out and chunk it into relevant information units.
Native speakers of English who do this will have difficulty in
assigning the word 'unlike' in line 7 to either the preceding or
following clause. The first reading yields the clause 'poverty
stricken patients unlike you'. This not only conflicts with ex-
pectations signalled through the preceding context but also
 The Linguistic Bases of Communicative Competence / 333

renders the remaining passage unintelligible. Speakers of

Filipino English who were consulted tended to assign 'unlike'
to the following clause and had no difficulty in recognizing
the speaker's intent to contrast 'there 1 with 'here'. Yet
native speakers of English are likely to have difficulty in
fitting a clause such as 'unlike you going to a medical center'
into the surrounding discourse frame.
The problem is a complex one, requiring more detailed analy-
sis than can be presented here. But the most likely expla-
nation lies in the discourse conventions for signalling co-
referentiality. To make sense of the 'unlike' clause, an Eng-
lish speaker would have to recognize it as a syntactically in-
complete clause in which 'this is' had been deleted or left un-
expressed immediately preceding 'unlike'. To recover such un-
verbalized information, English speakers look for a pronoun or
noun phrase that could signal coreferentiality. It is our in-
ability to locate such a phrase in the foregoing passage that
leads to processing difficulty. My hypothesis, which will, of
course, have to be tested through systematic research, is that
Filipino English speakers, even though they speak grammatical
English at the sentence level, nevertheless employ discourse
principles influenced by Tagalog and similar Austronesian lan-
guages, where coreferentiality is signalled by means other than
overtly lexicalized pronouns or noun phrases. The same pass-
age can thus be differently interpreted by listeners who pro-
cess it with different presuppositions.
Investigation of such multilevel signalling processes, and of
the role played by contextualization as well as by linguistic
and sociocultural presuppositions in the multilevel inferences
necessary to sustain verbal exchanges, could lay the founda-
tion of a universal theory of communicative competence capable
of providing new insights into the communicative problems that
affect our urban societies.
NOTES

Research on this paper was supported by grants from The

Institute of Advanced Studies in Princeton, N . J . , and the
National Institutes of Health.
1. Editor's note: This is a reference to Michael A gar and
Jerry Hobbs, who participated in a preconference session,
'Toward adequate formal models of discourse', before the 1981
Georgetown University Round Table on Languages and Lin-
guistics.
REFERENCES

Chafe, Wallace L. 1980. The deployment of consciousness in

the production of a narrative. In: The pear stories:
Cognitive, cultural, and linguistics aspects of narrative
334 / John J . Gumperz

production. Edited by Wallace Chafe. Norwood, N . J . :

Ablex.
Cook-Gumperz, Jenny, and John Gumperz. 1978. Context in
children's speech. In: The development of communication.
Edited by Katherine Snow and Natalie Waterson. London:
Wiley.
Erickson, Frederick, and Jeffrey Shultz. 1981. Talking to
the man. New York: Academic Press.
Gumperz, John. 1977. Sociocultural knowledge in conversa-
tional inference. In: Georgetown University Round Table
on Languages and Linguistics 1977. Edited by Muriel Saville-
Troike. Washington, D.C.: Georgetown University Press.
Gumperz, John, (in preparation) Discourse strategies.
Gumperz, John, and Deborah Tannen. 1979. Individual and
social differences in language use. In: Individual differ-
ences in language ability and language behavior. Edited by
C. J. Fillmore, W. Kempler, and W. S.-Y. Wang. New York:
Academic Press.
Habermas, Jurgen. 1970. Toward a theory of communicative
competence. In: Recent sociology II. Edited by H. P.
Dreitzel. London: Macmillan.
Halliday, M.A.K. 1967. Notes on transitivity and theme in
English, Part 2. Journal of Linguistics 3.2:199-244.
Ladd, Robert. 1980. The structure of intonational meaning.
Bloomington: Indiana University Press.
Reddy, Michael. 1979. The conduit metaphor: A case of
frame conflict in our language about language. In: Meta-
phor and thought. Edited by Andrew Ortony. Cambridge:
Cambridge University Press.
Sacks, Harvey, Emanuel Schegloff, and Gail Jefferson. 1974.
A simplest systematics for the organization of turn taking
for conversation. Lg. 50.696-735.
Shuy, Roger, ed. 1972. Georgetown University Round Table
on Languages and Linguistics 1972. Washington, D.C.:
Georgetown University Press.
 THE RHYTHMIC INTEGRATION
OF ORDINARY TALK
Ron Scolhn
University of Alaska

Rhythm, timing, tempo, pace. Rhythm, timing, tempo, pace.

These notions have appeared from time to time in our discourse
about discourse. Recently, Erickson (1980; Erickson and
Shultz 1977; in press) has shown that some notion of rhyth-
micity is central to our understanding of such very different
face-to-face interactions as junior college counseling interviews
and elementary school examinations. Much earlier, Chappie
(1939) pointed out the centrality of time in exchanging turns
between speakers in interviews. Householder was one of the
first linguists to take rhythmicity seriously in his thinking
about accent, juncture, and intonation (1957). He said:
The fact is, we can't hear noises repeated with fair regu-
larity at more than a certain average frequency without
grouping them rhythmically (as every subway-rider can
testify), and once a given pattern is established we will
hear it over and over till some new irregularity breaks
the rhythm and starts another pattern (1957:238).
Many of us, in reading Lenneberg (1967), were caught by
the notion of the critical period for language development and
missed his assertion that rhythm was the organizing principle
of speech and language. In 1970, it was still possible for non-
linguists such as Jaffe and Feldstein, working on face-to-face
interaction, to say that linguists had not taken rhythmic phe-
nomena seriously. Jaffe and Feldstein could say with confi-
dence (1970) that when linguists began looking at connected
speech (that is, at discourse), they would necessarily find
themselves wanting to address rhythmic issues. Now one sees
that linguists are indeed finding rhythmic phenomena of inter-
est. As a recent example, phonologists such as Ladd (1980)

335
336 / Ron Scollon

and Liberman and Prince (1977) have extended the work of

Bolinger (1964) to build notions of sentential stress and into-
nation around rhythm and timing. In short, while notions of
rhythm, timing, tempo, and pace have been around for some
time in our discourse about discourse, I believe that a point
has been reached where rhythmicity needs to become a more
central concern.
My purpose in this paper is to show how I have found the
idea of tempo to be useful in my thinking about discourse and
to try to clarify one or two confusions that seem to crop up
wherever people have wanted to use this notion in their dis-
course about discourse.
Before getting to my main point, though, I think it is useful
to clarify my reasons for an interest in discourse. While I
take my audience here to be primarily an audience of linguists,
I need to point out that I do not see myself as doing linguistic
work so much as using linguistic work to do other things. I
am interested in coming to understand how 'institutions' are
constituted in discursive practices. In this I rely heavily on
the thinking of Foucault (1976, 1977, 1980a, 1980b). Although
his use of the term 'discourse' is broader than the use made by
many linguists, it is not incompatible with linguists' use. I
think in this regard of the comment made by John Gumperz
(personal communication) that 'discourse is the key to histori-
cal process'. My interest, not unlike Chappie's (1980), is to
understand the constitution of our institutions in the process
of face-to-face interaction.
More specifically, in my work in Alaska I am involved with
a number of institutions--schools, from pre-school to post-
secondary institutions (Scollon 1981a), the criminal justice
system, the judicial system, resource development corporations,
federal regulatory agencies, and Alaska Native corporations--in
their relationships with a particular population, Alaska Natives,
who do not share institutional assumptions nor assumptions
about the nature of discursive practices within those institu-
tions. Let me emphasize that the preceding clause is intended
to be a restrictive clause. I mean to speak of the population
of Alaska Natives who do not share institutional assumptions.
There are many who do share those assumptions and I want to
be careful not to include those in this general concern. My
concern with these institutions is that many Alaska Natives find
participation blocked by ordinary processes of face-to-face
interaction. Building on the work of Erickson (1980; Erickson
and Shultz 1977; in press) and Gumperz (1977a; 1977b; 1978;
Gumperz, Jupp, and Roberts 1979; Gumperz and Roberts 1978;
Gumperz and Tannen 1979), I am concerned with understanding
how in these cases institutions fail to constitute themselves in
a way that includes these Alaska Natives.
The problem with face-to-face interaction. I started with the
idea that face-to-face interaction was discourse, and I have had
 The Rhythmic Integration of Ordinary Talk / 337

problems with that. My reasons for wanting to consider dis-

course to be face-to-face interaction were various. I have
been interested in the discourse of one-year-old children
(Scollon 1976), as well as the discourse of elderly tradition
bearers in nonliterate traditions (Scollon and Scollon in press
a). In looking at gatekeeping encounters, I have been inter-
ested in those critical few minutes of face-to-face interaction
in which major life choices are made. I have also kept in mind
Bateson's constant reminders that human communication is
originally and primarily mammalian communication and that means
face-to-face communication (Bateson 1972, 1979). The problem
is that much of our institutional communication is carried out in
print, by telephone, on audio- and videotape, on film, and now
increasingly by computer. Studies of face-to-face interaction
can tell us relatively little about such mediated forms of com-
munication. Hence our interest in 'literacy' research. Suzanne
Scollon and I have been engaged in a series of studies in which
we have tried to argue that children in a particular speech com-
munity are socialized into discourse conventions that are com-
patible with (if not actually determinative of) the preferred
medium of communication of that speech community (Scollon and
Scollon 1979; in press a; in press b; S. Scollon in preparation).
Throughout that work we have felt, somewhat uneasily, that
some critical element was missing, some concept that would con-
siderably simplify our discourse about these quite varied forms
of discourse. The notion of tempo is the key to the concept
we were looking for; the concept itself, borrowed from my
chamber music-playing past, is ensemble.
Tempo. My interest in tempo began with an interest in paus-
ing in oral narrative traditions. As I tried to think about ways
of representing Athabaskan oral narratives in a written medium,
the work of Tedlock (1972) and Hymes (1975, 1976, 1977) called
to my attention the great importance of the pause as an ele-
ment of style and even metrics. Other writers have also been
concerned with pausing and rhythms in oral traditions. Scheub
(1977) speaks of the centrality of repetition in oral narratives
as a means of establishing rhythms and Perkins (1980) has
written of the matching of rhythms between speakers in the
Hawaiian oral tradition, this matching being a means of mutually
attesting the degree of agreement among the participants.
I was not, myself, interested in rhythm so much as in paus-
ing. My concern was with understanding the interaction be-
tween a storyteller and his or her audience and the pause ap-
peared to be a critical moment for the interchange between par-
ticipants. This led me to look into quite another literature on
pausing. A number of researchers have had an interest in
pausing as a cognitive issue. In this research, pauses in
speaking have been argued to relate to cognitive processing
(Goldman-Eisler 1968; Pawley and Syder n.d.; Chafe 1979,
1980; Sabin, Clemmer, O'Connell, and Kowal 1979; Welkowitz,
338 / Ron Scollon

Cariffe, and Feldstein 1976). It has been argued that pauses

are the result of difficult lexical items, complex syntactic or
discourse structures, or limitations on processing because of
immaturity.
Another body of research has been more concerned with
affective domains. Chappie (1939) and more recently, such
researchers as Siegman (1979; Siegman and Pope 1972) have
argued that the length of the pauses taken by speakers in
interviews is an accurate indicator of states of anxiety, com-
fort, or interpersonal attraction. This research was of par-
ticular interest to me because of the finding that the pausing
mechanism was apparently tied to a process of attribution.
One group of researchers (Feldstein and Welkowitz 1978;
Feldstein, Alberti, and BenDebba 1979) found that a group of
negative stereotyped attitudes was attributed to speakers who
take longer pauses in speaking. These individuals were stereo-
typed as cold, withdrawn, or even hostile while speakers who
take shorter pauses were stereotyped as warmer, outgoing, or
socially concerned. It was of particular interest to me that
just these stereotypes tend to be attributed in social inter-
actions between Athabaskans and non-Athabaskans, and I have
had an interest in seeing to what extent this attribution of
qualities is tied to pausing phenomena (Scollon and Scollon in
press a; Scollon 1981b).
In these three bodies of research there were three different
kinds of phenomena being addressed: cognitive, affective, and
aesthetic. These were all being addressed at the same point,
the pause or silence in face-to-face interaction. As a further
complication, there was not only disagreement on definitions of
what might constitute a pause; there was absolute disregard
for other views. I was concerned with finding a way to begin
to speak in a coherent manner of all three of these kinds of
phenomena without stumbling over confusions in our notions of
what we might mean by a pause.
Erickson's work on rhythmicity in interpersonal interaction
provided the necessary insight. It was his insight that not
only is talk timed, not only are pauses critical in negotiating
turn exchange, but talk itself is rhythmically timed to a regu-
lar underlying metric or tempo. Erickson has shown that in
ordinary talk, people speak to each other in a regular meter
of regular beats, and time their entrances and exits to the
rhythm of these beats.
Building on the work of Erickson, I have begun to sample a
wide range of talk, at first to see if talk was really as metrical
as he had argued, and then to begin a closer examination of
the phenomenon of tempo itself. Since I have prepared the
details of the data I have used in another paper (Scollon
1981c), here I merely summarize my findings. My sample of
situations includes breakfast table talk, a family gathering with
three generations (including a newborn infant) present, several
university lectures, a radio symphony broadcast announcer, a
 The Rhythmic Integration of Ordinary Talk / 339

radio baseball game, Groucho Marx, Athabaskan tradition

bearers telling traditional narratives in both Athabaskan and
English, potlatch songs, and third grade reading lessons in
rural and urban Alaska and in Hawaii.
The central finding is that talk in apparently all contexts is
timed to an underlying tempo. This tempo is most easily repre-
sented by a simple duple measure. I use 2/4 time as the
representative measure. One might want to ask why this mea-
sure should be duple time and not something else. Hopkins
(1979), in a book on music for nonspecialists, argues that the
fundamental tempo is duple and that this is related to the duple
rhythms of such basic activities as walking, or perhaps the
human heart beat, or perhaps primitive rhythms such as the
in and out of the canoe paddle. Householder's comment quoted
earlier might be a preferable explanation: this tempo is heard
as duple because it is regular and it is the nature of human
perception to hear regular tempos as duple. In a moment I
return to the issue of whether this regularity of tempo is
'real' or 'perceived'.
As in music, the underlying tempo is not to be confused with
the rhythmic patterns superimposed on it. Some speakers
superimpose a pattern of relatively few syllables per beat while
others superimpose a pattern of a very high density. It came
as a surprise to me to find that Groucho Marx, in performing
on his radio show, 'You Bet Your Life', spoke in a very slow
tempo (75.9 beats per minute). What gives the impression of
rapid speech is the very high density of 4.62 words per mea-
sure. To trade on the parallel with music, it can be said that
some speakers speak in quarter notes while others, such as
Groucho Marx, speak in 32nd or even 64th notes. I find it
useful to refer to this phenomenon as 'density' and to treat it
as quite distinct from tempo.
A third phenomenon of relevance is the relative amount of
silence. In my work I am defining silence as beats on which
speech does not occur. Again, I find that our intuitive im-
pressions of 'lots of talk' relate more to the density and the
relative amount of silence than to the tempo. I have further
found it useful to distinguish among silent beats. Basing my
analysis on the work of Gumperz and his colleagues (e.g.
Gumperz and Roberts 1978, Gumperz and Tannen 1979), I
have used the tone group as a unit of analysis. I argue that
silent beats which follow the closure of a tone group are quali-
tatively different from those which occur within a tone group,
and suggest that it is the silent beats following tone group
closure which are interactively and cognitively useful. I have
been tentatively calling these silent beats 'useful silences',
that i s , silences useful for speaker interchange or for hearer
processing.
In this paper it is not my purpose to elaborate on the notions
of density or silence, whether useful or not. I am concerned
with the idea of tempo. What is most striking about tempo is
340 / Ron Scollon

its negotiability. In my whole sample, the range of mean

tempos is from 115.4 beats per minute in a radio spot announce-
ment to 70.2 beats per minute also in a radio program. Within
15 minutes of a conversation around a breakfast table there is
a range in tempos from 103.4 beats per minute to 60.9 beats
per minute. In other words, there is a highly variable range
of tempos both within situations and across situations. It is
also impossible to characterize individual speakers as having a
unique tempo. The variability within speakers is as great as
across situations. While density and silence do to some extent
characterize individual speakers, tempo appears to be used as
the means of negotiating the interaction between speakers. As
in music, it is the tempo that keeps the participants in touch
with each other. It is through the tempo that the performers
integrate their ensemble.
And so I find in ordinary talk, as in music and literary
metrics, there is syncopy (beats anticipated), hemiola (two
different rhythms built upon the same underlying tempo) ,
tempo rubato (arbitrary lengthening or shortening of the beats
with compensatory changes in the length of neighboring beats),
and anacrusis ('pick up' syllables or unaccented syllables pre-
ceding the downbeat). There is no reason to expect ordinary
speech to be less variable and more metronomic than the con-
ventionalized meters of music and poetry.
Processes of acceleration and retardation are central to the
integration of the rhythmic ensemble of two or more speakers.
Speakers time their entrances according to the tempo set by
preceding speakers. After entering in that rhythm, speakers
often accelerate or retard their tempo to establish what is in
effect a new tempo. It is very rare that any speaker will in-
dependently and arbitrarily begin speaking without first con-
firming the established tempo. Children at breakfast bang
their spoons in the prevailing tempo. Radio emcees make their
announcements in the tempo of their theme songs. Sometimes
teachers time their instruction to the tempos of their students
and sometimes they require their students to follow the teach-
er's tempo.
Not only do stressed syllables express this rhythmic matrix;
conversationalists also cough, sneeze, clear their throats, blow
their noses, and laugh in rhythmic ensemble. Often after a
long silence someone clears his or her throat in a gesture
which predicts the following tempo as accurately as a con-
ductor's silent 'one-two' before the orchestra's entrance.
Before picking up the notion of ensemble there are two points
that need to be considered, the question left hanging of how
'real' these tempos are, and a related question of how my use
of tempo relates to that of Chappie and those who have pur-
sued rhythmicity in that tradition of research. In trying to
address the first issue, we have found that in making tran-
scriptions our ability to 'hear' the tempo seems to wax and
wane. Some situations are very easy to hear while others are
 The Rhythmic Integration of Ordinary Talk / 341

much more difficult. In some situations, certain parts are

more easily transcribed than others. It is our perception that
it is the rhythmicity itself that is waxing and waning. As a
test of interjudge reliability, we have made multiple transcripts
of single events and found that across judges it is the same
points that are easily transcribed and the same points in which
difficulty is encountered. While our work at this point is
tentative, we can now suggest that it appears to be at the
boundaries of events that rhythmicity is more difficult to per-
ceive. We now also believe that it is at points where the
original participants themselves are negotiating tempo adjust-
ments that our own ability to perceive rhythms suffers. Our
degree of interjudge reliability gives us reason to believe that
the tempos we are speaking of are 'really there', while the wax-
ing and waning of our confidence in our accuracy leads us to
believe that the tempos themselves are waxing and waning in
their degree of regularity.
Chappie (1980) has provided us with an excellent overview
of his some 50 years of the study of rhythmicity. The con-
cern of Chappie, begun in the thirties, to introduce a high
level of objectivity into his recording has been consummated in
the work of Jaffe and Feldstein (1970), who have used the
computer in conjunction with voice-activated microphones to
achieve a standard of completely automated recording, as they
say, 'without human intervention 1 .
It is important now to consider Chappie's (and his followers')
use of the notion of rhythm, or as they sometimes call it,
tempo. This body of research is looking at the exchange of
speaking turns between two or more participants. Chappie
points out in his recent overview that the model he is using
is that of a relaxation oscillator. This type of rhythmicity,
which is characteristic of biological rhythms from the firing of
neurons to circadian potassium secretions, consists of two
phases, an active phase and a latent phase. During the
latent phase, some form of energy is built up until it reaches
some threshold. When the threshold is reached, there is a
triggering of the release of this stored energy. This release
of energy constitutes the active phase. Chappie very accur-
ately points out the difference between such relaxation oscilla-
tors and harmonic oscillators.
Harmonic oscillators such as clocks and musical strings do
not have a latent and an active phase. These oscillators depend
on a steady input of energy to produce some form of periodic
oscillation, usually represented by a sine wave. To couple
these two types of oscillation is for me an interesting issue.
The pendulum of the clock is a harmonic oscillator but the
spring that runs it is a relaxation oscillator. When it runs
down it must be wound up again. In other words, a clock
succeeds by virtue of the coupling of a relaxation oscillator
and a harmonic oscillator.
312 / Ron Scollon

To get from clocks back to Chappie, Chappie argues that

the give and take of conversational interaction operates as a
relaxation oscillator, the speaker being in the active phase and
the hearer being in the latent phase. The tempo of which
Chappie speaks is the pattern of active, latent, active, latent
in the exchange of speaking turns. It should be clear that
this is a very different use of the notion of tempo from mine.
Tempo as I use the notion is a harmonic oscillation. Chappie's
use of the term 'tempo' is an analogy to music but my use
claims that tempo in ordinary talk is the same phenomenon as
tempo in music. I believe that this difference is what has led
to such radically different methodologies. Chappie's concern
is based on tying social interaction to biological rhythms.
These relaxation oscillators are most appropriately studied by
objective means. The notion of tempo as I am using it requires
methods much more akin to those of musical criticism than of
biological science.
To use a musical analogy, one might make the observation
that Toscanini's tempo in the first movement of the Brahms
First Symphony is much quicker than the tempo taken by
Bruno Walter. The question is: where does one go from that
observation? An interest in the psychophysics of music might
take you into the conductor's biological rhythms or into the
acoustic properties of the hall in which it was performed. An
interest in the quality of the performance would take you into
a very different domain in which the absolute tempo would not
be of nearly as much interest as the relative adjustments of
that quick opening tempo as the movement progresses. It is
in this latter direction that my work is leading me. The idea
of tempo can lead into either the biological foundations of social
interaction, as it has for Chappie and others, or it can lead
into the idea of social ensemble, as it has for me. For what I
am interested in doing, the concept of ensemble provides the
most useful direction. While these two views of social inter-
action are conceptually very different, I do not see them as
any more incompatible than the clock's pendulum and its
spring. The trick will be in learning how to build an escape-
ment.
Ensemble. What I have written to this point is really a pre-
amble to allow me to say a few words about ensemble. As
musicians use the term, ensemble refers to the coming together
of the performers in a way that either makes or breaks the
performance. It is not just the being together, but the doing
together. And so a performance of a string quartet can be
faulted, no matter how impeccably the score has been followed,
if a mutual agreement on tempos, tunings, fortes, and pianos
has not been achieved. Ensemble in music refers to the extent
to which the performers have achieved one mind, or--to favor
Sudnow (1979a, 1979b), one body--in the performance of their
work. Of the elements which contribute to the achievement of
 The Rhythmic Integration of Ordinary Talk / 343

ensemble, tempo is the guiding element. While the note you

are now playing tells me about the loudness and tuning of what
I am now playing, it is the tempo that tells me when you and I
will play our next notes. Tempo is the temporal bond that al-
lows us to move together in real time. It gives us an account
of the immediate past and a basis for predicting the immediate
future. The ensemble of either musical performance or face-
to-face talk depends on this bond of immediate temporal pre-
dictability. Ensemble is what is 'real' about real time.
This gives me a vocabulary for talking about differences
among forms of communication. Such forms as face-to-face
and telephone conversation happen in real time. That is, the
participants must mutually attend to tempo to achieve their en-
semble. Other forms such as unedited audio- and videotaped
communications preserve the real time predictabilities of the
original event but are not heard or seen in that real time.
The listener or viewer may observe the ensemble but not par-
ticipate in it. Still other forms such as writing, whether of
words or musical scores, do not preserve real time predictabili-
ties except to the extent these become conventionalized.
And this is my interest now. What is the relationship betweer
a musical score and its performance? What is the relationship
between a book and its performance? I find it helps me in
thinking about reading and writing to think about that other
form of Western literacy, musical notation. It interests me
that in musical notation there is an elaborate set of conventions
by which the composer's intentions about real time performance
are represented. It also interests me that these elaborate con-
ventions in music developed during the period in which the
idea of 'prose' developed. Do we want to think of prose as
being a nonreal time genre of communication, a genre in which
the width of the page is a more central consideration than the
rhythms of the line? In any event, my interest in ensemble
leads me to ask: how does ensemble become conventionalized
in nonreal time communication?
Institutions and ensemble. Now it remains for me to sketch
in the outlines of how I think this vocabulary of tempo, en-
semble, and conventionalized rhythms helps me to think about
the constitution of institutions in practices of discourse. To
do this I need to make reference to Bateson and Gumperz. In
a number of places Bateson has argued that it was important to
recognize a kind of learning, his 'learning II' or 'deutero learn-
ing' (1972, 1979). This is the sort of learning we recognize as
insightful or creative. This is learning that looks beyond the
situation as given and looks into the context. In its pathologi-
cal forms, according to Bateson, it is also recognizable in
schizophrenia. What I am interested in is the idea that deutero
learning is learning about learning, learning about the contexts
of learning.
344 / Ron Scollon

Bateson tied his ideas about deutero learning to his ideas

about the double bind. In his view, it takes some form of
double bind to produce deutero learning. 1 Briefly, a double
bind is a situation in which one receives two simultaneous and
contradictory messages from someone along with a third con-
straint: that one cannot leave the situation. It is this sort
of bind, according to Bateson, that drives the learner to look
into the context of the situation. In its productive forms this
looking beyond is seen as insight or creativity, in its patho-
logical forms it is seen as schizophrenia.
I want to combine Bateson's perspective with that of Gum-
perz' notion of contextualization cues (Gumperz 1977a, 1977b,
1978). Gumperz, referring to Bateson's notion of metacommuni-
cation, has introduced the idea that the contexts in which com-
munication takes place cannot be taken as given but are con-
structed by the participants in the act of communicating. As
people talk, they must give attention not only to the 'message'
but to its contextualization as well. This perspective of Gum-
perz has proved highly productive in looking at issues of mis-
communication between members of different groups and is, I
believe, a major insight into the nature of face-to-face com-
munication.
The question I have had, however, is: how does contextuali-
zation actually work? What learning mechanism drives people to
pay attention not only to the message but also to the meta-
message? I think now the answer lies in looking back to Bate-
son's double bind. If he was right in arguing that insight into
the contexts of communication comes out of some form of double
bind, and if Gumperz is right that communication critically de-
pends on some looking into contextualization cues, I would argue
that all communication must then depend on some form of double
bind.
For this to be the case two elements would be needed: a
double and contradictory message, and a bonding that makes it
difficult to leave the situation. For the double and contra-
dictory message, I believe the work of Lakpff (1977) and Brown
and Levinson (1978) has suggested the source in the polarity
between two aspects of face. In any communication, the par-
ticipants are faced with the dilemma of respecting the other's
right to be left alone (negative face) and the other's right to
be accepted as a participating member of society (positive face).
This work on politeness phenomena has suggested that any
message must be a carefully concocted blend of the right
amounts of deference and solidarity (Scollon and Scollon in
press a).
What remained was to isolate the bond that ties participants
into the situation to produce the double bind. I believe the
temporal bond of ensemble completes the picture. In the view
I am now taking, it is ensemble which holds participants to-
gether in a mutual attention to the ongoing situation, and it
is the polarity of positive and negative face that forces the
 The Rhythmic Integration of Ordinary Talk / 345

attention to the communication of relationship. These in con-

sort produce a double bind which is the mechanism by which
conversants learn to learn.
This allows me to talk about real time communication and
learning, but what about the communication and learning that
takes place out of real time? Where do insight and creativity
(as well as schizophrenia) come from in such nonreal time com-
munications as prose? The answer that I am now pursuing is
that it comes out of learned conventions for the production of
ensemble. In music, ensemble is conventionalized in scores
and in conventions for performance. In my current view, in-
stitutions are best regarded as conventionalizations of ensemble.
Institutions in this view are the necessary bond to make non-
face-to-face communication work. You do not get literacy with-
out binding social institutions.
Olson (1977, 1980) has argued for the intimate relationship of
literacy and schooling. My current view leads me to believe
that it could be no other way. Literacy as a means of achiev-
ing insightful learning may well be fundamentally inseparable
from the binding social institutions that direct, manage, and
prescribe its use.
Finally, to return to my most general concern, the constitu-
tion of institutions in discursive practices, I believe that the
concept of ensemble will ultimately allow me to argue that our
institutions are not structural entities that impinge on our
achievement of ensemble in face-to-face interaction. These
institutions might better be seen as the conventionalization of
that ensemble itself.
NOTE

A number of colleagues, students, and friends have worked

with me in our studies of tempo and ensemble. I would like
to thank Richard Dauenhauer, Suzanne Scollon, Carol Barn-
hardt, Cecilia Martz, Bob Maguire, and Meryl Siegel for their
help and hasten to point out that none of them should be im-
plicated in what I have presented here.
1. It is unfortunate that the double bind has become so
strongly tied to situations of pathological communication. As
Bateson has argued more recently (1979), there are no mono-
tonic values in biology. That is, in Bateson's view, the double
bind must be viewed as a situation that with optimal values pro-
duces insight and creativity but with maximized values pro-
duces such pathological forms of communication as schizophrenia.
It is critical in my view that we view 'normal' communication
and 'pathological' communication as different states of the same
phenomenon, as Bateson's double bind allows us to do.
346 / Ron Scollon

REFERENCES

Bateson, Gregory. 1972. Steps to an ecology of mind. New

York: Ballantine Books.
Bateson, Gregory. 1979. Mind and nature: A necessary
unity. New York: E. P. Dutton.
Bolinger, Dwight. 1964. Around the edge of language:
Intonation. Harvard Educational Review 34.282-296.
Brown, Penelope, and Stephen Levinson. 1978. Universals
in language usage: Politeness phenomena. In: Questions
and politeness: Strategies in social interaction. Edited by
Esther Goody. New York: Cambridge University Press.
Chafe, Wallace. 1979. The flow of thought and the flow of
language. In: Syntax and semantics, Vol. 12: Discourse
and syntax. Edited by Talmy Givon. New York: Academic
Press.
Chafe, Wallace. 1980. The deployment of consciousness in the
production of a narrative. In: The pear stories: Cognitive,
cultural, and linguistic aspects of narrative production.
Edited by Wallace Chafe. Norwood, N . J . : Ablex.
Chappie, E. D. 1939. Quantitative analysis of the interaction
of individuals. Proceedings of the National Academy of Sci-
ences 25.58-67.
Chappie, E. D. 1980. The unbounded reaches of anthropology
as a research science, and some working hypotheses. Dis-
tinguished lecture for 1979. American Anthropologist 82.4:
741-758.
Erickson, Frederick. 1980. Timing and context in everyday
discourse: Implications for the study of referential and
social meaning. Sociolinguistic Working Paper Number 67.
Austin, Texas: Southwest Educational Development Labora-
tory.
Erickson, Frederick, and Jeffry Shultz. 1977. When is a con-
text? Institute for Comparative Human Development News-
letter 1(2). 5-10.
Erickson, Frederick, and Jeffry Shultz. (in press) Talking
to 'the man': Organization of communication in school
counseling interviews. New York: Academic Press.
Feldstein, Stanley, Luciano Alberti, and Mohammed BenDebba.
1979. Self-attributed personality characteristics and the pac-
ing of conversational interaction. In: Of speech and time.
Edited by Aron W. Siegman and Stanley Feldstein. Hillsdale,
N . J . : Lawrence Erlbaum Associates.
Feldstein, Stanley, and Joan Welkowitz. 1978. A chronography
of conversation: In defense of an objective approach. In:
Nonverbal behavior and communication. Edited by Aron W.
Siegman and Stanley Feldstein. Hillsdale, N . J . : Lawrence
Erlbaum Associates.
Foucault, Michel. 1976. The archeology of knowledge. New
York: Harper and Row.
 The Rhythmic Integration of Ordinary Talk / 347

Foucault, Michel. 1977. Discipline and punish. New York:

Pantheon Books.
Foucault, Michel. 1980a. The history of sexuality, Volume I:
An introduction. New York: Vintage Books.
Foucault, Michel. 1980b. Power/knowledge: Selected inter-
views and other writings 1972-1977. Edited by Colin Gordon.
New York: Pantheon Books.
Goldman-Eisler, F. 1968. Psycholinguistics: Experiments in
spontaneous speech. New York:. Academic Press.
Gumperz, John. 1977a. Sociocultural knowledge in conversa-
tional inference. In: Georgetown University Roundtable on
Languages and Linguistics 1977. Edited by Muriel Saville-
Troike. Washington, D.C.: Georgetown University Press.
191-212.
Gumperz, John. 1977b. The conversational analysis of inter-
ethnic communication. In: Interethnic communication.
Edited by E. Lamar Ross. Proceedings of the Southern
Anthropological Society, University of Georgia. University
of Georgia Press.
Gumperz, John. 1978. The retrieval of sociocultural knowl-
edge in conversation. Paper presented at the American
Anthropology Association, November, 1978.
Gumperz, John, T. C. Jupp, and Celia Roberts. 1979.
Cross-talk: A study of cross-cultural communication. A
film and notes. London: BBC and the National Centre for
Industrial Language Training.
Gumperz, John, and Celia Roberts. 1978. Developing aware-
ness skills for interethnic communication. Middlesex, Eng-
land: National Centre for Industrial Language Training.
Gumperz, John, and Deborah Tannen. 1979. Individual and
social differences in language use. In: Individual differ-
ences in language ability and language behavior. Edited by
Charles J . Fillmore, Daniel Kempler, and William S-Y. Wang.
New York: Academic Press.
Hopkins, Antony. 1979. Understanding music. London:
J. M. Dent.
Householder, Fred. 1957. Accent, juncture, intonation and
my grandfather's reader. Word 13.234-245.
Hymes, Dell. 1975. Folklore's nature and the Sun's myth.
Journal of American Folklore 88.350:345-369.
Hymes, Dell. 1976. Louis Simpson's 'The deserted boy'.
Poetics 5.119-155.
Hymes, Dell. 1977. Discovering oral performance and mea-
sured verse in American Indian narrative. New Literary
History 8.431-457.
Jaffe, Joseph, and Stanley Feldstein. 1970. Rhythms of dia-
logue. New York: Academic Press.
Ladd, D. Robert. 1980. The structure of intonational mean-
ing: Evidence from English. Bloomington: Indiana Uni-
versity Press.
348 / Ron Scollon

Lakoff, Robin. 1977. Language and society. In: A survey

of applied linguistics. Edited by R. Wardaugh and H. Brown.
Ann Arbor: University of Michigan Press.
Lenneberg, Eric. 1967. Biological foundations of language.
New York: Wiley.
Liberman, M., and A. Prince. 1977. On stress and linguistic
rhythm. Linguistic Inquiry 8.249-336.
Olson, David. 1977. The languages of instruction: On the
literate bias of schooling. In: Schooling and the acquisition
of knowledge. Edited by R. C. Anderson, R. J. Spiro, and
W. E. Montague. Hillsdale, N . J . : Lawrence Erlbaum Associ-
. ates.
Olson, David. 1980. Some social aspects of meaning in oral
and written language: In: The social foundations of lan-
guage and thought: Essays in honor of Jerome S. Bruner.
Edited by David Olson. New York: W. W. Norton.
Pawley, Andrew, and Frances Syder. ( n . d . ) Sentence formu-
lation in spontaneous speech: The one-clause-at-a-time hy-
pothesis. Unpublished MS.
Perkins, Leialoha Apo. 1980. Toward a modern Pacific
aesthetics for poetry. In: Kingdoms of the heart. By
Leialoha Apo Perkins. 2322 Maile Way, Honolulu:
Kamalu'uluolele.
Sabin, Edward J . , Edward J. Clemmer, Daniel C. O'Connell,
and Sabine Kowal. 1979. A pausological approach to speech
development. In: Of speech and time. Edited by Aron W.
Siegman and Stanley Feldstein. Hillsdale, N . J . : Lawrence
Erlbaum Associates.
Scheub, Harold. 1977. Performance of oral narrative. Paper
presented at the Annual Meeting of the American Association
for the Advancement of Science, February, 1977, Denver,
Colorado.
Scollon, Ron. 1976. Conversations with a one year old: A
case study of the developmental foundation of syntax.
Honolulu: University Press of Hawaii.
Scollon, Ron. 1981a. Gatekeeping: Access or retention?
Working Papers in Alaska Native Studies, No. 1. University
of Alaska, Fairbanks: Alaska Native Studies Program.
Scollon, Ron. 1981b. The machine stops: Silence in the
metaphor of malfunction. Paper presented at the session,
The functions of silence. Georgetown University, March
1981. To appear in: The functions of silence. Edited by
Muriel Sayille-Troike and Deborah Tannen.
Scollon, Ron. 1981c. Tempo, density, and silence: Rhythms
in ordinary talk. University of Alaska, Fairbanks: Center
for Cross-Cultural Studies.
Scollon, Ron, and Suzanne B. K. Scollon. 1979. Linguistic
convergence: An ethnography of speaking at Fort Chipewyan,
Alberta. New York: Academic Press.
 The Rhythmic Integration of Ordinary Talk / 349

Scollon, Ron, and Suzanne B. K. Scollon. (in press a)

Narrative, literacy, and face in interethnic communication.
Norwood, N . J . : Ablex.
Scollon, Ron, and Suzanne B. K. Scollon. (in press b)
Cooking it up and boiling it down: Abstracts in Athabaskan
children's story retellings. In: Spoken and written lan-
guage: Exploring orality and literacy. Edited by Deborah
Tannen. Norwood, N . J . : Ablex.
Scollon, Suzanne B. K. (in preparation). Reality set, social
interaction, and linguistic change. Doctoral dissertation.
University of Hawaii.
Siegman, Aron W. 1979. The voice of attraction: Vocal
correlates of interpersonal attraction in the interview. In:
Of speech and time. Edited by Aron W. Siegman and Stan-
ley Feldstein. Hillsdale, N . J . : Lawrence Erlbaum Associates.
Siegman, Aron W., and B. Pope. 1972. The effects of ambi-
guity and anxiety on interviewee verbal behavior. In:
Studies in dyadic communication. Edited by A. W. Siegman
and B. Pope. New York: Pergamon Press.
Sudnow, David. 1979a. Ways of the hand. New York:
Bantam Books.
Sudnow, David. 1979b. Talk's body. New York: Alfred A.
Knopf.
Tedlock, Dennis. 1972. On the translation of style in oral
narrative. In: Toward new perspectives in folklore. Edited
by Americo Paredes and Richard Bauman. Austin: University
of Texas Press.
Welkowitz, J . , G. Cariffe, and Stanley Feldstein. 1976. Con-
versational congruence as a criterion of socialization in chil-
dren. Child Development 47.269-272.
 HARD WORDS: A FUNCTIONAL BASIS
FOR KALULI DISCOURSE

Steven Feld
Bambi B. Schieffelin
University of Pennsylvania

0. Introduction. This paper is concerned with cultural con-

structions that frame appropriate Kaluli discourse and with some
kinds of discourse that operate within that frame. We begin
with ethnographic and metalinguistic materials scaffolding the
Kaluli notion of 'hardness', the Kaluli conception of language
and speech, and the specific idea of 'hard words'. These con-
structs illustrate the pervasive character of a Kaluli distinction
between 'langue' and 'parole'. Based on these systematic notions
of language form, socialization, and behavior we analyze some
situated discourse examples that indicate both how these cul-
tural constructions are learned and how they operate in every-
day interactions.
0.1 People and place. The Kaluli people are part of a popu-
lation of about 1,200 who live in several hundred square miles
of tropical rain forest just north of the slopes of Mt. Bosavi,
on the Great Papuan Plateau of Papua New Guinea (E. L.
Schieffelin 1976). They are one of four culturally identical
but dialectically different subgroups who collectively refer to
themselves as Bosavi kalu 'Bosavi people'. The Kaluli reside
in longhouse communities made up of about 15 families (60-90
people), separated by an hour or so walk over forest trails.
Subsistence is organized around swidden horticulture, the pro-
cessing of wild sago palm to make a staple starch, and hunting
and fishing. In broad terms, Kaluli society is highly egali-
tarian, lacking in the 'big man' social organization characteristic
of the Papua New Guinea Highlands. Men and women utilize ex-
tensive networks of obligation and reciprocity in the organiza-
tion of work and sociable interaction.
Kaluli is one of four dialects of Bosavi, a non-Austronesian
verb-final ergative language. Most speakers are monolingual.

350
 Hard Words: A Functional Basis for Kaluli Discourse / 351

While Tok Pisin (Neo Melanesian), is known by some younger

men, it is almost never heard in daily discourse. Recently
introduced literacy programs have affected few people.
Kaluli everyday life is overtly focused around verbal inter-
action. Talk is thought of and used as a means of control,
manipulation, expression, assertion, and appeal. It gets you
what you want, need, or feel owed. Extensive demarcation of
kinds of speaking and speech acts further substantiate the ob-
servation that Kaluli are energetically verbal; talk is a primary
way to be social, and a primary indicator of social competence
(B. B. Schieffelin 1979; B. B. Schieffelin and Feld 1979).
More generally, the realm of sound yields the most elaborated
forms of Kaluli expression. In the tropical forest and village
longhouse it is difficult to find auditory privacy or quiet.
Greetings, comings and goings, announcements, arguments,
meetings, and all soundings are projected into aurally public
space. No comparable variety, salience, or exuberance exists
for Kaluli visual or choreographic modes of expressions.
1. 'Hard 1 , 'words', 'hard words': Putting a construction on
life and language

1.1 Halaido 'hard 1 . Halaido 'hard' is a pervasive Kaluli

notion that applies broadly in three cultural-semantic domains.
The first is growth and maturation, where the socializing inter-
actions in the acquisition of language are what 'makes (it) hard'
(halaido domeki); the development of strong teeth and bones in
the uncoordinated infant who is 'without understanding' (asugo
andoma) is a process of 'hardening' (halaidan). In these cases,
the process of becoming 'hard' is a literal and metaphoric con-
struct for physical and mental development and for cultural
socialization. A second domain for halaido is the fully adult
consequence of this maturation process. A kalu halaido or
'hard man' is one who is strong, assertive, and not a witch;
a major component in this person's projection of his 'hardness'
is the acquisition and command of to halaido 'hard words', the
fully developed capacity for language. x The final area in which
halaido is prominent is dramatic style. In ceremonial perform-
ance, songs are intended to be evocative and make the audience
weep. The climax in the development of aesthetic tension,
where the manner of singing and the textual elements coalesce,
is what promotes the 'hardening' (again, halaido domeki) of a
song. A performance that does not' 'harden' will not move
listeners to tears and will not be considered successful.
Furthermore, the ability to 'harden' a song is an important
compositional (particularly in textual craft) and performative
skill.
The cultural construction and prominence of halaido in Kaluli
growth, adulthood, and presentational style can in part be
traced to an origin myth which tells how the world was once
muddy and soft; a megapode and Goura pigeon together stamped
352 / Steven Feld and Bambi B. Schieffelin

on the ground to make it hard. Like the hardening of the

land which symbolizes the necessity of physical and geographi-
cal formation, the hardening of body, language, character, and
dramatic style symbolizes the necessity of human socialization
in order to develop cultural competence.
One term used in opposition to halaido is taiyo 'soft'. Within
this oppositional frame, taiyo is 'soft' in the senses of: mushy
foods, things which decay and rot, or debilitation. It signifies
a stage in the process" of decay, and all connotations with this
state are unpleasant. Food taboos constrain the eating of cer-
tain soft substances (such as eggs) while young lest one not
'harden'. Children, moreover, do not eat the meat of certain
birds who have 'soft' voices or redundant and otherwise strange
calls, lest their language not harden and they grow up to
speak unintelligible sounds. (On the topic of children's food
taboos vis-a-vis hardness, see B. B. Schieffelin 1979:62-65,
and Feld 1982:Chapter 2.) Similarly tabooed are all animal and
vegetable foods which are yellow; like the leaves of plants,
things yellow as they decay. Witches are said to have yellow
soft hearts, while the hearts of 'hard men' are dark and firm
(E. L. Schieffelin 1976:79, 128). In short, the passage from
'hardness' to 'softness' is undesirable, synonymous with debili-
tation, vulnerability, and decay, states which must be avoided.
The desired progression in all things is from softness (infant)
to hardness (adult); once hard in body, language, and dra-
matic style, Kaluli must stay that way.
Another term utilized in opposition to halaido is halaidoma
'unhard', 'without hardness', formed by the word 'hard' plus
the negative particle -ma. Something which is potentially hard
--or which should be, but is not--is 'unhard'. For instance,
when one of us was learning the Bosavi language (SF), his
verbal behavior was judged as to halaidoma and his mistakes
greeted assuringly with towo halaidesege 'when your language
has hardened 1 . Never was this speech ability referred to as
*to taiyo 'soft words', a construction which was laughed at
when suggested. 'Soft words' is neither an appropriate nor
utterable phrase; language is either 'hard' or 'unhard', i.e.
in the process of hardening, or in the state of becoming un-
hard, as in sickness or delirium.
1.2 To 'words/language 1 . Kaluli observe a langue/parole
distinction. This is marked by the distribution of the terms
to and tolema 'words', 'language' and imperative 'talk words/
language' (langue) and sama imperative 'speak' (parole). 2
To and tolema refer to the systematic form of language or its
capacity; in contrast, sama refers to the manner or act of
speaking. To illustrate langue we examine the items in (1).
(1) Bosavi to Bosavi language
bali to 'turned over words' = systematic linguistic
irony/euphemism, metaphor, or obfuscation
 Hard Words: A Functional Basis for Kaluli Discourse / 353

malolo to 'narrated/told words' (= myths and stories)

mugu to 'taboo words'
In these examples, the noun to refers to the system or form of
talk. All of these nominal forms can be followed by the ha-
bitual verbs solan 'one speaks/says', asulan 'one understands',
or dadan 'one hears'. These indicate that one may speak,
understand, or hear any of these systems of talk or different
languages. The use of tolana contrasts with constructions us-
ing sama ('parole'), for instance; (here with sama in the pres-
ent habitual form salan).
(2) wonoli-salan one speaks secretly, stealthily
tede-salan one speaks in a deep voice
hala-salan one speaks with mispronunciations
In these instances (and a multitude of similarly constructed
ones), salan concerns the behavior of speaking, or some de-
scription of how speaking is performed.
From our analysis the Kaluli theory of language and speech
is one in which to 'words' are the prime substance of language;
tolana is the doing or speaking of words.
Figure 1.

to
'words' 'language'

tolema
'do language', 'speak/say words'

to
'words'
ele sama
'like that' 'speak/say'
(imperative)

As can be seen in Figure 1, tolema is formed by adding to

'words' and -elema, imperative 'do/say/speak like that'. The
item elema is the contracted form of ele sama, 'like that' plus
the imperative 'say/speak'. Many Kaluli verbs are formed in
this way, by adding a substance or onomatopoeic root to
-elema. For instance, the verb for 'weep' is yelema, composed
by contracting the onomatopoeic representation of the sound of
weeping, ye, and the imperative 'say/speak like that' (Feld
1982: Chapters 3 and 4 contain materials on these formations in
354 / Steven Feld and Bambi B. Schieffelin

Kaluli metalinguistics; B. B. Schieffelin 1979: Chapter 3 con-

tains materials on elema and elz sama in interaction).
Everyday interactions make clear that the contrast between
these two notions is salient for Kaluli. To take a simple in-
stance, SF was once questioning some men about the fact that
certain birds are claimed to speak some Bosavi words. He
asked about bolo, the friar bird in the Kaluli myth about how
birds received human tongues.
(3a) Bolo-wo, Bosavi to salano?
'As for bolo, does he speak Bosavi words/language? 1
Two answers followed:
(3b) Bosavi to salan.
'He speaks Bosavi words/language.'
Mugu tolan.
'He talks taboo language.'
The first response is the usual specific one ('parole'), while
the second was a response from a Christian referring to the
way the systematic form of boZo's talk consists of words Chris-
tians consider 1 taboo ('langue'). Yet in the context of listening
to a tape recording of specific calls by bolo, the same man
noted, mugu to salab 'he is speaking/saying taboo words/lan-
guage' , implying: in that specific instance.
In everyday talk the distribution of inflected verb forms for
sama and tolema further exemplifies the importance of speaking
as a situational act and language as a fundamental capacity.
Part of the paradigm includes the items in (4).
(4) tolema sama imperative immediate
tolebi selebi imperative future
tolomeno selemeno future first person
tolab salab present third person
tolan salan habitual third person
but:
•tolol solol present first person
*tolo siyo past
The fact that the present first person form and past form are
blocked for tolema is consistent with the general nature of to
as 'words/language' and tol&na as 'talk'. Moreover, *tolol
contrasts with:
towo solol 'I speak/say words/language'
towo motolan 'It doesn't talk words/language' (can be
said only about animals whose communication is assumed
 Hard Words: A Functional Basis for Kaluli Discourse / 355

to be a system based upon a substance other than

'words'.)
*ele tolo
*el£ tolema
ele siyo 'said like that'
ele sama 'say like that'
Use of 'like that' is also blocked with to and tolema because of
lack of reference to a specific situation or context.
The metalinguistic area provides further examples of the
distribution and further evidence for the cohesiveness of ways
of describing related modalities of soundmaking. In one exam-
ple across modalities, gese, the root of gesema 'make one feel
sorrow or pity' is only blocked for tolema as illustrated in (5).
(5) gese-salan one speaks sadly (plaintively; with descend-
ing intonation)
gese-yelan one weeps sadly (plaintively; with descend-
ing intonation)
gese-holan one whistles sadly (plaintively; with de-
scending intonation)
gese-molan one sings sadly (plaintively; with descend-
ing intonation)
but:
*gese-tolan inappropriate because gese describes the
manner of speaking and is not applicable
to the system or capacity of talk
In these cases the verbs deal with modes of soundmaking while
the adverbs describe the manner of performance; like other
verbs of soundmaking, sama refers to the behavioral aspect of
speech; to and tolema refer to its form and capacity.
A major area of metalinguistic denomination is marked by use
of sa. By itself, sa means 'waterfall'; the term also prefixes
all verbs of soundmaking to indicate that the sound has an 'in-
side' or text. This usage derives from the metaphor that texts
are composed 'like a waterfall flowing into a waterpool'; the
sound is 'outside' and the text, like a waterfall, is the part
that flows down and inside. Verbs of soundmaking turn into
musical or compositional terms when prefixed by sa in this way,
as in (6) (with verbs all in a present habitual form).
(6) salan 'one speaks' sa-salan 'one speaks inside the
words/one speak poetically'
ydan 'one weeps' sa-yelan 'one weeps with text'
holan 'one whistles* sa-holan 'one whistles with words
in mind'
molan 'one sings' sa-molan 'one sings inside' i.e.
'one composes'
356 / Steven Feld and Bambi B. Schieffelin

but:
tolan 'one talks' *sa-tolan inappropriate because
one cannot have an inner text
to language capacity
Sa-salan, sa-sama, and sa-siyo all indicate an intention to mean
more than what is said. To and tolema do not participate in
this paradigm; *sa-to and *sa-tolema are blocked because there
cannot be an 'inside' or inner text to the capacity or system of
language. 'Insides' are specific and contextual, related to
situated performances only.
1.3 To halaido, 'hard words/talk/language 1 . Given the cul-
tural importance and pervasiveness of 'hardness' as a construct
underlying mature social process and capacity, and the role of
'hardness' in the distinction Kaluli observe between langue and
parole, we turn to the specific importance of 'hard words'. In
the most general sense, to halaido is the system of and capacity
for grammatically well-formed and socially appropriate language.
It is the substance of what Kaluli adults know and act upon in
their verbal behaviors. It is what is normally acquired, the
competence to perform, what Kaluli should 'have in mind' when
they speak. The opposite of to halaido is not *to taiyo 'soft
words'; when language is in the process of forming, it is to
halaidoma 'unhard words'.
Nevertheless, when asked if there is any language which is
neither hard nor in the process of becoming hard, Kaluli3 indi-
cate that such is the situation for the language of song. This
is a special poetic system called o be go no to 'bird sound
words'. Songs are said to be composed and sung from a bird's
point of view, arid not a human one. They achieve their plain-
tive quality and ability to move people to tears in this way be-
cause birds are the spirit reflections of Kaluli dead. Song
language is thus not human and hard, but birdlike, sad, senti-
mental, reflective.
The contrast between to halaido 'hard words' and obe go no to
'bird sound words' is basic. 'Hard words' are assertive and
direct language forms which engage speakers in face-to-face
talk that is interactive and mutual, and are intended to get
speakers what they want or need out of social situations. On
the other hand, 'bird sound words' are reflective and nostalgic,
and are supposed to make a listener empathize with a speaker's
message without necessarily or generally responding to it
verbally. 'Bird sound words' involve linguistic means that
communicate affect by revealing the speaker's state of mind
and moving a listener to feel sympathy for that state.
It is not the case that the difference between these two con-
structs is simply one of referential/expressive or ordinary/
nonordinary. Certain message forms and contents can appear
in either; the different way that messages are interpreted
 Hard Words: A Functional Basis for Kaluli Discourse / 357

depends on judgments about intention deriving from contextual

constraints, as well as from placement in an ongoing textual
chain. Consider example (7).
(7) Dowo ge oba hanaya?
'Father, where are you going?1
There are numerous daily contexts in which this might be
uttered by a person to someone called 'father'. Depending on
the intonational contour, the utterance could be a request for
information, a challenge, or a rhetorical question--all of which
might be benign or threatening. However, when we shift from
conversation to song usage, the implications shift radically and
the audience immediately knows that the message is that a
father has died and left someone behind. The person asking
the question is in the resultant state of abandonment and ap-
pealing to the audience for sympathy. The form of the words
is 'hard' in the sense that they are well formed and could be
uttered in appropriate daily situations. However, in a song
context the words show their 'inside', set, and this is why they
are 'bird sound words'. What is implied in the saying context
and manner of saying is more important than the referential
equivalents of the words which are said.
2. Learning and speaking 'hard words'

2.1 Imperatives. To exemplify how the process of learning

the model for discourse is the learning of 'speaking' and 'hard
words', we turn to some discourse examples from tape-recorded
family interactions. While these examples involve much adult-
child speech, the same forms are used among adults (though
perhaps not as frequently or with the same concentration in an
episode, since child-adult speech involves more direction and
repetition). Imperatives form an important class of examples
since they provide major instances of learning by instruction.
In addition to indicating specific rhetorical strategies for getting
what one wants, imperatives teach directness, control, speaking
out, sequencing, and cohesion in the flow of talk.1* This is
further strengthened by the unambiguous relation of speaker/
addressee in imperatives, as evidenced by frequent deletion of
the optional subject pronoun or a vocative. Moreover, impera-
tives are favored forms for requesting both actions and objects
because Kaluli does not express requests indirectly with forms
like 'would you, could you'. Additionally, language structure
provides great flexibility, range, and specificity for impera-
tives. For example, Kaluli morphologically differentiates pres-
ent and future imperative, marking iterative and punctual
action, with various degrees of emphasis or seriousness, all
of which can be indicated for single, dual, or plural subjects.
358 / Steven Feld and Bambi B. Schieffelin

In the examples that follow, sama, elema, and to/tolema

clearly distribute according to whether specific instances of
speaking or general prescriptions to talk are encouraged.
For the Kaluli infant, involvement in verbal interactions
starts about a week after birth. A mother holds her infant so
that it faces another child; she moves the infant as one might
a ventriloquist's dummy, speaking for it in a nasalized falsetto
voice. Her speech is well formed and clearly articulated, with
the complexity of a 4-year-old's speech. The child to whom
the baby is 'speaking' engages in conversation directed to the
baby for as long as interest can be maintained. Through these
verbal interactions the baby is presented as a person, an indi-
vidual, and is made to appear more independent and mature
than it actually i s , largely through the mother's speech and
her manipulation of the infant's body. These 'three-party'
interactions, as well as the much less frequent direct talk be-
tween mother and infant, are said to 'give words/language
understanding or meaning' (to samiab).
The use of language and rhetoric in interaction are the major
means of social manipulation and control in Kaluli life. Thus,
one of the most important achievements in childhood is to learn
to speak Kaluli effectively to a variety of individuals with whom
one participates in everyday activities. Kaluli say that lan-
guage (to) has begun once the young child uses two critical
words, no 'mother' and bo 'breast'. Children who only name
other people, animals, or objects are said to do so 'to no pur-
pose' (ba madali); they are not considered to have begun to
use language. This is evidence for the essentially social view
of language taken by the Kaluli, a view which emphasizes not
only the learning and using of words per se, but the use of
specific words to express the first social relationship a person
has, namely, the mother-child relationship mediated by food
from the breast. This is a basic theme in Kaluli social life.
The giving and receiving of food is a major way in which re-
lationships are mediated and validated (E. L. Schieffelin 1976;
Feld and B. B. Schieffelin 1980).
Once a child has begun to use the words 'mother' and
'breast', Kaluli begin to 'show language' (to wzdan). Kaluli
say that children must be 'shown language' by other Kaluli
speakers, principally by the mother. Kaluli use no baby talk
lexicon as such, and claim that children must hear to halaido
'hard language', if they are to learn to speak correctly. When
a Kaluli adult wants a child to say something in an ongoing
interaction, a specific model is provided for what the child is
to say, followed by the imperative 'say like that' elema. The
word elema is a contraction of ele 'like this /that' and sama
'say/speak' present imperative. While the adult occasionally
asks the child to repeat utterances directly back to him or
her, correcting the child's language or initiating a game, the
vast majority of these directives to speak concern instructions
 Hard Words: A Functional Basis for Kaluli Discourse / 359

to the child to say something to someone else. 5 An example of

this type of interaction is given in (8).
(8) Meli (female, 25 months) and her mother are in the
house. Mother has tried to get Meli into an elema rou-
tine, and Meli has been distracted. 6 Finally, she settles
down. Grandfather is not in sight.
1. Mother-•Meli: Sit on this. (Meli does)
now speak words.
ami to ena sama
2. Mother -•Meli —»•• Grandfather:
Grandfather! elema
(softly) 3. Grandfather/

4. Mother -•Meli:
speak more forcefully/loudly,
ogole sama
(louder) 5. Grandfather! /

6. Mother -•Meli — • • Grandfather:

I'm hungry for meat! elema
7. I'm hungry for meat! /
(This continues for 14 turns, which consist of requests
to grandfather to get different foods.)
In line 1, Meli's mother encourages her to 'speak words/
language' {to sama), to engage verbally with someone. She
has the addressee and utterances in mind, which she will pro-
vide followed by the imperative 'say like that' elema. The
addressee is named, but Meli does not call out loudly enough,
and in line 4 her mother tells her how to speak, using sama.
This is followed by a specific utterance, and another directive
to speak, with which Meli complies. Thus, to sama refers to
the activity of speaking and saying, where a sequence of
utterances are followed by elema. While in this episode the
addressee, Grandfather, is not in the vicinity and therefore
does not respond to Meli's requests, the majority of such epi-
sodes involve responses from a third person to the child's di-
rected utterances. These sequences often involve extensive
and cohesive turns of1 talk. This 'showing the language1 helps
the language 'harden (halaido domeki) and thus is consistent
with the general goals of socialization and development: the
360 / Steven Feld and Bambi B. Schieffelin

achievement of 'hardening' which produces, an individual who is

in control of himself or herself, and who is capable of verbally
controlling others.
Directives to speak, using the imperative, occur in a variety
of speech situations, but are most frequent in those involving
shaming, challenging, and teasing. The interactional sequence
in (9) illustrates several of the rhetorical strategies used in
such situations, and demonstrates the sensitivity young chil-
dren develop about the consequences of what they say.
(9) Wanu (male, 27 months), his sister Binalia (5 years),
cousin Mama (3£ years), and Mother are at home. The
two girls (Mama and Binalia) are eating salt belonging to
another child.
1. Mother^Wanu
Whose is it?! dema
2. Whose is it?! /
3. Is it yours?! dema
4. is it yours?! /
5. Who are you?! elema
6. who are you?! /
7. Binalia-*-Wanu —»-•- Mother:
Is it yours?! dema
8. is it yours?! /
inalia
9. Mother -*-Wanu — » ^ ? . :
Mama
It's mine! dema
10. Mama-*-Binalia: Don't speak like that!
dedo selasabo!
Rhetorical questions, such as those found in lines 1, 3, 5,
and 7 in example (9), are frequent in family interactions in-
volving the use of zl&na. They are intended to shame the
addressee so that he or she will terminate undesirable behavior.
Kaluli frequently utilize teasing, shaming, and other means of
verbal confrontation that focus on an addressee who cannot
answer rhetorical questions without the admission of fault.
These strategies of confrontation and their component rhetori-
cal skills set the tone of many interactions, while the use of
directives (such as 'put the salt away') or physical intervention
is much less common. Although children may challenge adults
in certain situations (and are encouraged to do so), here Mama
(age 3i) tells Binalia 'don't speak like that', referring to
Binalia's attempt to get Wanu to challenge his mother. When
asked about that utterance, Kaluli said that Mother could get
angry and take the salt away. Thus, even children evidence
a sensitivity to how language is being used in interactions,
 Hard Words: A Functional Basis for Kaluli Discourse / 361

sensing t h e consequences of particular kinds of talk. This

further serves t h e functional importance of directly p u t t i n g
the burden on t h e a d d r e s s e e .
The use of elema in these interactions i s consistent with t h e
mother's treatment of h e r preverbal infant, in which she p u t s
words into his mouth. She pushes h e r young language-learn-
ing child into social interaction, providing t h e words he cannot
say o r may not b e interested in saying. This practice p r o -
vides t h e opportunity for t h e child to acquire t h e verbal skills
that a r e needed later o n , when mother has h e r next baby and
t h e child becomes part of a peer g r o u p . It i s t h e ability not
only t o repeat rhetorical questions such as 'who a r e you?! f ' I s
it y o u r s ? ! ' b u t to u s e them, spontaneously in t h e appropriate
c o n t e x t s , that lead Kaluli to comment about a young child, to
halaido momada salab ' h e / s h e is s t a r t i n g to speak hard lan-
guage'.
It i s important t o note that throughout interactions using
derma, assertion prevails. In teaching language, mothers are
teaching their small children assertion itself. For Kaluli this
implies s t r e n g t h and independence. In interactional terms this
means to request with imperatives, to challenge and confront,
and to say something powerful so others will bend or give.
Mothers never u s e dema to i n s t r u c t their children in begging,
whining, or appealing to o t h e r s for sympathy. In learning t h e
t y p e s of things one says with dema, Kaluli children a r e learn-
i n g culturally specific ways in which to b e t o u g h , i n d e p e n d e n t ,
and a s s e r t i v e , which reinforces t h e cultural value of acting in
a direct, controlled manner.
In addition to t h e imperatives sama and dema, t h e imperative
tol&na is also used in conversations. In contrast to t h e act of
speaking (sama), u s e of tolema calls attention to t h e importance
of verbal interaction as an activity in which children a r e e n -
couraged to participate.

(10) Meli (female, 25 months) i s with h e r father in t h e

house. She i s not involved in any activity. Mama
is not in s i g h t .

1. Father-*-Meli •••Mama:
Mama! call o u t .
holema
2. Mama/
3. Come and talk together with
me! elema
neno to tomeni meno!
4. come and talk
together with
me/
362 / Steven Feld and Bambi B. Schieffelin

(There is no response. Seeing another child)

5. Father-•• Meli: Now you and Babi go in order
to talk.
ami Babi gain tome'hamana
6. (Meli puts marble in her mouth) Take out the
marble! After taking it out with your hand,
. you will talk!
to tolebi
In this episode, Father is trying to get Meli established in a
verbal activity, made explicit in line 3 as a directive (dema)
to invite Mama to come and talk (to tomeni meno). The word
eletna marks what is specific to be said, and the concatentated
form (tomzni 'in order to talk' + meno 'come' imperative) marks
the general activity to take place. A similar concatenated form
is used in line 5, this time directing Meli to go in order to
talk. And finally (line 6), to tolebi (future imperative) is 7used
to indicate what Meli should do, but not what she will say.
In this situation, talking is being established as a way to
engage and be social. Parents assume the importance of inte-
grating children into adult verbal activities and additionally
encourage the organization and maintenance of verbal exchanges
among children themselves. This establishes talk as a topic of
talk, instructions to talk as instructions to be social, and talk
as a modality that promotes social cohesion.
In addition to both the desire and the necessity to develop to
halaido, children must learn to converse, to kudan 'one puts
language/words together'. The expression i kuduma 'put wood
together', is used to tell someone how to build a successful
fire, by taking a stick with an ember, putting another stick to
it to make contact and transferring the heat. Just as putting
wood and sticks together makes a successful fire, talk must
also be put together to be successful. Commenting on the
language of a 2-year-old who wasn't collaborating with or
building on the other's utterances, a Kaluli said, to mokudab
'he doesn't put language together'. The same expression was
used with regard to a conversation between two adults, in
which they had not agreed on what they were, in fact, talking
about.

(11) As father is leaving Meli (age 25 months)

my child! as for me, I'm going to converse.
niyo to kudumeni
You stay here.
ge ya tebi.
The use of to kudan in these contexts indicates the importance
Kaluli attach to verbal interactions which are mutual, collabora-
tive, and cohesive.
 Hard Words: A Functional Basis for Kaiuli Discourse / 363

As has been seen, utterances directing a child to use lan-

guage (tolema) and specifying what to say (elema) and how to
say it (sama) are used to promote and support young children's
involvement with others in a variety of everyday interactions.
The Kaiuli say that without this kind of direction children would
not learn what to say and how to say it. The idea is that after
a child is 'shown' what to say, he or she will spontaneously use
language to respond, to initiate, sustain, and control verbal
interactions. However, children themselves initiate and par-
ticipate in language interactions that are unlike any that their
parents have shown them. Many of these exchanges are termi-
nated by Kaiuli mothers when they feel that these could impede
language development or promote an undesirable effect. These
situations provide an opportunity to examine what is and is not
acceptable language behavior for small children, and the cul-
tural reasons for these differences.
(12) Meli (30i months) and her cousin Mama (45 months) are
at home with Meli's mother, who is cooking and talking
to several adults. Mama initiates a sequence of word
play involving Meli which is marked by repetition, high
pitch, staccato delivery, and exaggerated prosodic con-
tours. After 10 turns this dissolves into sound play
marked by overlap within turn pairs, higher pitch,
vowel lengthening and shifting, and repetition. This
continues for 15 more turns, at which point Meli's
mother suddenly turns to the girls and says in a loud,
authoritative voice:
Wai! Try to speak good talk! This is bird talk!
Wai! to nafa se seleiba! obe towo we!
The girls suddenly become quiet.
The mother's abrupt termination of the children's verbal/vocal
interaction was not due to mild irritation caused by the noise
these girls were making, since similar sound levels caused by
other kinds of verbal activity would never have prompted this
reaction. Her response, which was consistent with that of
other Kaiuli mothers in similar situations, grows out of Kaiuli
ideas about language development and the broader notion of
taboo.
As mentioned earlier, Kaiuli have very definite ideas about
appropriate verbal behavior for language learning children.
When asked about this word /sound play, Kaiuli said it had no
name and was 'to no purpose'. Purposive language is encour-
aged in interactions and the vocalizations between Meli and
Mama violated these cultural expectations.
However, in addition to their ideas about how a young child's
language should sound, Kaiuli say that children and birds are
361 / Steven Feld and Bambi B. Schieffelin

connected in a number of complex ways (Feld 1982: Chapter 2).

In addition to prohibiting young children from eating certain
birds lest they, too, only 'cooT and never develop hard lan-
guage, children must not sound like birds, even in play.
Thus, in order to insure that 'hard language' develops, the
mother prevents a dangerous association by terminating this
vocal activity. Furthermore, she makes it explicit to the chil-
dren and to the others around them, that children are to
speak 'good talk', not 'bird talk'. It is important to emphasize
that Mother does not want them to stop speaking, but to
speak properly.
Another form of verbal behavior that is not tolerated by
Kaluli mothers is the imitation and distortion of a younger
child's speech by an older child. It is important that older
children do not engage in language interactions with younger
children that are contradictory to the efforts made by adults
to ensure 'good talk' and 'hard talk'. Consider example (13).
(13) Abi (27i months) and his sister Yogodo (5£ years) are
alone in the house, as Mother has gone out to get wood.
Following Abi's utterances, Yogodo repeats what he says,
phonologically distorting his words to tease him. When
mother returns, Yogodo continues to repeat everything
Abi says to her, leaving him very confused and frus-
trated. After hearing eight turns of this, mother turns
to Yogodo and says:
speak words/language!
to sama
Mothers see this type of activity as not only mocking or teasing
the young child's not as yet well-formed language, but as con-
fusing the younger child about language, its correct form and
appropriate use. Thus, an undesirable language interaction is
terminated with the explicit directive to 'speak language' (to
sama). By focusing on the form of talk rather than its specific
content, the children are not discouraged from speaking to one
another but encouraged to do it properly, on the model of 'hard
words'.
By the time a child is about 3 | years old, and elema direc-
tives have stopped, that child's language is considered suf-
ficiently hard so that the playing of word and sound games
with peers is acceptable. While closely timed, repetitive,
formulaic utterances involving teasing and challenging are
appropriate for older children, mothers do not want these chil-
dren negatively influencing younger ones whose speech is not
yet well developed.
(14) A mother, her son (28 months), and three siblings
(ages 5-8), are sitting around a fire cooking bits of
food. The three siblings are playing a teasing game
 Hard Words: A Functional Basis for Kaluli Discourse / 365

about who will and will not eat, which involves speaking
rapidly and distorting words. After watching this for
16 turns, the little boy attempts to join the interaction
by interjecting nonsense syllables. The mother turns to
the older children saying:
speak hard!!
halaido sama!
to which one of the older children responds (teasing):
huh?, followed by the mother's repetition with empha-
sis:
speak hard!!
halaido same!!
'Speak hard' implies that until this point, speech has been 'un-
hard'. Such a reference is always to speech in an ongoing
context. In this situation, as in many others like it, mothers
are careful that their young children do not sound less mature
than they actually are in their speaking. This is consistent
with the goals of language socialization: to enable children to
be independent and assertive by the time that they are 3-3£
years old. Independence and assertion in speech and action
are functionally valued in this egalitarian society; ability to
speak out is one important way to get what one needs.
Next, we examine situations with negative imperatives, where
selesabo (sama) and tolesabo (tolema) are used. The use of
selesabo 'don't say (that/it)' (parole) implies that one knows or
suspects what is about to be said, and is telling another not
to say that thing. It is also used with reference to a specific
body of knowledge or secrets. One may say 'don't say that'
or 'don't tell them' with reference to specific information. Note
example (15).
(15) A number of people are socializing and eating in the
longhouse. A guest enters, having walked through the
muddy jungle paths; leeches have attached themselves
to his ankles. A child runs up to alert the guest to
this fact, and an adult intervenes, saying: selesabo!
'don't say it!', thus directing the child not to say the
speech specific word 'leech' while others are enjoying
their meal. Kaluli etiquette strongly prohibits the say-
ing of this word while people are eating.
The use of selesabo contrasts with the use of tolesabo.
Tolesabo means 'don't talk' in the sense, 'be quiet', 'shut up',
or 'don't engage in language' (langue). The meaning is 'stop
talking' or 'do something else besides engage in language'.
366 / Steven Feld and Bambi B. Schieffelin

(16) Isa <age 8) is teasing her brother Wanu (32 months)

about who will be his wife. Father tells him to counter
her teasing with:
1. Father -*-Wanu—»-^Isa:
no! glema
2. no! /
3. that's mother! elema
4. that's mother! /
5. One doesn't speak/say like
that! dema
eledo mosalano!
6. One doesn't
speak/say like
that! /
7. Father-»-Isa: girl, Isa, you . . .
that's being bad.
Shut up! Shut up!
tolesabowo!
In this sequence, an adult uses zl&na to instruct a young
child in how to provide an appropriate response to his sister's
teasing. In addition, in line 5 the child is directed to say 'one
doesn't say that', calling attention to the inappropriateness of
what is being said. This response is yet another way to
counter teasing. In such interactions the conventions of lan-
guage use are made explicit to younger members who may not
as yet know them or may need to be reminded of them. This
sequence ends when the father, being angry at his daughter,
tells her to 'stop talking 1 . This instructs the children as to
what is and is not out of bounds and further draws attention
to the social need to control the flow of talk by forcefully end-
ing undesirable speech.
A final example completes the point that in some interactions
the issue is not to say what you want to say better, but to
stop talking completely.
(17) A group of children are loudly talking and playing,
and mother turns to them:
Sosas, shut up!
tolesabo!
Sosas is the name of a very noisy bird, one whose sounds are
considered unpleasant. By comparing the children to sosas
birds, the mother emphasizes the irritating nature of the group
noise, further marking the general injunction to stop the annoy-
ing verbal activity and do something else. Tolesabo is used
here quite in contrast to selesabo; the children are being told
 Hard Words: A Functional Basis for Kaluli Discourse / 367

to stop the activity of talking, not to stop saying specific

things.
In these examples of learning and speaking 'hard words',
children are provided both with an explicit cultural model of
the importance of verbal activity, and with the importance of
saying or not saying the right thing. Functionally, such a
model promotes social integration into a coherent world con-
structed upon the importance of direct, controlled, forceful
face-to-face communication. Kaluli children learn to focus
upon what they want and need, even when this requires
challenge or confrontation. They learn that discourse is a
means to social ends, and they openly utilize sequential talk
following that model. Imperatives are often heard in the lan-
guage of adults to children and adults to each other, and the
ability to utilize language in interaction requires an under-
standing of when to demand specific speech and when to de-
mand verbal closure.
When something has been said or done, or might be said or
done, the ability to refer appropriately, report, or challenge
is one consequence of the way Kaluli learn 'hard words'.
Such situations continually reflect the choice of formulations
about what has been said in order to focus the specifics of
the situation. If one reports benignly to another that 'some-
one said something to me . . . ' , and the listener immediately
wants to challenge the substance of the remarks, a common
interruption at this point would be ba madali siyo 'it was said
for no reason'. Remarks on the truth or intentions of what
was said are very commonly the subject of initial interruptions
in conversation, immediately letting the speaker know the
listener's point of view on the reported speech. Remarks about
the circumstances of what has been said must be formulated
with siyo 'said', or ele siyo, 'said like that'; these refer to a
specific instance of speech or the 'said' of a report in a cer-
tain context. *Tolo can never appear in these situations be-
cause one cannot have the capacity or system of language in
the past; in fact, the construction is inappropriate in any
utterance about the language of deceased persons.
More pointed rhetorical strategies for dealing with the re-
ports or references of speakers are formulated with two common
phrases: ge siyowo dadaye?! 'Did you hear what I said?!' and
ge oba siyoWo? 'What did you say?' While these can be re-
quests for information, confirmation, or acknowledgment, they
are often found breaking into or responding to the stream of
discourse in order to focus reaction and challenge what is be-
ing said. Neither construction can be formulated with to and
tolzma, as both exemplify the necessity of controlling a spe-
cific instance of speaking.
Rhetorical challenge can be pushed a degree farther; esca-
lation to threat is an important way not just to register re-
sponse but to prohibit or shame someone who is doing some-
thing that is inappropriate or not approved of. In such cases
368 / Steven Feld and Bambi B. Schieffelin

the threat is registered simply with: sameib! 'someone will say

(something)!' The implicit threat is that someone will say 'who
are you?!', 'is it yours?!', or other pointed rhetorical questions
that shame the addressee. Use of sameib! to control inter-
actions that may get out of hand, rather than use of physical
control, emphasizes the concern Kaluli exhibit about speaking
as an instrument of social action and accomplishment. Such a
threat cannot be formulated with Holomeib! because it is the
implied 'something' that will be said that is so important to
shaming as a regulatory action.
In these examples of learning, speaking, and controlling
'hard words', it is clear that Kaluli must understand when it
is appropriate to talk about language, and when it is appropri-
ate to talk about speaking. Kaluli discourse then is taught
and utilized as an integration of linguistic and metalinguistic
practice which is shaped and scaffolded by having a place in
a culturally coherent world of beliefs about 'hardness', con-
trol, direct action, and assertion. Kaluli discourse must be
analyzed in relation to the belief system that constructs its
organization and goals, as well as the social ends which it
accomplishes for participants. Cultural analysis then is an ex-
plicit manner of connecting form and function. We have found
that constructing an analysis from the bottom up satisfies both
the demands of ethno graphically situated explications and the
demands of explaining the ordinary and routine ways that
Kaluli interactions actualize cultural expectations about language
use and meaningful social behavior.

3. Closure. To close a story, a speech (or, in a recent

adaptation among the few literate Kaluli, a letter), Kaluli utilize
the phrase ni towo kom 'my talk/words/language are finished'.
It is fitting that we close this paper by explicating why this
phrase is appropriate and why the contrasting *ni siyowo kom
'what I have said is finished' is inappropriate and not utter able.
For Kaluli, verbal closure implies directly that there is nothing
left to talk about, at least for the moment. What is finished is
the action of language, the invocation of words, the activity of
talk. No such boundary is appropriately imposed upon the
'said' of speaking in a specific setting, which is always open-
ended and ongoing. Verbal activities are closed by a boundary
on talk, not a boundary on what has been said. The function
of reaching closure, again, underscores the direct manner in
which Kaluli control situations and behaviors by viewing talk
as a socially organized and goal-directed actualization of the
capacity for language, 'hard words'. Ni towo kom.
NOTES
Fieldwork in Bosavi during 1975-1977 was supported by the
National Science Foundation, the Wenner-Gren Foundation for
Anthropological Research, the Archives of Traditional Music,
 Hard Words: A Functional Basis for Kaluli Discourse / 369

and the Institute of Papua New Guinea Studies. We gratefully

acknowledge their assistance. Detailed reports of our sepa-
rate work are Feld (1982) and B. B. Schieffelin (1979). The
order of author's names was determined by geomancy.
1. Kalu specifically means 'man' (opposing kesale 'women')
but can generally refer to 'person' or 'people'. Kaluli see the
ideal form of 'hardness' modelled on maleness; women, however,
are clearly supposed to be competent language users. Sex
role socialization is clear in the speech of mothers to children;
little boys are encouraged to use language to be demanding,
while little girls are encouraged to use language to be more
complacent. These issues are addressed in detail in B. B.
Schieffelin (1979: Chapter 2).
2. It is worth noting that, in contrast to some aspects of
metalinguistics, Kaluli do not directly verbalize about the im-
portance of a distinction between to and sama. The clear
langue/parole distinction is consistent, however, in all of our
elicited or tape-recorded naturally occurring data. Further
discussion of how this distinction affects Kaluli poetic concepts
can be found in Feld (1982:Chapter 4).
3. There is one additional context where the term to halaido
or halaido to is found. This is in the talk of debate, heated
discourse, anger, dispute, or confrontation (as, for example,
in a bridewealth negotiation). This sense of to halaido is far
less prominent than the broader usage. The morphological
marking -ait is used only to indicate anger; it is not prominent
in our sample of recorded speech (83 hours of family inter-
actions, 50 hours of song, myth, texted weeping, and more
formal modes).
4. We are speaking here about interactions in an assertive
frame. These characterizations do not apply equally to frames
of appeal. On Kaluli assertion and appeal, see E. L. Schieffelin
(1976:117-134) and B. B. Schieffelin (1979: Chapters 3 and 4).
5. In casual adult interactions, elema may be used to direct
a response to a speaker who is slow to respond to teasing or
joking. A more marked and deliberate adult usage occurs in
funerary weeping, where women improvise sung-wept texts to
a deceased person lying before them. Often these texts con-
tain lines like, 'Look up to the treetops, elema . . . ' , indicating
that the weeper is telling the deceased to say these words back
to her. The grammaticality and pragmatics here rest on the
notion that while the deceased is next to the woman in body, he
or she is going elsewhere in spirit, in the form of a bird. The
commanded words marked with elema must therefore be in the
form of an appropriate utterance to a living person from one who
is now a bird. 'Look up to the treetops 1 is such a line because
it indicates that from then on the weeper will only see the de-
ceased as a bird in the treetops. Feld (1982:Chapter 3) con-
tains an analysis of elema in sung-texted-weeping.
6. Transcription conventions are described in B. B. Schief-
felin (1979). Child speech is on the right and the speech of
370 / Steven Feld and Bambi B. Schieffelin

others plus contextual notes are on the left. Single arrow

indicates speaker to addressee; double arrow indicates speaker
to addressee who is to address a third party. Kaluli glosses
are provided only where to, tolema, sterna, and sama, or other
forms of these verbs, are used. Full transcripts of all exam-
ples with morpheme by morpheme glosses can be obtained by
writing to the authors.
7. The use of concatenated forms also appears with sama,
particularly in interactions with dona, where the child is too
far from the intended addressee and is told to 'go in order to
speak', semcni hamana.
REFERENCES
Feld, Steven. 1982. Sound and sentiment: Birds, weeping,
poetics and song in Kaluli expression. Philadelphia: Uni-
versity of Pennsylvania Press.
Feld, Steven, and Bambi B. Schieffelin. 1980. Sociolinguistic
dimensions of Kaluli relationship terms. Paper presented at
the Annual Meeting, American Anthropological Association.
Schieffelin, Bambi B. 1979. How Kaluli children learn what
to say, what to do, and how to feel: An ethnographic study
of the development of communicative competence. Unpub-
lished Ph.D. dissertation. Department of Anthropology,
Columbia University. [To appear: Cambridge University
Press.]
Schieffelin, Bambi B . , and Steven Feld. 1979. Modes across
codes and codes within modes: A sociolinguistic analysis of
conversation, sung-texted-weeping, and stories in Bosavi,
Papua New Guinea. Paper presented at the Annual Meeting,
American Anthropological Association.
Schieffelin, Edward L. 1976. The sorrow of the lonely and
the burning of the dancers. New York: St. Martins Press.
 THE MEDICINE AND SIDESHOW PITCHES

Fred 'Doc' Bloodgood

Editor's Introduction. One of the highlights of the 1981

Georgetown University Round Table on Languages and Linguis-
tics preconference sessions was a presentation by Fred 'Doc'
Bloodgood, the last known living medicine show pitchman. A
transcript of Mr. Bloodgood's presentation is included here in
order to preserve an example of a once flourishing, now ex-
tinct American folk discourse genre which is not otherwise
available. No medicine show or sideshow pitch of the twenties
or thirties was ever tape recorded or written down. As Doc
Bloodgood put it in a letter to me, until now the pitch
was not printed, stamped, stained, marked or engraved ...
on anything movable or immovable, capable of receiving the
least impression of a word, letter, syllable, or character,
which might have become legible or intelligible, to any
person or persons under the blue canopy of heaven.
A tape recording of Mr. Bloodgood's introductory remarks and
demonstration pitches is also available from Georgetown Uni-
versity Press.
Although the transcript fails to give an adequate sense of the
oral presentation, it shows dramatically that the medicine show
and sideshow pitches were constructed in ways similar to those
identified by Lord (1960) for oral epics: formulaic phrases
woven together in a flexible but structured sequence to yield a
text that sounds memorized because it is astoundingly fluent.
Mr. Bloodgood produced only a few false starts and only three
instances of fillers ('uh') in more than a half-hour of talk. The
pitches make use of repeated rhythmic patterns, sound play,
and specific details to create immediacy and vivid imagery—
features found in poetry, both oral and written.
What follows is a verbatim transcript of Doc Bloodgood's in-
troductory remarks and sample pitches. Mr. Bloodgood did have

371
372 / Fred 'Doc1 Bloodgood

a chance to go over the transcript and make minor corrections.

He inserted commas and occasional underscores, corrected the
spelling of names, corrected the transcription in a few places,
and added two sentences to the pitches. No other changes were
made.
On the matter of the names of the tonic ingredients, Mr.
Bloodgood wrote the following disclaimer in a letter to me, which
I pass on:
I will not vouch for the accuracy of the names of the roots,
herbs, leaves, gums, bark, berries, and blossoms. But
we have the distinct advantage that no one known to the
medical world can dispute their veracity.
In case of a challenge on this, just remember that some
of these are probably now extinct ... like the medicine show
people themselves . . . of yesteryear. (Down the valley of a
thousand yesterdays).
... the corrections I have listed are the nearest I could
come. Anyway, as my dad always told me, 'Never let a
grain of truth interfere with the story.'
I am very grateful to Mr. Bloodgood for his participation and
to Samantha Hawkins of Georgetown University and Steve Zeitlin
of the Smithsonian Institution for helping to arrange for his
appearance. Marta Dmytrenko generously volunteered to tran-
scribe the presentation.

Doc Bloodgood1 s oral introduction. Fifty years ago the medi-

cine show was the only entertainment that some people ever saw.
It was their one connecting link with the outside world, and
they looked forward to its arrival with a very great deal of
anticipation. I suppose one could say it was the forerunner of
the commercial on radio or television as we see it today. The
show was usually comprised of about seven to as many as four-
teen performers: singers, dancers, comedians, a piano player,
fiddler, and sd forth. And after about thirty minutes of enter-
tainment, then/frame the pitchman or the lecturer, followed
hopefully by ttie sale.
It was imperative that the moment the sale started, the music
also started Joud and lively. And then to add further interest
to that magical moment, all of the performers acted as agents
for the product. In other words, they would stand poised with
a bottle of tonic in hand, waiting for a hand. The moment a
hand shot up, that particular agent would loudly shout, 'Sold
out, doctor!1 And then all the other six or seven agents would
immediately echo that cry, 'Sold out, doctor!' Now on the way
back to the stage to replenish his stock, that is to get a new
bottle of medicine, that man would again scream, 'Sold out,
doctor,' and with all this repetition, you can readily understand
 The Medicine and Sideshow Pitches / 373

that the impression was created that a tremendous amount of

tonic was being sold, although unfortunately such was not al-
ways the case.
At some point in the program came the inevitable candy sale.
This was a box of prize candy usually put out by the Gordon
Howard Candy Company of Kansas City, Missouri, which sold
for as little as a dime or as much as a quarter. Now each
package contained a prize of some value. However, some of
the prizes were much much too large to be contained within the
package, so they had coupons calling for them up on the stage.
And these called for various articles, such as: safety razors,
silk hose, silk lingerie, opera glasses, field glasses, pen and
pencil sets, garter and hose sets, boxes of stationery, beacon
blankets, tilt-top tables, many other valuable and useful articles.
Now incidentally, it was the sale of this candy that was often
instrumental in allowing the show to move from one town to an-
other. Remember, this was in the very depths of the Great De-
pression. And that, combined with the fact that down in
Georgia they had not yet quite recovered from Sherman's march
to the sea, it made selling a real challenge. Now most of us
owned cars. But the show itself usually moved on a very large
truck, the sides of which would let down to form a stage. One
particular one that I recall was on a show I was with in Texas,
that was a Packard truck--a 1912. It had solid rubber tires.
What a collector's item that would be today!
Medicine shows were once as plentiful as the buffalo on the
western plains, and people thought this was something that
would be with us forever. And yet, a year ago this last fall,
when the Smithsonian Institution decided that this was part of
our American heritage which should be preserved, they had
great difficulty in locating anyone who was ever even a part of
it. So they placed an ad in the Billboard magazine. (That's
the showman's bible.) And even after combing the entire
country, they were able to locate just fourteen performers and
two pitchmen. Actually they located one pitchman; the other
one passed away last summer.
One of the difficulties, of course, was the fact that not one
word of the material was ever written down. Any of the bits,
the skits, the after pieces or the lectures. Not a bit of it.
Everything was passed from word of mouth, one to the other.
And what with old age taking its toll, and ancient memories
growing dim, it is truly only a question of time until the medi-
cine show of yesteryear will have vanished down the valley of
a thousand yesterdays. I suppose it would be safe to assume
that we are an endangered species.
Now by a stroke of very good luck, my son in California
happened to see this ad in the Billboard, and he telephoned to
me and he said, 'You surely should answer t h i s ' . And under
extreme duress I finally did. The duress came because I felt
that this would never come back. I hadn't done any of these
in forty years, these pitches. Or anything like that. But I
374 / Fred 'Doc1 Bloodgood

did call the Smithsonian people, and they suggested that I sub-
mit a tape of a sample of an oldtime medicine show pitch, which
I did, and immediately I received a call from them inviting me
to be a participant in the folk life festival in October.
And my arrival on that lot was a very emotional thing, I
assure you. They had reconstructed an old-time medicine show
stage just exactly as I remember it, complete in every detail
even to the two model T trucks, one on either side of the stage,
beautifully restored. When I stood there it was as though time
had turned backward in its flight a half a century. And a
great lump came to my throat, and a thrill ran up and down my
spine, and--remember, all this, in the actual shadow of the
Washington Monument. To me it was the absolute apogee of
spine-tingling enchantment.
And when those banjos started to play and those fiddlers
started, it all came back to me just exactly as I hope it will
today. We gave a full and complete performance each day plus
many interviews and reminiscences, all of which was recorded
for the Archives of the Smithsonian Institution. They told me
that even two hundred years from now (Mr. Zeitlin told me that)
that two hundred years from now, if anyone were doing research
on medicine shows, that would be the voice they would hear.
It was very awe-inspiring to me to think that I might be able
to contribute something to our American heritage. Actually, I
suppose the only real difference between David Farragut and
Doc Bloodgood is that instead of the stirring battle of cry of
'Don't give up the ship!' etched indelibly upon the yellowing
pages of history, my contribution may be, 'Sold out, doctor,
four dollars change'.
Now just a very few words as to how I got started in this
business. As a youth I can't ever remember ever wanting to
do anything else. When I was seven my Dad took me to the
circus, and the circus was just great, but the sideshow! The
sideshow was incredible. And I thought to be able to stand on
that platform, in front of that long line of pictorial paintings,
and actually convince anybody of the benefits to be derived by
actually visiting that congress of freaks, curiosities, and mon-
strosities, that would be the greatest occupation a man could
ever have. So, the day I finished high school, I hitch-hiked
to the nearest circus, and there followed eleven of the most ex-
citing, incredible years that a youth could ever have. With
circus and sideshows in the summer in the north, and then to
the medicine show circuit consisting of Georgia, Mississippi,
Alabama, and Texas, in the winter, and all this, I remembered
that, I think one of the things that convinced me, that to t>e
able to sell your goods sight unseen, and collect the money in
advance--that would be the ultimate in selling.
That wonderful week in Washington that we had a year ago
last October was not without its sad moments, because as I was
standing on that platform on the last and final day, it suddenly
occurred to me that it was quite conceivably possible that I was
 The Medicine and Sideshow Pictures / 375

now giving the last and final medicine show lecture that the
world would ever know.
But how very wrong I was. Because last summer, they tele-
phoned me again and said that because of the success of the
Washington situation, that they were going to do a documentary
film of a medicine show in Bailey, North Carolina. So, again,
we flew to NC and met fourteen venerable old performers, most
of which we already knew. All had been with medicine shows
in the old days, one of which was Roy Acuff, and may I just
say that he probably is one of the very nicest people I've ever
had the privilege of meeting. So, again, we gave a performance
every day and the film company from the Smithsonian took over
seven miles of film which will be edited and then shown on edu-
cational TV next fall sometime.
Their attention to detail was absolutely phenomenal, even to
this time the Grand Free Street Parade. They had a galaxy of
model T Ford trucks with performers and musicians playing up
and down the main street of Bailey, NC, while the cameras
rolled. And it was my privilege to lead the parade perched on
the back of a rumble seat of a Model A Ford roadster, with a
bottle of tonic in one hand, and waving to an enthusiastic
audience with the other. Oh, and the last two nights I really
had an opportunity to sell medicine again. Now actually this
was just colored water and the people were apprised of that.
It was a souvenir bottle. But that which I'd waited for so long
happened. When I got ready and offered that medicine for sale,
a thousand people became a veritable sea of hands, each hand
holding a dollar bill in it. How often I had dreamed of just this
situation!
Those wonderful people of North Carolina gave us a banquet
or a barbecue almost every night, consisting of whole roast
pigs and other southern specialties. Then throughout the day,
after the show they would come back and ask for autographs
and the Girl Scouts would ply us with cookies and cold drinks.
It was a memorable week in our lives, but it was also a memora-
ble week in history. Because on Saturday, September 6, 1980,
the medicine show of yesterday joined the silent ranks of the
dinosaur and the dodo and the passenger pigeon. That was the
end forever. And as the guitars and fiddles faded into the
night air, the phenomenon of the oldtime medicine show also
faded, forever. And the evening breeze seemed to sigh with
sadness as it swirled the papers and the refuse of an emptying
lot, as the last clarion cries of 'Sold out, doctor' vanished for-
ever.
I've been asked to do an actual pitch one more time. It'll be
just as it was in 1928, with one exception. There won't be any
sale this time. I hope that you'll enjoy hearing it as much as
I'll enjoy doing it for you. And if at the conclusion of our per-
formance any time remains, I would be just more than happy to
answer any or all questions, on a no charge basis, without cost
or obligation.
376 / Fred 'Doc1 Bloodgood

Oh, just one more thing. It was a different era. Remember,

things that were dead serious in those days, hopefully today
they will prove to be entertaining or even amusing to you.
But it was an entirely different age.
You will notice one thing also, that whether it be a medicine
show pitch or a sideshow pitch, we used a great deal of
alliteration or euphonious phrases. May I give you just a short
example? For instance, in the process of describing alliteration:
Now one particular show I had was a deep sea diving show con-
sisting of a diver, a hard hat diver, and a giant deep sea devil
fish or octopus. And at one point in the program I would say,
And ladies and gentlemen, I'm going to send the young man
to the very bottom of that steel and glass tank for a hand
to hand encounter with that death dealing demon denizen of
the deep and, if fortunate, bring him to the surface for your
very close inspection.
Or an example of a euphonious phrase: I once worked on a
living skeleton with the Rubin and Cherry show; this was in
about 1929: Walter R. Cole. Now, Walter Cole was 5'11" tall
and weighed just 63 lbs. Now at a certain point in the pro-
gram, I would ask Walter to raise that right arm of his, and he
would put that tiny little arm over the top of the canvas, and
I would say,
Ladies and gentlemen, there he sits in there, slowly wasting
away, slowly becoming atrophied, slowly becoming ossified,
slowly becoming petrified, and slowly turning to stone. Un-
able to move, carried around from one place to another by
his nurse just like a mother would a babe, and yet--
apparently happy, for I have never heard him complain.
I want you to also if you will notice, that there's a great use
of comparisons in all of these pitches too. And it's not by acci-
dent that we used the long £ or the AH sound when we named
freaks. The reason for that was very obvious, really. We
didn't have all this sound equipment in those days and the E
or the AH sound had a much better carrying quality. That's
why all my 'Geek' shows or wild girl shows, were either Neva
or Neola. I'm thinking of one show that I worked on where
they incorporated both the E and the AH sound, was Leah-Lee--
half man, half woman, alive.
And when you get on the inside, ladies and gentlemen, I
want you to draw an imaginary line from the very top of its
head down to the very tip of its toes. And on one side
you'll find this strong arm, the muscular limb, the coarse
beard, the heavy features. And on the other side, the
 The Medicine and Sideshow Pitches / 377

beautiful features of the feminine sex. Father, mother,

brother and sister, in one body alive.
And now for the pitch of the medicine show.
THE MEDICINE SHOW PITCH
Good evening, ladies and gentlemen, and welcome to the Clif-
ton Comedy Company. We have come to your city to stay one
week, bringing you clean, moral, refined entertainment which
is absolutely free. We bring with us a company of fourteen
performers, each and every one an artist in his or her line.
But more than that, they are all ladies and gentlemen and can
conduct themselves as such. In other words, there will be
nothing seen, heard, said, or done, to mar the impunity or in-
jure the propriety, in any way, shape, form or manner of the
most fastidious little lady in the community. (If you'll pardon
me a second, then, you'll see how these are usually in multi-
ples of two, four, or six. It seems to work out that way; I
don't know why.)
We're sent here by the Finley Medicine Company, forty-one
hundred and fifty-one Olive Street for the express purpose of
introducing and advertising their product. Now we have just
two products, the Hospital Tonic, and the Instant Liniment.
The Hospital Tonic is a harmless preparation consisting of roots,
herbs, leaves, gums, barks, berries, and blossoms. Including
ginseng root, diana emma leaves, sinco, salfamettaberries, iron
phosphate, cassian, mandrake, Canadian snake root, bitter
apple, Chinese dragon flower, and gimico oil. And now you're
going to say, 'Well, will it cure everything?'
Ladies and gentlemen, we don't have a cure-all. If I did
have, it wouldn't be necessary to come out here with a show-
I would set up in some small town or city and in three days'
time I'd be having more business than I could handle. Friends,
we don't have a cure-all, and if I were to tell you we did have,
I would be lying to you. And I'm not going to lie to you.
Now throughout the week you're going to hear people calling
me 'doctor', and actually I'm really not a doctor. I did attend
Northwestern for two years. I'm not licensed, I'm not allowed
to make calls. Soon after that, I decided I didn't want to hang
out a small shingle in some little town, but I would prefer to go
down the highways and into the byways in an attempt to allay
the sickness and suffering that mankind is heir to. And if you
could look as I do upon that vast multitude of people, people
that I see going to and from me daily that I've taken off from
canes, off from crutches, out of the sick bed, ah, you might
say snatched off the operating table with the use of that tonic,
then you wouldn't blame me for preaching.
Now, really, that's all I know. Our product is good for three
things: the stomach, the liver, the kidneys. The three princi-
pal blood-making organs, or any disease arising therefrom, such
378 / Fred 'Doc' Bloodgood

as sour stomach, indigestion, constipation, female weaknesses,

rheumatism, catarrh. Any disease arising from a disorderly
stomach, impure liver, deranged kidneys, with the exception
of Bright's disease, and friends, let me say this. If your
kidney complaint has reached that stage, don't you buy a bottle.
It wouldn't do you any more good than that much rain water.
I would much rather you wouldn't have it.
And as I say, that's really all I know: just the stomach,
the liver, the kidneys. I have a car sitting out there, a Buick.
As long as it'll run, I can drive it. But if it stops, I don't get
out and try to fix it. I just hail the first passing motorist.
He may turn over a wire, a nut, or a bolt. I'll put my foot on
the starter and it'll go along all right. Now, if Mr. Buick had
made the car with stomach, liver and kidneys, then I could
have fixed it. He didn't, I don't know anything about it, I
don't want to know anything about it.
Now ladies and gentlemen, for me to earn my salary, it is
going to be necessary for me to say just a very few words
each night about the product. Tonight I would like to say a
few words about the subject of rheumatism: how, as a general
rule, it's treated, how, as a general rule, it's cured. On
either side of that lumbar spine there are two little organs.
When in a healthy condition, they weigh from two to four
ounces. These are known as the kidneys, or, in other words,
the sieve of the human body. And their duty in the body is
the same as a sewer system is in a large city. That is, to
cleanse and purify the blood.
Now let us suppose the kidneys are not in a good condition,
and they allow too much uric or lactic acid to go out into the
system. This in turn crystallizes, becomes like a lot of
powdered glass. In fact, friends, if I were to take that light
bulb, break it, pulverize it, powder it into the very finest of
glass, and then cut an incision in my wrist, fill it full of the
powdered glass, sew it up, it would heal up and, to all exter-
nal appearances, seem sound and well. And perhaps I wouldn't
even feel the pain--until I went to move that muscle. There
would be just one difference. I couldn't take medicine in my
system strong enough to dissolve the powdered glass. But if
the uric acid is properly treated, it can be driven back into
circulation.
And now you're going to say, 'how do we do it?' Well, I'm
not going to say that a good hot liniment properly applied will
not bring you temporary relief. I've had men come in my office,
with rheumatism so bad they couldn't raise their arm above their
shoulder, and in five minutes' time I'd have him putting his arm
high above his head, free from any pain and he'd be elated.
He'd say 'You've cured me! You've cured me, doc!' But I
really hadn't. I hadn't done anything of the kind. I had
driven that uric from that point to some other point.
Now, notice this, that usually it'll be settling in a point that's
been overstrained previously. For example, a man who's done a
 The Medicine and Sideshow Pictures / 379

lot of lifting, it'll settle in the small of the back, the lumbar
region of the spine. And they say the poor man has lumbago.
If it's someone like a postman that's done a lot of walking, it'll
settle in the sciatic nerve of the thigh, and they say he has
sciatica. In the face it's neuralgia. I don't care where it is,
it's all one and the same thing--too much uric or lactic acid.
Now, friends, I do have a preparation that will cure that
condition. Why? Because it goes right to work on those kid-
neys. Puts them in a strong, workable, healthy condition, so
they can perform the work which Nature's intended them for.
And it doesn't stop there. It goes to work on those other two
organs, the stomach and the liver, and let me say this, that if
it doesn't help you in three days, then it won't help you in
three years. You bring it back to me. I'll give you another
dollar for the bottle.
Now, I've had people come in my office also and say, 'I
haven't taken a dose of medicine in five years or ten years'.
And if you would stop and think, just think, a person wouldn't
make a remark like that. Let me paint you a word picture that
the smallest boy or girl in my audience can understand. Those
of you that keep house, have sitting at your back door what we
call a garbage can or a slop bucket. And when you get through
with your breakfast dishes, you scrape those dishes into that
bucket. You do the same thing with your lunch, same thing
with dinner. And when the bucket gets full, you take it out
and bury it or feed it to the pigs. I don't care what you do
with it, but just keep that bucket in that capacity for one
week's time, and then I want you to see the condition that it's
in. See the filth that adheres to the sides. Smell the stench
that comes from it, and stop and think, 'I've been putting that
same food into my stomach not for a day, a week, but for five
years--or ten years--and I have never cleaned it out!' (See,
I told you that these would be mildly amusing, but nobody ever
laughed in those days.) And, I will guarantee, ladies and
gentlemen, that the very first dose of the Hospital Tonic will
bring from your body, double handfuls of filth, slime, mucus,
corruption, fecal matter, maggots, and even worms.
And not very long ago, we asked the Finley Medicine Com-
pany to add one more ingredient in the product--something that
would pass a tapeworm—head and all. And I'm proud to say
that that condition now exists. In fact, I have some specimens
back there in my office. I have one in particular that I remem-
ber from a Mr. Adams, in Sanger, Texas--a brakeman on the
Baltimore and Ohio Railroad. He got a bottle of the tonic on
Monday night, and on Friday he came down with that in a tin
can. And I washed it and measured it. It's a tapeworm that
measures just over 16 ft. in length! And I have Mr. Adams'
sworn statement that he used no other medicine but the Hospital
Tonic in the passing of that worm!
Now there is just one more thing and then I'm all through.
One more thing that I think makes that product stand head and
380 / Fred 'Doc1 Bloodgood

shoulders above anything else ever offered on the market. But

there is one thing about it. I can't say much before a mixed
audience of ladies and gentlemen. But I will say this: If
there is a man within the hearing my voice that goes to his
home tonight, and he sees that poor wife, sister, mother,
sitting there with her head tied up, and you say, 'What's the
matter, Mary', and she says, 'Nothing. Nothing at all, John',
don't you believe there's nothing the matter. There is some-
thing--something she's not going to confide in you. She's not
going to tell you all her troubles, why--you know the disposi-
tion of a woman. The majority of them will drag themselves
around as long as they can keep going, and finally they break
down', and then you have an invalid to take care of the balance
of your days.
Friends, I talk to you like I would my own mother, my own
folks in my own home, and if I thought it'd do any good, I'd
get down on this platform, on my knees, I'd beg you to take
that woman home a bottle of that tonic. Oh, if you've got a
woman like that at home, you see she's on the toboggan, on the
downhill path. You want to bring the roses back to her cheeks,
make her step pick up, make her feel like she should again,
you'll take my advice, and take her home a bottle of that tonic.
The price? The price is so low you cannot afford to miss it.
It's a dollar a bottle, and with every bottle you will receive one
hundred votes for the most popular lady or baby in the Com-
munity. I'm only going to have our agents pass among you
just one time, and I'll ask you to raise your hand, turn on
your lights, I'll be glad to wait on you. Thank you. There
goes one right over in there. Thank you sir, very much,
thank you.
And that's the way a medicine show pitch sounded in 1928.
Just a couple of more things, and then I'm all through.
THE SIDESHOW PITCH
During those incredible years, I accumulated enough memories
to last two lifetimes. It was for instance my very great privi-
lege to tour the United States with one of the most amazing, one
of the most astounding, one of the most bewildering sights: a
human oddity, one of the strangest ever exhibited under the
blue canopy of heaven. That was Neola, that strangest of all
strange creatures. It was the very same girl that was brought
here during that great evolution trial that took place in Dayton,
Tennessee in 1925 between the late William Jennings Bryan and
Clarence Darrow, the great criminal lawyer. She was examined
at that time by some of our leading psychiatrists, Dr. Mullen,
head of the Department of Psychiatry at Columbia. Dr. Mullen
claimed that she had less intelligence than that of the chimpanzee
or the monkey family. She was found by that great antiquarian
explorer and trapper, Dr. Carter, who was exploring for animals
 The Medicine and Sideshow Pitches / 381

in the lowlands or swamp regions of Abyssinia. He heard this

strange tale of some fanatic natives, of a strange and curious
animal that lived in the center of a deep dark cave, and in-
vestigating their story, he found it to be fact. But he found
not an animal, but a human being, crouched upon a huge flat
rock exactly as you're going to see her in there tonight, com-
pletely surrounded by hissing, seething monsters, some of
them larger than a man's upper arm, some that could crush a
human being with one coil of those enormous bodies just as
you or I would crush a piece of food between our teeth.
Let me paint you a mental picture of her. She stands just
3 feet tall. Long, long arms that hang way down below the
knees. Eyes that pop out and glare just like two red-hot coals
of fire. But I think the most peculiar thing about her of all
(and I think you'll agree when you see her) is the shape of
the head. The head tapers at the top just like that of a coco-
nut. She doesn't speak any language. Doesn't know any
creed. Neither walks nor talks, just creeps and crawls and
spends her lifetime down in that steel-bound arena, down in
that steel cage, where you would not expect a dog to live for
an hour.
Now all afternoon they've been asking, 'When and what time
are you going to feed her?' And that moment has now arrived!
Once, on this very performance, I am going to feed her exactly
as you'd see her in her own native land, Abyssinia, in the
north of Africa. You'll see her leap clear across that steel-
bound cage, the eyes will pop out and glare like two red-hot
coals of fire. Now it's way beyond feeding time. I'm going to
feed her positively within the next three minutes. The price,
so low you cannot afford to miss it, usually 25 for the adults,
10 for little ones. I'm going to lay away the regular adult
tickets, turn back the pages of time, make children out of all
of you. Now I'll say this, for a period of three minutes and
three minutes only by the clock, if you can lay as much as the
price of a children's ticket, that's 100, it's 10<f to each, 10 to
all. You may get tickets here, and I'm going to feed her within
three minutes, whether one of you go, all of you go, or none
of you go. Hold it Doctor, don't feed her just yet!

'Doc' Bloodgood's concluding remarks, Y'know, during those

years, I suppose in retrospect it's natural to remember only the
nice things, the exciting things that happened. One has a
tendency to forget the endless rains, and the muddy lots, and
the stuck trucks, and the winds that often destroyed our tents.
Somewhere a half a century ago, I heard a little verse that
seems to tell it all. It says,
Looping around the mountain,
Dragging across the plain,
Gay in the flowered sunlight,
And drab in the droning rain,
382 / Fred 'Doc1 Bloodgood

Man from his gypsy childhood

Is a houseless vagabond
Who pitches his tent and passes
'Til he camps in the dim beyond.
And now, ladies and gentlemen, as we used to say on the
medicine show, I want to thank you one and all, for your very
kind, courteous and undivided attention, wishing you a safe
return to your respective homes and destinations, and a very
fond good night. I thank you.
Printed in the United States
1971
780878 40116'
Contents

Alton L. Becker. On Emerson on language. 1

Walter J. Ong, S.J. Oral remembering and narrative structures. 12
Robin Tolmach Lakoff. Persuasive discourse and ordinary conversation,
with examples from advertising. 25
Frederick Erickson. Money tree, lasagna bush, salt and pepper:
Social construction of topical cohesion in a conversation among
Italian-Americans. 43
Emanuel A. Schegloff. Discourse as an interactional achievement:
Some uses of 'uh huh* and other things that come between sentences. 71
Malcolm Coulthard and David Brazil. The place of intonation in the
description of interaction. 94
Roger W. Shuy. Topic as the unit of analysis in a criminal law case. 113
Catherine E. Snow and Beverly A. Goldfield. Building stories:
The emergence of information structures from conversation. 127
Georgia M. Green. Competence for implicit text analysis: Literary style
discrimination in five-year-olds. 142
Joseph E. Grimes. Topics within topics. 164
Teun A. van Dijk. Episodes as units of discourse analysis. 177
J.L. Morgan. Discourse theory and the independence of sentence
grammar. 196
V. Melissa Holland and Janice C. Redish. Strategies for understanding
forms and other public documents. 205
William Labov. Speech actions and reactions in personal narrative. 219
Charles J. Fillmore. Ideal readers and real readers. 248
William Bright. Literature: Written and oral. 271
Sally McLendon. Meaning, rhetorical structure, and discourse
organization in myth. 284
Joel Sherzer. The interplay of structure and function in Kuna narrative,
or: How to grab a snake in the Darien. 306
John J. Gumperz. The linguistic bases of communicative competence. 323
Ron Scollon. The rhythmic integration of ordinary talk. 335
Steven Feld and Bambi B. Schieffelin. Hard talk: A functional basis
for Kaluli discourse. 350
Fred 'Doc' Bloodgood. The medicine and sideshow pitches.
ISBN 0-87840-116-4

780878 401161
Georgetown University Press
GEORGETOWN UNIVERSITY PRESS
www.georgetown.edu/publications/gup/