Contrasting Languages

Trends in Linguistics
Studies and Monographs 51

Editor
Werner Winter

Mouton de Gruyter
Berlin · New York
Contrasting Languages
The Scope of Contrastive Linguistics

by
Tomasz P. Krzeszowski

Mouton de Gruyter
Berlin · New York 1990
Mouton de Gruyter (formerly Mouton, The Hague)
is a Division of Walter de Gruyter & Co., Berlin.

© Printed on acid-free paper which falls within the guidelines of the ANSI to ensure
permanence and durability.

Library of Congress Cataloging in Publication Data

Contrasting languages the scope of contrastive linguistics /

edited by Tomasz P. Krzeszowski.
p. cm. (Trends in linguistics. Studies and mono-
graphs 51)
Includes bibliographical references and index.
ISBN 0-89925-590-6 (cloth acid-free paper)
1. Contrastive linguistics. I. Krzeszowski, Tomasz P.
II. Series.
P134.C58 1990
410 —dc20 90-45502
CIP

Deutsche Bibliothek Cataloging in Publication Data

Krzeszowski, Tomasz P.:

Contrasting languages the scope of contrastive linguistics /
by Tomasz Krzeszowski. Berlin New York Mouton de
Gruyter, 1991
(Trends in linguistics Studies and monographs ; 51)
ISBN 3-11-012133-6
NE: Trends in linguistics / Studies and monographs

© Copyright 1990 by Walter de Gruyter & Co. D-1000 Berlin 30

All rights reserved, including those of translation into foreign languages. No part of this
book may be reproduced or transmitted in any form or by any means, electronic or
mechanical, including photocopy, recording, or any information storage and retrieval
system, without permission in writing from the publisher.
Typesetting: Arthur Collignon GmbH, Berlin — Printing: Gerike GmbH, Berlin — Binding:
Lüderitz & Bauer, Berlin — Printed in Germany
Acknowledgements

I am indebted to all my friends and colleagues who took pains to read

the draft of this book in its entirety or in parts and suggested improve-
ments. The complete list would have to include many students of mine,
the first patient experiencers of the ideas contained in this volume. Facing
the impossible task of mentioning all the numerous persons whose com-
ments influenced my thinking, let me single out the following scholars:
Prof. Dr. Edmund Gussmann, Prof. Dr. Ruta Nagucka, Doc. Dr. Elzbieta
Muskat-Tabakowska, Dr. Tadeusz Danilewicz, Doc. Dr. Barbara Kryk,
Dr. Wojciech Kubinski, Dr. Krzysztof Kwasniewicz, Ms. Janina Ozga,
and Dr. Ewa Willim.
The book owes its existence to the tradition of contrastive studies in
Poland, inspired, promoted, and supported by their spiritus movens Prof.
Dr. Jacek Fisiak, to whom I hereby express my special thanks.
I am also very grateful to Marc Weinstein, who carefully read my
manuscript and suggested numerous stylistic improvements.
I also owe a debt of gratitude to Ms. Olga Sokolowska for her
assistance in compiling the indexes.
If the book suffers from inaccuracies, misinterpretations, and other
flaws, it is partly because its author did not pay sufficient heed to the
illuminating comments of its critics. Assuming total responsibility, the
author apologizes to both: the critics, for not listening carefully enough,
and to the readers, for obliging them to endure the remaining inadequa-
cies.
Gdahsk, March 1988 Tomasz P. Krzeszowski
Contents

Introduction 1

Chapter I

What is contrastive linguistics? 9

Chapter II

Tertium comparationis 15

Chapter III

Towards a classification of contrastive studies 23

Chapter IV

Three steps in "classical" contrastive studies 35

Chapter V

Contrastive studies at various levels of linguistic analysis 47

Chapter VI

Linguistic models and contrastive studies 107

Chapter VII

Towards a theory of semanto-syntactic equivalence 147

Chapter VIII

Contrastive Generative Grammar 169

Quantitative
Error
Chapteranalysis,
XIX contrastive
interlanguage,
studies
and Contrastive Generative Grammar 203
189
VIII Contents

Chapter XI

Cognitive linguistics and contrastive studies 213

Chapter XII
"Theoretical" and "applied" contrastive studies 233

Notes 245
References 251
Index of subjects 267
Index of authors 283
Introduction

Contrastive studies do not enjoy much respect among linguists. Although

many scholars of repute occasionally or systematically do practise what,
loosely speaking, falls within the domain of contrastive studies by evoking
cross-language evidence to support some theoretical claims (for example,
R. Lakoff 1972, Comrie 1976, or Siewierska 1984), they only do so
compelled by general linguistic motivation rather than by the require-
ments imposed by the rules of an altogether different game called con-
trastive linguistics.
Those who, by their own admission, undertake contrastive studies only
involve themselves in that part which we shall presently refer to as
contrastive analysis proper, paying insufficient heed to matters of prin-
ciple, which motivate the analyses and provide them with methodological
tools. In any case, the emphasis falls on actual practice and applications
to the detriment of the theory and methodology of contrastive studies.
Whatever issues arise in connection with these latter two aspects of
contrastive studies, they are treated only marginally, as it were, in passing,
and without sufficient attention paid to matters of finer detail. Conse-
quently, the number of works explicitly and exclusively devoted to the
theory and methodology of contrastive studies is negligible (but see Di
Pietro 1971 and James 1980).
For some years, I have been engaged in providing contrastive studies
with a more rigorous format. Several other linguists have advanced
illuminating comments and criticisms, notably Bouton (1976), Chester-
man (1980), James (1980), and, most extensively, Van Buren (1976). Thus,
although only embryonically, something that might be called a method-
ology of contrastive studies has begun to emerge.
The present book attempts to bring under one cover some discussions
and controversies connected with contrastive studies, to suggest possible
answers to some critical comments, and to provide a synthetic outlook
on the state of the art of contrastive studies with some modest suggestions
of improvement.
Contrastive studies have a very long history. As early as ca. 1000 A. D.
Aelfric wrote his Grammatica, a grammar of Latin and English, based
2 Introduction

on the implicit assumption that the knowledge of grammar of one lan-

guage may facilitate the learning of another language.
Among later grammarians, John Hewes, in the 17th century, was the
first to explicitly express the view that the knowledge of the native
grammar can not only facilitate learning a foreign language but also
interfere with it. In his A perfect survey of the English tongue taken
according to the use and analogie of the Latine, published in 1624, he
devoted a long introductory section to presenting fundamentals of English
in order to provide the learner with a "right knowledge or censure of
their owne Mother tongue, in regard it holdeth a great difference in it
selfe from the dialect of the Latine" Having provided some intricate
contrastive analyses of Latin and English, Hewes provides numerous
translational exercises to counteract what appear to be the effects of
negative transfer, focusing his attention on those phenomena which are
different in Latin and English.
Many other grammarians, like Howel(l) (1662), Coles (1675), and
Lewis (1670?) applied the idea of facilitation (positive transfer in modern
terms) by adjusting their grammars of English or of Latin to the needs
of speakers of various native languages. It is very interesting to note that
those early contrastive studies were motivated in almost the same way as
modern contrastive studies in the United States were motivated. As early
as 1670, Mark Lewis wrote the following words:
The most facil (sic!) way of introducing any in a Tongue unknown is
to show what Grammar it hath beyond, or short of his Mother tongue;
following that Maxime, to proceed a noto ad ignotum, making what
we know, a step to what we are to lean (sic!).
One wonders whether nearly three centuries later Charles C. Fries was
aware of these words, when he wrote the following:
The most efficient materials are those that are based upon a scientific
description of the language to be learned, carefully compared with a
parallel description of the native language of the learner (Fries 1945:
9)·
So, although the word "contrast" with reference to different phe-
nomena across languages had not appeared until the end of the 18th
century, when James Pickbourne (1789: 18) first used it ("I thought it
would be useful to contrast [italics supplied] the English verb with the
verb in other languages"), comparisons of languages for pedagogical
purposes probably go to the very beginnings of foreign language teach-
 Introduction 3

ing, while systematic written records of such procedures go back to at

least the 15th century (cf. Krzeszowski 1985, 1986 a, and in prepara-
tion; Meech 1935).
The early contrastive analysts did not concern themselves with meth-
odological problems, although they did work out a method of comparison
known as the "sign theory", the first method in contrastive studies
(Krzeszowski 1985). For many years contrastive studies were practised
and applied in the classroom in a more or less intuitive way, toutes
proportions gardees, like folk medicine, without much theory and without
much explanation.
But modern linguistic theories that began to flourish in the 20th century
could not fail to affect the state of affairs in contrastive studies. Interest
in methodology and theory of contrastive studies began to grow. Lin-
guistic explorations into the nature of language, its complex, multilayer,
and hierarchical structure, its systematic but changing nature, its function
in communication, and its relation to the human mind became a subject
of very close scrutiny of modern linguistics. This is not to say that many
of these problems never arose before. For centuries, people wrote gram-
mars (including contrastive grammars) and for centuries they were inter-
ested in how languages reflect human thoughts. But modern linguistic
theories have given new dimensions to old problems and have created
new problems (cf. Fisiak 1975 b; Fisiak — Lipinska-Grzegorek — Zabrocki
1978). Obvious things ceased to be obvious, and completely new ap-
proaches to language were proposed (e. g., generative grammar). Modern
linguistic theories have made contrastive analysts sensitive to methodo-
logical and theoretical problems in their own field. Contrastive studies
began to aspire to the status of a rigorous scientific discipline. What was
once a relatively simple, intuition-based procedure, began to assume the
format of an algorithm, which culminated in the attempt to construct
Contrastive Generative Grammar (see Chapter VIII). The rigorous, math-
ematical approach to contrastive studies has revealed a number of thorny
problems which at best make such efforts extremely difficult and at worst
make them hopeless. Ironically, the attempt to give contrastive studies a
rigorous, mathematical format could turn out to be self-defeating: it
could reveal the impossibility of conducting formalized contrastive studies
(cf. Van Buren 1976: 315).
The theoretical problems which bedevil contrastive studies can be
summarized as three paradoxes: the grammatical paradox, the semantic
paradox, and the pedagogical paradox.
4 Introduction

The grammatical paradox

Any grammatical contrastive studies performed without reference (at

least implicit) to meaning are doomed to failure. Yet, witness another
paradox, the revival of contrastive studies in the 20th century was pro-
moted by rather extreme versions of American structuralism, in which
the semantic aspect of language was removed from the mainstream of
linguistic analysis.
What makes a purely structural approach to contrastive studies the-
oretically impossible is the fact that any comparison presupposes simi-
larity as tertium comparationis (TC) against which differences can be
stated. Therefore, languages are structurally comparable in the extent to
which they are structurally similar. It must be remembered that early
structuralists were not yet familiar with the notion of underlying structure
nor with the notion of semantic representation from which, according to
later theories, surface structures in various languages are derived. These
concepts were acknowledged and appreciated in contrastive studies as
soon as they had made their way in the linguistic world. Originally,
however, contrastive studies concerned only those structures which later
became known as surface structures.
If one considers a taxonomy of surface structures in various languages,
it is possible to envisage a sort of a cline of similarities for every pair of
languages within the taxonomy. Theoretically, at one extreme, there will
be a pair of languages so radically different that no common set of
grammatical categories can be established for them, and at the other
extreme, there will be two languages identical in all respects. In-between,
there will be a spectrum of pairs of languages with varying degrees of
similarities. In such a hypothetical cline, the languages situated at the
two extremes would be completely incomparable in purely structural
terms. The two languages that would have nothing in common could not
be compared since there would be nothing that could serve as tertium
comparationis. The two structurally identical languages at the other ex-
treme, being identical, could not be compared either since there would
be nothing different to consider.
Graphically the cline can be represented as in Fig. 1, where L t and L 2
have no features in common while L n and L n + , are identical in all respects.
In fact, Fig. 1 represents a fragment of a more complex situation since,
in principle, the number of totally different languages and totally identical
languages may be more than two, and the respective numbers of such
 Introduction 5

Figure 1. Languages in a cline ranging from those with nothing in c o m m o n to

those that are identical.

languages cannot be determined a priori. Therefore, more accurately, our

cline should be represented by the following formulas: Let U stand for
the set of all possible human languages

U = [L„ L 2 Ln]

and let σ stand for the similarity ratio, i.e.,

σ = s/d

where s stands for the number of similar elements and where d stands
for the number of different elements, 1 and where both s and d > 0 , in
order for the L's to be comparable.
Let us assume, further, that reliable descriptions of all the languages
constituting U are available, and that these descriptions are all made
within the same theoretical framework or a set of mutually compatible
frameworks such that equivalence can be established for these frame-
6 Introduction

works. Provided these conditions are met, all languages in U can be

arranged pairwise on the scale of σ from η to m, where η > 0 and m < oo.
The grammatical paradox consists in the resulting situation, viz. that
languages are comparable only to the extent to which they are similar;
thus, the lower the σ for a given pair of languages, the less comparable
they are. It follows that the most voluminous contrastive analyses should
be performed on those languages for which σ would be the highest, while
contrastive analysis becomes increasingly impossible with the diminishing
σ. In either direction, purely structural contrastive studies reach an
impossible impasse, which cannot be overcome, even in principle. The
additional paradox, as we pointed out earlier, is that contrastive studies
in the 20th century began to flourish within the structural framework.
The paradox underlying such studies led to the situation in which,
thriving as they were, they were at the same time completely atheoretical.

The semantic paradox

Even before the difficulties connected with comparability within the

structural framework had been formulated, practising contrastive analysts
abandoned the strictly structural positions in favour of employing se-
mantic criteria in establishing comparability. At first, semantics was
employed implicitly as translational equivalence, the term borrowed from
the theory of translation as formulated by Catford (1965: 27 — 34) and
used in the context of contrastive studies by Halliday et al. (1964:
111 — 134). Translation equivalence was employed as the main criterion
in deciding what constructions and sentences in various languages are
comparable. The following principle was formulated:
If the items are not at least sometimes equivalent in translation, they
are not worth comparing (Halliday et al. 1964: 115).
This clearly implied the presence of semantics on the contrastive scene,
since translational equivalence presupposes at least some degree of se-
mantic resemblance: two sentences are mutually translatable if they share
at least some semantic properties.
Later, a more radical hypothesis was formulated. It claimed that
equivalent sentences across languages have identical semantic represen-
tations. This hypothesis was connected with the so-called universal base
hypothesis (Bach 1968: 91; Fillmore 1968: 51 ff.), which claimed that all
 Introduction 7

sentences in all languages are derived from a universal semantic base, for
which various metalanguages of representations were proposed, such as
a modified predicate calculus (Fillmore 1968), a system of roles (Fillmore
1968), or a system of labelled graphs (Krzeszowski 1974). All these
proposals had one thing in common: they assumed the existence of some
universal, underlying semantic representation, free of language-specific
syntactic categories, from which all sentences in all languages are derived
through language-specific categorical and syntactic rules. These rules can
be compared and contrasted in various languages, which provides a new
dimension for contrastive studies (see also Lipinska 1975: 50 ff.).
The hypothesis concerning the identity of semantic representations of
equivalent sentences leads to the semantic paradox, which is based on
the fact that what is identical is not subject to comparison, and what is
different is not comparable. Since equivalent sentences across languages
have identical semantic representations, it follows that differences occur
at less abstract levels, due to the operation of different, language-specific
grammatical rules. Therefore, languages differ at more superficial levels,
while in their deeper structure they are increasingly similar, and at the
level of semantic representation they are presumably universally identical.
Thus, the most interesting linguistic insights and generalizations provide
the least promise for contrastive studies since with the increasing univer-
sality of the grammar (whether expressed in terms of the universal base
hypothesis or in quite different terms, such as more recent versions of
the standard theory, as expounded by Chomsky 1975, 1982, 1984), there
is less and less for contrastive studies to deal with. Here is how Preston
formulates this paradox:
That the drive for universality should deny comparative detail between
even related languages should seem paradoxical is understandable at
a superficial level, but if we recall Chomsky's assertion that universal
grammar will eventually leave only idiosyncratic odds and ends and
irregularity behind in particular grammars, the conclusion that better
grammars provide less and less detail for contrastive analysis is self-
evident (Preston 1975: 69).
Therefore, what is left for comparison are essentially incomparable
idiosyncracies, such as suppletion, pronominal irregularities, morpholog-
ical curiosities, and specific selection features, as the only items of any
interest in the grammar of a particular language. According to Preston,
even if contrastivists do trouble themselves with all these idiosyncratic
details across languages, they will "make no real contribution to linguis-
tics" (Preston 1975: 65).
8 Introduction

The pedagogical paradox

Accepting the inevitable conclusion that what is universal must be familiar

to all foreign-language learners from their native languages and consti-
tutes the main focus of interest of theoretical linguists, we must also face
the conclusion that what remains to be learned are "non-comparable bits
of peculiar and idiosyncratic information generally ignored in contrastive
studies" (Preston 1975: 65).
So the pedagogical paradox is that what theoretical linguists consider
as trivial and uninteresting may be of utmost pedagogical importance.
Language teachers all over the world are particularly sensitive to errors
in the realm of morphology, transgressions against rare, exceptional forms
or unique pronunciations. All these phenomena, mainly connected with
surface structures, traditionally constitute the learner's hell and, if exces-
sively focused on, occasion many a failure in learning foreign languages
in classroom situations. Languages abounding in such phenomena have
the reputation of being "difficult" in contrast to those languages which
are relatively free of such surface phenomena as inflections.
Neither the semantic nor the pedagogical paradox undermine the
validity of contrastive studies since there is no way of telling what is
universal and what is idiosyncratic without conducting thorough con-
trastive studies of as many languages as possible. One cannot accept the
view that
a thorough investigation of [!] large number of languages is really
unnecessary for the universal features necessary to explanatory
adequacy are natural by-products of close, descriptively adequate work
on the structure of a particular language (Preston 1975: 67).
T. Zabrocki presents some convincing evidence demonstrating that
certain theoretical linguistic hypotheses cannot be verified without ref-
erence to cross-language data. In fact, Zabrocki is of the opinion that in
order to test any claim assigning a universal value to particular syntactic
rules or categories one should ideally test these claims "on a number of,
possibly all, languages, within general theoretical linguistic studies" (Za-
brocki 1976: 101).
The monograph presented here attempts to resolve these paradoxes
by suggesting such methods in contrastive studies which would be theo-
retically plausible and practically useful. In this way problems and con-
troversies beclouding contrastive studies may lose some of their edge,
which will give contrastive studies a chance to survive as an art, if not
as an algorithm.
Chapter I

What is contrastive linguistics?

When two or more languages are compared, it is possible to focus either

on similarities or on differences. 1 When a learner learns a new language,
he usually focuses attention on differences and remains largely unaware
of similarities. If he discovers some similarities, he is amused and surprised
since he ordinarily does not expect to find them. Grammarians, on the
other hand, quite early became interested in discovering what various
languages have in common, in the belief that making such similarities
explicit for the learner may facilitate the process of foreign language
learning. Early contrastive studies were motivated precisely by this as-
sumption (Krzeszowski 1985: 485).
There are several approaches to linguistic comparisons. The 19th
century witnessed the development of historical linguistics and the related
field of comparative historical studies, which aimed at finding the com-
mon genetic background for whole groups of languages. Another field
of linguistic comparisons emerged when languages were grouped on the
basis of various characteristics which they share. The area of linguistics
concerned with such comparisons is called typological linguistics. Lan-
guages grouped together in the same typological group need not be
genetically (historically) related. For example, English and Chinese, which
are not genetically related, share a large number of grammatical prop-
erties, such as relatively fixed and grammatically constrained word order,
paucity of inflections, and prominence of function words. These shared
features place the two languages quite close in the typological groupings,
in spite of the genetic distance separating them.
Contrastive linguistics is connected with yet another kind of compar-
ison: noting and describing similarities and differences in languages rather
than grouping them genetically or typologically. Ultimately, of course,
all kinds of comparisons may yield results which are relevant to linguistic
theory in general, as in the search for linguistic universals. Therefore,
differences between typological and contrastive linguistics are largely a
matter of focus: typological linguistics focuses on clusters of languages
10 Chapter I

united by some common feature or features, while contrastive linguistics

focuses on pairs of languages and explores similarities as well as differ-
ences between them.
Contrastive linguistics, like descriptive and historical linguistics, is
dependent on theoretical linguistics since no exact and reliable exploration
of facts can be conducted without a theoretical background, providing
concepts, hypotheses, and theories which enable the investigator to de-
scribe the relevant facts and to account for them in terms of significant
generalizations. But contrastive linguistics is also dependent on descriptive
linguistics since no comparison of languages is possible without their
prior description. In brief, then, contrastive linguistics is an area of
linguistics in which a linguistic theory is applied to a comparative de-
scription of two or more languages, which need not be genetically or
typologically related. The success of these comparisons is strictly depend-
ent on the theory applied. As will be seen later, in extreme cases, the
linguistic framework itself may preclude comparison. Therefore, contras-
tive linguistics imposes certain demands on the form and nature of the
linguistic theory which is to be "applied" in such comparisons. In many
less extreme situations the results of comparisons are strictly dependent
on the theoretical framework adopted in the comparisons (see Chapter
VI).
Originally, all contrastive studies were pedagogically motivated and
oriented. In recent years, however, distinctions have been drawn between
"theoretical" and "applied" contrastive studies (see Fisiak 1981: 2 — 9).
According to Fisiak
Theoretical CS give an exhaustive account of the differences and
similarities between two or more languages, provide an adequate model
for their comparison, determine how and which elements are compa-
rable, thus defining such notions as congruence, equivalence, corre-
spondence, etc. Applied CS are part of applied linguistics. Drawing
on the findings of theoretical contrastive studies they provide a frame-
work for the comparison of languages, selecting whatever information
is necessary for a specific purpose, e.g. teaching, bilingual analysis,
translating, etc. (Fisiak 1981: 9).
In Chapter XII we shall investigate to what extent this is a valid
distinction. Assuming, however, that "applied contrastive studies" are
sufficiently distinct from "theoretical contrastive studies", the former, as
part of applied linguistics, especially when related to teaching, must
necessarily depend not only on theoretical, descriptive, and comparative
 What is contrastive linguistics? 11

linguistics but also on other disciplines relevant to teaching; among them

are psycholinguistics, sociolinguistics, didactics, psychology of learning
and teaching, and possibly other areas which may be important in ways
difficult to evaluate at the present moment.
Finally, some comments are needed about terminology. Although the
word "contrastive" is used most frequently with reference to cross-
language comparisons of the sort described above, various authors have
been trying to replace it with other terms, such as "cross-linguistic
studies", "confrontative studies", and some even more esoteric terms, for
example, "diaglossic grammar" (Dingwall 1964 a), which enjoyed but a
brief existence. The word "contrastive" is likely to outlive all the com-
peting terms since it appears in titles of monographs and collections of
papers on the subject (cf. James 1980; Fisiak 1980, 1981, 1984).
Terminological complications manifest themselves in a rich array of
nouns which have been attested in collocation with the adjective "con-
trastive" In the current literature we find such terms as "contrastive
linguistics", "contrastive studies", "contrastive analysis", and "contrastive
grammar"
Although consistency is certainly wanting, there is an observable
tendency to select a particular collocation to refer to particular domains
of cross-language comparisons. And so the term "contrastive studies"
appears to be the least marked, as it fits all contexts in which other
collocations with "contrastive" are also appropriate. The term "contras-
tive linguistics" is also often used with reference to the whole field of
cross-language comparisons, with a slight tendency to focus on those
instances when theory or methodology of comparisons come into play.
The collocation "contrastive analysis" is often used interchangeably with
the above two terms, but there is a tendency to restrict its scope of
reference to comparison proper. In that restricted sense "contrastive
analysis" would refer to the third of the three steps in classical contrastive
studies, viz. description, juxtaposition, comparison proper, respectively
(see Chapter IV).
Finally, the collocation "contrastive grammar" is often used to refer
to the product of contrastive studies, as a bilingual grammar highlighting
differences across languages. In this sense, "contrastive grammars" con-
stitute an outcome of "contrastive linguistics"
Whether these observations are accurate or not, it would be desirable
to aspire towards some consistency in the use of these terms along the
lines suggested in the above generalizations.
12 Chapter I

Unfortunately, these observations do not exhaust all the contexts in

which the adjective "contrastive" appears. Many other collocations can
be encountered in an astonishingly varied assortment. Each such col-
location consists of the adjective "contrastive" followed by all manner
of nouns; so we get "contrastive pragmalinguistics" "contrastive soci-
olinguistics", "applied contrastive studies", "contrastive discourse anal-
ysis" "contrastive pragmatics", "contrastive syntax", "pragmatic con-
trastive analysis" "contrastive generative grammar", "contrastive phon-
ostylistics" "contrastive language studies", "contrastive lexicon", "con-
trastive considerations", "transformational contrastive studies", "theo-
retical contrastive studies", "classical contrastive studies", "deeper and
deeper contrastive analysis", "contrastive description", and quite a few
more.
The existing confusion is typical of a field where insufficient atten-
tion has been paid to various matters of principle. Preoccupied with
analytic details, investigators often lose sight of general distinctions
and aims of their research, and they do not question certain funda-
mental assumptions, which are often taken for granted. In contrastive
studies, the situation is further aggravated by the deeply-nourished
conviction of many practitioners in the field that theoretical problems
of relevance arise and can be solved only in the domain of pure and
descriptive linguistics. In the view of these linguists, contrastive studies
are merely a set of procedures involving mechanical application of
various findings in theoretical and descriptive linguistics. Implicit in
this position is the belief that contrastive studies do not require any
special theoretical framework to be fully effective and to bring forth
whatever results are expected of such analyses.
There is a handful of linguists concerned with contrastive analyses,
who occasionally feel guilty of the "sin of omission", and they stop to
ponder over certain vagaries of comparative procedures, trying to see
them in more general terms than those available from actual practice
(e.g., Van Buren 1974; Schwarze 1978; Kühlwein 1983). These few in-
vestigators are aware that contrastive linguistics requires its own theo-
retical framework in addition to specific linguistic models employed in
the description of the compared languages.
One of the purposes of this book is to remedy this situation by
highlighting and discussing some crucial issues which bedevil contrastive
linguistics. We are going to address notorious problems such as tertium
comparationis, equivalence, the relation between linguistic theory and
 What is contrastive linguistics? 13

contrastive linguistics as well as between contrastive linguistics and for-

eign language teaching. With respect to the existing terminological jungle,
referred to earlier on, we are going to suggest a taxonomy of contrastive
studies, which will provide motivation for some terms and render other
terms superfluous.
Chapter II

Tertium comparationis1

To compare them would be tantamount to

putting ten-ton lorries and banana skins
in the same class on the grounds that nei-
ther ought to be left on footpaths!
Carl James

One of the reasons why contrastive studies continue to perform the role
of the Cindarella of linguistics is the fact that its most fundamental
concept, tertium comparationis, remains as hazy as ever. The existing
contrastive analyses involve various platforms of interlinguistic reference,
determined by specific linguistic models which they employ and specific
levels of analysis which they embrace. Thus different tertia comparationis
are used for comparisons in lexicology, in phonology, and in syntax. In
few of these studies is explicit mention of any tertium comparationis made
or any justification for a specific choice presented. 2
All comparisons involve the basic assumption that the objects to be
compared share something in common, against which differences can be
stated. This common platform of reference is called tertium comparationis.
Moreover, any two or more objects can be compared with respect to
various features and, as a result, the compared objects may turn out to
be similar in some respects but different in others. Thus, a square and a
rectangle are similar in that both consist of four sides at right angles.
But they are also different, since in a square, but not in a rectangle, the
four sides are of equal length. 3 If we compare squares and rectangles
with respect to the angles, we ascertain that the two types of figures are
identical. If, on the other hand, we consider the length of their sides, we
find them to be different. Depending on the platform of reference (or
tertium comparationis) which we adopt, the same objects turn out to be
either similar or different.
In cross-language comparisons, the choice of tertium comparationis
will also constitute the determining factor in establishing similarities and
differences between the phenomena compared (cf. Lipinska 1975: 48;
Fisiak et al. 1978: 15). Since language is a complex hierarchical structure,
operating at various levels of organization, and since it manifests itself
16 Chapter II

as texts produced by its users, every aspect of language at every level of

organization, as well as every text and its constituents, can undergo
comparison with equivalent elements in another language. Therefore,
various kinds of contrastive studies can be distinguished, depending on
the tertium comparationis adopted and the kind of equivalence involved.
Before we attempt to classify contrastive studies, let us take a closer look
at various possible tertia comparationis.
Theoretical discussions tend to be limited to only two types: formal
correspondence and semantic equivalence (e.g., Lado 1957: 52 — 53; Spa-
latin 1969; Ivir 1969, 1970). Even a cursory glance at the wealth of the
existing contrastive studies suffices to notice that these two types of
tertium comparationis are not the only ones that are used in practice.
Formal correspondence and semantic equivalence can serve as tertia
comparationis for certain types of contrastive studies, such as syntactic
and lexical. Other types of contrastive studies, for example phonological,
pragmatic, or quantitative, must be based on other tertia comparationis.
Moreover, neither contrastive studies based on formal correspondence
nor those based on semantic equivalence are free from difficulties. For
example, it has been pointed out that formal likeness alone cannot serve
as a tertium comparationis without support from semantic equivalence
(Liston 1970: 44; Lipinska-Grzegorek 1977: 1 — 10). At best a comparison
based on formal criteria alone is incomplete, at worst it cannot be
performed at all, and in many cases it is misleading (see also Spalatin
1969: 3 1 - 3 4 ) .
If, for example, one compares Polish and English personal pronouns, a
formal analysis will ascertain the equivalence between the English you and
the Polish ty/wy and will be accurate as far as it goes. But such an analysis
is incomplete as it leaves out such forms as Pan/Pani and other possible
equivalents of you. These equivalents can only be established if other than
formal criteria are employed (see Chapter IV). English articles cannot be
compared to anything in those languages in which there are no articles, if
only formal criteria are considered. Finally, in the case of such phenomena
as the present perfect tense in English and passe compos0 in French, a formal
analysis is misleading since the formal similarity is not matched, at least in
this case, by semantic similarity, which creates a kind of situation which
often causes considerable learning problems (see Politzer 1968). Therefore,
it is generally recognized that a contrastive analysis based on purely formal
criteria falls short of both theoretical and practical expectations. We shall
return to this problem in Chapter VI.
 Tertium comparationis 17

Somewhat less obviously, a contrastive analysis based on semantic

similarity alone can also be inadequate and misleading. In the contrastive
practice, semantic equivalence is often erroneously identified with trans-
lation equivalence:

To establish that these [systems of deictics] are comparable, we first

need to show their contextual equivalence; this can be done most
simply by reference to translation (Halliday et al. 1964: 115).
Chapter VII will discuss differences between semantic equivalence and
translation. It will be shown that translation equivalents are often se-
mantically non-equivalent. At this point, it must only be noted that
semantic equivalence must be constrained formally, while translation
equivalence may, but does not have to, be thus constrained. When one
translates, one departs from semantic equivalence due to three types of
reasons: (1) errors in translation; (2) formal properties of respective
languages; and (3) what is loosely called "stylistic" reasons. These three
types of reasons lead to situations in which actual translation practice,
with the exception of that concerned with legal texts, seldom involves
semantic equivalents in the sense defined below in Chapter VII. This
means that only some translations can be used as data for systematic
contrastive studies (cf. Ivir 1969), while translation as a method of
contrasting must be regarded with caution:

Translation must be viewed amorphously as the rendition of a text

from one language to another. This is translation from the standpoint
of la parole: the text, the act of speech or writing, is the thing. Or it
may be viewed as a systematic comparison of two languages: this is
translation from the standpoint of la langue (Bolinger 1966: 130).
In fact, the use of translation in systematic contrastive studies is highly
limited:
Translation equivalence serves merely to help us isolate items of struc-
ture with shared meanings in the two languages (Ivir 1970: 15).
Even if we do distinguish translation equivalence from semantic equiv-
alence and base contrastive studies on the latter, we still face problems.
As has been stated earlier, semantic equivalence involves "formal" con-
straints. Thus, semantic equivalence is inherently connected with at least
some degree of formal correspondence. But the meaning of both concepts
is richer than is commonly recognized in contrastive studies. For instance,
"formal" can be extended to cover the entire plane of expression (cf.
18 Chapter II

Hjelmslev 1961: 59), whereas in most American studies the word "formal"
is restricted to word order, function words, inflections, affixation, and
suprasegmentale. In a broader perspective, "formal" would also embrace
such aspects of expression as aliteration, rhymes, and rhythm. Many of
these "formal" properties would find their place in the study of function-
ally (pragmatically) equivalent texts (see Chapter XI).
The notion "semantic" is also often extended to cover matters of
pragmatics, especially by those authors who identify semantic equivalence
with translation equivalence:
Our experience is that languages can be effectively contrasted only on
semantic basis, specifically, on the basis of translation equivalence
(Spalatin 1969: 34).
In reality many authors have shown that semantic equivalence is not
a necessary prerequisite of a good translation (cf. Rülker 1973: 29 — 35;
Krzeszowski 1974: 13, 1981 a; Kopczynski 1980: 4 1 - 4 2 ) . What is ex-
pected of a good translation is pragmatic or functional equivalence (see
Chapter V). It cannot be denied that pragmatic equivalence can serve as
tertium comparationis for contrastive analyses of such matters as the
structure of discourse, stylistic properties, and quantitative aspects of
texts. But syntactic contrastive studies, the primary concern of earlier
contrastive studies, must be conducted within the limits of the semantic
component of the language, or more specifically that part of the semantic
component which can be systematically and predictably correlated with
the grammatical structure of sentences. This restricted sense of "semantic"
still embraces some aspects of meaning which are traditionally relegated
to "pragmatics" or "interpersonal function" of sentences (Halliday 1970:
143). According to Halliday, the systems of mood and modality are
precisely those systems which relate sentences to their interpersonal func-
tions. It seems obvious that the notion "sentence semantics" should cover
those elements of "pragmatics" which can be correlated with the structure
of sentences, even if consistency in this area is definitely out of the
question; declarative, interrogative, and imperative sentences do not nec-
essarily perform the functions of statements, questions, and commands,
respectively. In so far as some correlation between form and function
does exist, those "functional" aspects of sentence structure constitute the
border area between pragmatics and semantics and should be included
in any semanto-syntactic contrastive studies (for details see Krzeszowski
1974). Therefore, James (1980) suggests that for the purposes of contras-
tive analyses translation equivalents should be limited to those which are
 Tertium comparationis 19

both semantically and pragmatically equivalent. However, this proposal

also raises doubts.
Presumably, what James means by "semantic" refers to Halliday's
"ideational" function of sentences, while Halliday's "interpersonal" and
possibly "textual" functions fall under "pragmatic" Under James' pro-
posal many sentences across languages would exhibit both "ideational"
(semantic) and "interpersonal" (pragmatic) equivalence; yet one would
hardly wish to use them as data for syntactic contrastive studies. For
example:
(1) Did he kill that dog?
(2) Zabil tego psa?
(literally: 'killed-he that dog?')
are equivalent, both ideationally (agent, transitive verb, patient) and
interpersonally (general question); yet, they falsely suggest a relationship
between syntactic types represented by (1) and (2) in English and Polish.
A systematic syntactic equivalence would have to be ascertained between
(1) and (3) rather than between (1) and (2):
(3) Czy on zabil tego psa?
(literally: 'whether he killed that dog?')
since (3) typically represents interrogative sentences in Polish, just as (1)
is a typical interrogative sentence in English.
In an earlier work (Krzeszowski 1981 b: 123), I suggested that syntactic
contrastive studies should be performed on data restricted in the following
way: a contrastive grammar will take as its primary data (to be assigned
the status of semanto-syntactic equivalence) the closest approximations
to grammatical word-for-word translations and their synonymous para-
phrases, if such forms exist. Such a constraining of primary data as the
basis for syntactic contrastive studies bypasses the inherent difficulties of
the proposals suggesting the use of unrestricted semantic equivalence as
the basis for comparison. Accepting any translation as a possible basis
for syntactic contrastive studies leads to two mutually exclusive and
undesirable consequences. Either (1) no comparative generalizations be-
come possible, as the number of well-formed translations of a particular
sentence into another language cannot be predicted a priori; or (2) purely
arbitrary decisions concerning formal correspondences in unconstrained
translations must be made. Any non-arbitrary decision involves circular-
ity: the investigator has to assume formal correspondences on the basis
of syntactic and/or morphological features which the compared texts
20 Chapter II

share. This circularity is even reflected in the use of the word "compa-
rable" in certain contrastive grammars. For example, Stockwell et al.
thus write about determiners in English and Spanish:
Both English and Spanish have two sets of determiners, commonly
referred to as definite and indefinite articles. In many respects they
are comparable [emphasis is my own]; in others they are different
(Stockwell et al. 1965: 65).
The circularity consists in the following: we compare in order to see
what is similar and what is different in the compared materials; we can
only compare items which are in some respect similar, but we cannot use
similarity as an independent criterion in deciding how to match items for
comparison since similarity (or difference) is to result from the compar-
ison and not to motivate it.
To avoid this undesirable circularity, in deciding about formal corre-
spondences, one needs a common tertium comparationis outside the for-
mal properties. The underlying meaning of the closest approximations to
well-formed word-for-word translations provides such a tertium compar-
ationis. Sentences and constructions sharing identical semantic represen-
tations at the level of sentence semantics (but necessarily exhibiting certain
idiosyncratic differences at the level of word-semantics) are semanto-
syntactically equivalent and constitute a constrained set of data for
syntactic contrastive studies. The approach through constrained trans-
lations does not require the initial recognition of shared syntactic cate-
gories as tertium comparationis for syntactic contrastive studies. Such a
recognition would illegitimately anticipate the results of contrastive stud-
ies. A detailed proposal along these lines will be presented in Chapter
VIII. (See also Krzeszowski 1974 and 1979).
Summarizing, let us say that formal properties alone do not provide
an adequate tertium comparationis for syntactic contrastive studies, while
a semantic tertium comparationis must be constrained through restricting
the scope of translation equivalents as primary linguistic data for syntactic
contrastive studies.
Such constrained but rigorous contrastive studies have a very limited
pedagogical relevance. Any extension of the scope of contrastive studies
to make them pedagogically more useful increases the likelihood of their
becoming less rigorous and hence less respectable as a "scientific" pro-
cedure. One has to look for ways of extending the scope of contrastive
studies without losing any of the rigour characterizing syntactic contras-
tive studies. Formal and semantic tertia comparationis, discussed so far,
 Tertium comparationis 21

will not suffice as bases for extended contrastive studies. For example,
phonetic and phonological contrastive studies cannot rely on semantic
equivalence as tertium comparationis.
The crucial notion in identifying various kinds of tertia comparationis
and determining their character is the concept of equivalence or the
relation which provides justifications for why things are chosen for
comparison, keeping in mind that only equivalent items across languages
are comparable. The various principles motivating equivalence and, eo
ipso, contrastive studies will provide grounds for dividing tertia compar-
ationis and, consequently, contrastive studies into various categories, each
being connected with a specific kind of equivalence which motivates the
comparisons (see Chapter III). In other words, equivalence is the principle
whereby tertium comparationis is established inasmuch as only such ele-
ments are equivalent for which some tertium comparationis can be found,
and the extent to which a tertium comparationis can be found for a
particular pair of items across languages determines the extent to which
these elements are equivalent. Thus, equivalence and tertium compara-
tionis are two sides of the same coin.
Chapter III

Towards a classification of contrastive studies1

Contrastive studies can be divided and subdivided according to various

criterial principles. Fisiak's division into "theoretical" and "applied"
contrastive studies is based partly on the aims of contrastive studies and
partly on their methodology. 2 According to Fisiak, theoretical contrastive
studies are performed for their own sake, while applied contrastive studies
are performed for the purpose of some application. But Fisiak reinforces
this division by claiming that applied contrastive studies are directional
while theoretical contrastive studies are not. This roughly means that
theoretical contrastive studies

do not investigate how a given category present in language A is

represented in language Β they look for the realization of a universal
category X in both A and Β (Fisiak et al. 1978: 10).

Another kind of taxonomy can be based on specific linguistic models

applied in the description of languages involved in contrastive studies.
Since contrastive studies can be conducted in a variety of models, we can
speak of structural, transformational, stratificational, or systemic con-
trastive studies. Yet not all models are equally suitable as frameworks
for contrastive studies. In extreme cases, a particular model may turn
out to be almost totally useless (cf. Van Buren 1974: 293; Lipinska 1975:
7ff.). A survey of some models as applied to contrastive studies and their
evaluation in terms of their suitability as frameworks for contrastive
studies will be presented in Chapter VI. It is important to note here that
a taxonomy of contrastive studies based on criteria external to compar-
isons themselves can lend credence to linguistic theories and models rather
than to contrastive studies. Therefore, external criteria will not be con-
sidered in the taxonomy which we are going to suggest later in this
chapter.
An interesting though incomplete taxonomy of contrastive studies is
presented by Di Pietro (1971: 17 — 19), who divides contrastive studies
24 Chapter III

into autonomous and generalized, on the one hand, and into taxonomic
and operational, on the other.
In autonomous contrastive studies no conscious, explicit reference is
made to any universal, underlying structure which the compared lan-
guages might share. In such contrastive studies each language is described
independently and in its own right. In generalized models, explicit ref-
erence is made to those layers of structure which the compared languages
share, not only on account of their typological or genetic similarity, but
mainly because of the universal grammar which is believed to underlie
all human languages.
The division into taxonomic and operational models pertains only to
generalized models. Taxonomic models are restricted to stating similarities
and differences across languages and to stating their "hierarchical im-
portance" Operational models seek to formulate

a series of conversions performed on the source language in order to

produce the forms of the goal language (Di Pietro 1971: 18).

This procedure would lead to the formulation, in linguistic terms, of

the steps which would have to be taken by the learner to acquire a foreign
language. Ideally such models would lead to the formulation of algorithms
of foreign language acquisition.
From the pedagogical point of view, autonomous contrastive studies
have not been of much use in the area of language teaching since neither
the descriptive apparatus that they employ nor the results which they
yield can be easily related to any psychological or pedagogical reality.
On the other hand, operational models are extremely difficult, if at all
possible, to construct since still not enough is known about both learners
and grammars with relation to psychological reality.
In principle, Di Pietro's taxonomy could be augmented by generative
models, i.e. such models which generate equivalent constructions and
sentences across languages and assign appropriate structural descriptions,
including similarities and differences, to the enumerated equivalents. A
detailed proposal along these lines will be presented in Chapter VIII.
Since our own classification of contrastive studies is based on distin-
guishing various kinds of equivalence (and hence tertium comparationis),
it is apropriate to mention Kühlwein's typology of equivalence, designed
in connection with lexicological contrastive studies situated in the context
of socio-semiotics (Kühlwein 1983). His typology embraces the following
kinds of equivalence:
 Towards a classification of contrastive studies 25

formal — based on linguistic structure;

derivational-semantic — connected with the "depth" of the der-
ivation;
paraphrase procedure — which yields "regressum ad infinitum"·
translation procedure — limited by truth conditions and culture-
specific considerations; and
functional-communicative — involving "mental processes of cog-
nition and associative connotative components" (Kühlwein 1983:
6).

This division suffers from the following shortcomings:

a) It is not based on a set of homogeneous criteria: e.g., paraphrase
procedure seems to be only loosely connected with contrastive studies;
b) it does not mutually relate these various types of equivalence, nor
does it suggest that there is a relationship;
c) it leaves out certain other types of equivalence, for example, substan-
tive equivalence;
d) it does not explicitly relate the concept of equivalence to the concept
of tertium comparationis, although, admittedly, Kühlwein is certainly
aware of the importance of the relationship between the two concepts,
at least in the preamble to his typology.
The taxonomy which we are going to present is intended to be free of
these inadequacies. It is based on the assumption that various kinds of
contrastive studies can be distinguished in a strict relation to various
tertia comparationis adopted and, consequently, to various kinds of equiv-
alence.
The first division is drawn between text-bound and systematic (or
projective) contrastive studies. It is based on the familiar distinction
between la parole and la langue. Text-bound studies involve comparisons
of texts in two (or more) languages and do not go beyond such texts to
generalizations about grammars, i.e. rules and systems that generate
those texts. Projective contrastive studies are related to text-bound con-
trastive studies in the same way in which the study of language is related
to the study of texts. Such studies go beyond primary linguistic data
found in texts in order to grasp and formulate generalizations about
various aspects of the compared languages.
At this point, it is useful to introduce the term 2-textj'tu:tekst/ to refer
to any pair of texts, written or oral, in two languages, which are used as
data in contrastive studies. Every 2-text can be described in terms of a
26 Chapter 111

binary distinction: [ +translation] (henceforth [ ± trans]). A 2-text marked

as [ +trans] is such a 2-text of which it can be asserted that its constituent
texts can function as translations. Such 2-texts usually provide data for
qualitative contrastive studies, which constitute the main bulk of con-
trastive studies. 2-texts which are not translations, marked [ — trans], can
be used as data for quantitative contrastive studies, which will be dis-
cussed in detail in Chapter X.
Text-bound contrastive studies are corpus-restricted if no systematic
generalizations outside the original data are made. Quantitative contras-
tive studies are necessarily corpus-restricted, even if they enable one to
make statistical predictions concerning other, similar texts. Quantitative
text-bound contrastive studies may also be corpus-restricted as long as
they do not aim at drawing systematic generalizations about the languages
of the 2-text. But they may also serve as basis for projective generaliza-
tions, if clearly stated constraints on the selection of the relevant 2-texts
are formulated and implemented. The relevant 2-texts serve as linguistic
data on which contrastive grammars as generalizations about differences
and similarities in the compared languages are based.
We can now see that tertium comparationis is in fact the reason why
any two texts are brought together as a 2-text and/or why any two items
in two languages are juxtaposed for comparison. Each type of contrastive
studies has its own type of tertium comparationis. Within each type of
tertium comparationis, it is possible to distinguish more specific subtypes,
subsubtypes, etc. unique within each type. Each type of tertium compar-
ationis is connected with a specific type of equivalence.
Text-bound contrastive studies may involve statistical comparative
studies, and, as was said earlier, the relevant 2-texts need not be [ +trans].
However, to prevent comparisons of incomparables, one has to establish
a tertium comparationis (and consequently an equivalence). The tertium
comparationis will restrict the class of texts that can undergo comparisons.
Thus, it may be necessary to require that texts constituting a particular
2-text, be written in the same register or at least deal with the same topic
or represent the same literary genre. Whatever requirements on the
"sameness" of the constituent texts are imposed, they will determine the
tertium comparationis relevant in these texts. If the compared texts are
translations, no additional requirements are necessary. Statistical equiv-
alence can be established on 2-texts which are either [ + trans] or [ — trans],
but in the latter case the extra requirements, referred to above, must be
met. Statistical equivalence, marked as (1) in Fig. 1 below, obtains be-
tween various systematically equivalent items (see below) which appear
 Towards a classification of contrastive studies 27

in 2-texts and which have maximally similar frequencies of occurrence.

Let us note that in order to qualify as statistically equivalent, two items
across languages need not be in the strictly delimited semanto-syntactic
equivalence (relation type (4) below); but, to be comparable at all, they
have to be equivalent in some sense. In many instances, however, statis-
tically equivalent constructions are not semanto-syntactically equivalent.
Consider as an example the English participial constructions in such
sentences as
(1) I saw Peter entering the house.
The semanto-syntactic equivalent of (1) in Polish is
(2) Widzialem Piotra wchodzgcego do domu.
as the closest approximation to an acceptable word-for-word translation
of (1) (cf. Krzeszowski 1974,1979,1981 b). Yet, if we consider quantitative
data, (1) and (2) will turn out to be non-equivalent on at least two counts:
intralinguistic and interlinguistic (contrastive). The intralinguistic count
for English and an analogous count for Polish will reveal that the
frequency of occurrence of the construction NP\ V NP2 Ving X, repre-
sented by (1), in comparison with other English verb-complement con-
structions, is higher in English texts than the frequency of occurrence of
the semanto-syntactically equivalent Polish construction NPt V NP2 Vgcy
X in comparison with other Polish complement constructions appearing
in Polish texts. In Polish texts such constructions as 7VP, V Subordinate
S (" Widzialem, jak Piotr wchodzil" — "I saw how Peter entered") or NP]
V Nom Ngen ("Widzialem wejscie Piotra" "I saw Peter's entry") are
more frequent. Thus, although the nearly congruent Polish construction
(2) is available, it is less frequently used than other semantically similar
constructions, which are more favoured by native users of Polish. 3
The same result can be obtained cross-linguistically and more directly
by looking at various translations of a given construction into another
language. The equivalent construction which is most frequently relative
to other, nearly synonymous constructions, will also be used most fre-
quently in translations. Thus, statistical comparisons can be conducted
both on texts which are attested as translations and on texts which are
not translations but are comparable on account of being written on a
similar topic, by similarly qualified authors using similar registers, etc.
We can generalize the above remarks about statistical equivalence as
follows: two linguistic items across languages are statistically equivalent
if they occur as the most frequent translations of each other and/or if,
28 Chapter III

in comparison with other synonymous constructions, they have maximally

similar frequency of occurrence in the relevant texts.
2-texts with the feature [ +trans] consist of texts which are in the
relation of translation equivalence, marked as (2) in Fig. 1. It is important
to note that such translations need not be "correct" or "acceptable" In
fact, they often display considerable deviations from other kinds of
equivalence. In particular, translations are usually at variance with what
we shall later on describe as semanto-syntactic equivalence. Such devia-
tions are not necessarily due to errors in translation, but often have their
source in various pragmatic considerations, which override the demands
of semanto-syntactic equivalence. All the same, it frequently happens
that particular texts which purport to be translations exhibit errors due
to the translator's incompetence, negligence, and other such factors.
Usually such data will be rejected by investigators exploring translation
equivalents, unless pathology of translation becomes the focus of interest
and translation errors become the object of study. The latter pursuit does
not strictly fall within the scope of contrastive studies. More will be said
about the relation between translation and semanto-syntactic equivalence
in Chapter VII.
Systematic contrastive studies involve comparisons of constructions,
systems, and rules. Of these three types, contrastive studies of construc-
tions are based on semanto-syntactic equivalence, which constrains 2-
texts for the purpose of such studies in the way referred to above. More
will be said about this type of equivalence in Chapter VII.
The comparison of systems requires resort to syntagmatic considera-
tions, because any paradigmatic analysis, i. e., any analysis involving
systems, must be linked to a syntagmatic analysis since isolating a par-
ticular system in a particular language requires an analysis of syntagmatic
arrangements of linguistic units, i.e., of constructions. For example, in
order to isolate the system of personal pronouns in English in the
nominative and oblique forms, it is necessary first to investigate various
constructions in which such pronouns appear and then generalize on the
basis of such syntagmatic data by extracting the paradigmatic system of
personal pronouns. In order to obtain an equivalent system in another
language, one must investigate equivalent constructions in that language
and extract the relevant paradigmatic system from the data. Therefore,
equivalence of systems cannot be ascertained without the foundation
provided by the equivalence of constructions in which elements of these
systems appear. In contrastive practice, equivalent systems across lan-
guages are usually juxtaposed on the basis of the investigator's intuition
 Towards a classification of contrastive studies 29

(his knowledge of the two languages), corroborated by the established

grammatical tradition expressed in the common terminology. It is usually
tacitly assumed that such things as pronouns in one language and pro-
nouns in another language are comparable by virtue of the common label
"pronoun" used to refer to a specific set of words in one language and
a specific set of words in another language. In this way, the concept of
system equivalence (3) is employed as tertium comparationis. However,
system equivalence is dependent on construction equivalence, since the
former can be made explicit only through the examination of construc-
tions in which the elements of the compared systems appear, i. e. via the
notion of semanto-syntactic equivalence, relating the relevant 2-texts as
primary data.
Likewise, any comparison of rules cannot be divorced from an implicit
comparison of constructions on which these rules operate. Most rules
have a construction as input and a construction as output. Therefore,
semanto-syntactic equivalence also underlies rule equivalence (5). We
shall return to this sort of equivalence in Chapter VI.
Phonological and lexical contrastive studies are based on the type of
tertium comparationis which can be called substantive (6) insofar as it is
connected with the material substance outside language, with which
language is joined through its phonological interface, on the one hand,
and through its semantic interface, on the other (Hjelmslev 1961). In the
case of phonological contrastive studies, acoustic, articulatory, and, in
principle, auditory phenomena provide the substantive tertium compar-
ationis. Most phonological contrastive studies make reference to articu-
latory parameters, less frequently to acoustic ones, and never to auditory
parameters alone.
In the case of lexical contrastive studies, the external reality, or, strictly
speaking, its psychic image in the minds of language users, provides the
substantial lexical tertium comparationis. In this way, lexical items across
languages are compared with respect to differences and similarities con-
cerning their reference to various elements of the reality in the world at
large and the ways in which this reality is reflected in the minds of
language users.
Both phonological and lexical contrastive studies are mainly paradig-
matic, even if, especially in various kinds of generative frameworks,
syntagmatic arrangements, and consequently rules accounting for those
arrangements, are also comparable. However, in the case of phonological
and lexical contrastive studies, we are faced with a situation which is a
mirror image of semanto-syntactic studies: paradigmatic studies are now
30 Chapter III

central, while syntagmatic studies are secondary and impossible to con-

duct without ultimate reference to substantive tertia comparationis.
Semanto-syntactic equivalence is not required in the case of 2-texts
compared in respect of their styles or registers. Such texts need not be
semanto-syntactically equivalent but must be acceptable translations,
which means that they have to be pragmatically equivalent. Obviously
erroneous translations must naturally be disregarded, as well as those
translations which fail to convey some relevant pragmatic functions,
especially if alternative, more adequate translations are available. Al-
though constraints on the suitability of 2-texts for stylistic contrastive
studies are less rigorous than the constraints imposed on 2-texts as data
for syntactic contrastive studies, they are no less important and must be
stated clearly and unequivocally, lest contrastive studies fail to grasp the
relevant generalizations concerning the pragmatic aspect of the compared
texts. Pragmatic equivalence (7) as tertium comparationis for stylistic and
sociolinguistic contrastive studies is thus a relation that holds between
constituent texts of 2-texts selected in such a way that they evoke maxi-
mally similar cognitive reactions in the users of these texts. Becka (1978)
describes this sort of relations in the following way:

Since, however, contrasting does presuppose, apart from agreements

and differences under observation, also a certain common foundation,
one proceeds from the fact that the fundamental functional differen-
tiation of styles holds in roughly the same way for cultured languages
and that it makes itself felt in roughly the same tendencies even though
not always realized by the same means of expression. In other words:
French scientific style will be characterized by analogous stylistic
tendencies as its counterpart in Czech, in the same way as the basic
features distinguishing the belles-lettres narrative style from descriptive
style will be the same in English as, say, in Italian. This is due to the
impact of social, i. e. extralinguistic communicative needs which e. g.
in languages of the European cultural sphere in the given period bear
on the whole an analogous character. This is why we engage in
comparing discourses of belles-lettres prose in various languages, or
of scientific style in various languages and so on. In this manner a
certain common foundation is gained upon which contrastive analysis
of styles can be built (Becka 1978: 131 - 1 3 2 ) .

Pragmatically equivalent texts may, and actually often do, correspond

to translations and as such provide data not only for pragmatic and
 Towards a classification of contrastive studies 31

sociolinguistic contrastive studies but also for text-bound contrastive

studies.
The seven types of equivalence discussed in this chapter and the related
tertia comparationis characterizing various types of contrastive studies
are presented in Figure 2. All these types of equivalence and the related
tertia comparationis embody a number of more specific tertia compara-
tionis, unique in every kind of contrastive studies. For example, semanto-
syntactic contrastive studies employ a large number of various specific
tertia comparationis at various levels of generality, corresponding to the
level of abstraction at which a given contrastive study takes place. This
is connected with the fact that all descriptively adequate models of
language give an account of the hierarchical structure characterizing
language in both the syntagmatic and the paradigmatic dimension. This
hierarchical structure can be represented by means of various conven-
tional notations, such as trees or labelled bracketings with labels of nodes
or of brackets corresponding to linguistic units at various levels of
abstraction.
The device called Contrastive Generative Grammar, to which Chapter
VIII is devoted, is based on the claim that equivalent constructions across
languages are derived from identical semantic representations, which
means that down to a certain level of derivation, before they are diver-
sified by language-specific grammatical rules, they have identical repre-
sentations and that somewhere in the course of the derivation they begin
to differ. The exact place in the grammar where particular pairs of
constructions begin to diverge are different for various pairs of construc-
tions. Therefore, Contrastive Generative Grammar provides all the nec-
essary tertia comparationis for semanto-syntactic contrastive studies, be-
ginning with the universal, language-neutral, semantic representation,
which provides the most general tertium comparationis for all compared
constructions and ending with the Immediately Relevant tertium com-
parationis, i. e. that place in the derivation of equivalent constructions
which immediately precedes the first diversification. Given that Contras-
tive Generative Grammar is based on grammars organized along five
components: semantic, categorial, syntactic, lexical, and post-lexical, ar-
ranged hierarchically from the most abstract to the most concrete, it is
possible to expect diversifications to occur on all subsemantic levels.
According to Contrastive Generative Grammar, all equivalent construc-
tions share a certain number of rules, before they become diversified by
the first rule which is different. The input to the rule which accounts for
the first diversification in each case provides the Immediately Relevant
32 Chapter III

tertium comparationis by specifying the common grammatical categories

in terms of auxiliary symbols employed in the rules.
The Ultimately Relevant tertium comparationis for syntactic contras-
tive studies is language-neutral semantic representation, satisfying the
requirement that all tertia comparationis must be expressed in terms of
concepts outside the compared categories.
Also, pragmatic contrastive studies are connected with a hierarchy of
tertia comparationis from Ultimately Relevant tertium comparationis to
Immediately Relevant tertia comparationis for every specific pragmatic
phenomenon. Various sociolinguistic parameters, such as age, sex, edu-
cation, occupation, etc. can serve as Immediately Relevant tertium com-
parationis for comparisons of specific linguistic phenomena across lan-
guages (Janicki 1985). In such instances, translations will be used as 2-
texts and pragmatic equivalence will the Ultimately Relevant tertium
comparationis.
The concept of Immediately Relevant tertium comparationis can be
clearly seen in the domain of phonology. As was stated earlier, in pho-
nology, substantive equivalence provides the Ultimately Relevant tertium
comparationis in acoustic, articulatory, and possibly also auditory sub-
stance. If we focus on articulatory substance, most readily employed not
only in contrastive studies but also in separate language descriptions, one
can set up the Immediately Relevant tertium comparationis at various
levels of "delicacy" For example, comparing consonants in two lan-
guages, one must make sure that the Immediately Relevant tertium
comparationis delimits the scope of sounds called consonants and brings
together for comparison all those sounds which, when articulated in
isolation, require occlusion in some place of the vocal tract. At the level
of Immediately Relevant tertium comparationis, the contrastive studies
will recognize a number of sounds thus articulated in both languages, all
of them articulated with occlusion. The contrastive studies will become
more and more delicate, when more subtle distinctions are considered.
The successive Immediately Relevant tertia comparationis can be evoked
by such features as aspiration and its various degrees, voicing and its
degrees, force and its degrees, as well as various places of articulation.
These features characterize various types and subtypes of consonants in
the compared languages. When enough features are evoked, one gets
down to comparisons of individual sounds, which are characterized by
matrices of articulatory features serving as Immediately Relevant tertia
comparationis for specific pairs of sounds contrasted across languages.
 Towards a classification of contrastive studies 33

Figure 2 only shows Ultimately Relevant tertia comparationis. Each

of the seven types of constrastive studies will have its own hierarchy of
Immediately Relevant tertia comparationis, which have to be stated and
described relative to the relevant Ultimately Relevant tertium compara-
tionis and to the factual data that undergo comparisons.
We do not distinguish between pedagogically oriented and pure con-
trastive studies since we believe that this distinction is irrelevant. Whether
directional or adirectional, contrastive studies may yield results relevant
to teaching or other fields of application, but the potential implementa-
tions of contrastive studies do not, as a matter of principle, determine
the course and the direction of the analysis. Modern studies of language
and of particular languages (especially the recently flourishing cognitive
approaches) offer a very broad perspective within which language is
described, not as a set of semanto-syntactic objects called sentences
the area roughly corresponding to items (4) and (5) in Fig. 2 — but as a
symbolic organization entrenched in human experience and human so-
ciety. Conducted in this broad perspective, contrastive studies yield results
which are naturally relevant to teaching and other practical domains.
The usefulness of these studies will be properly evaluated by applied
linguists, syllabus designers and teachers. All those interested will be
responsible for transforming contrastive studies into pedagogical con-
trastive studies, or rather for writing pedagogical grammars based on the
relevant results of contrastive studies. Contrastive studies themselves will
merit the epithet "pedagogical" or "applied" only to the extent to which
proper use will be made of them in some practical domains. Pragmatic
contrastive studies and statistical (quantitative) contrastive studies have
been and will continue to be particularly fruitful in this respect.
34 Chapter III

oo
U
t/5
ω
• Ψ-4
Ό
3
U
>

-*V
-»h
C
Ο
ο

>1
00
ο
"ο
α
Ρ
Iii
νa
.bo
Chapter IV

Three steps in "classical" contrastive studies

By classical contrastive studies we mean studies in which the contrastive

procedures have no generative format of the type suggested in Chapter
VIII of this book. Classical contrastive studies, even if they are based on
generative grammars of the compared languages, are still taxonomic in
nature, since they are limited to producing inventories of differences and
possibly of similarities between equivalent systems of the compared gram-
matical structures, between equivalent sentences and constructions, and
between equivalent rules operating at various levels of derivations. Com-
parison of rules can only be conducted within those frameworks in which
the concept of rule appears, as in various versions of transformational
generative grammar.
A classical contrastive analysis consists of three steps, not always
clearly distinguishable in the analysis itself but always tacitly assumed:
(1) description; (2) juxtaposition; (3) comparison, i. e., contrastive analysis
in the strict sense (cf. Halliday et al. 1964: 113 — 114).

1. Description

No comparison is possible without a prior description of the elements to

be compared. Therefore, all contrastive studies must be founded on
independent descriptions of the relevant items of the languages to be
compared. The fundamental demand on such descriptions is that they
should be made within the same theoretical framework. It will not do to
describe one language in terms of transformational grammar and another
language in terms of, say, relational grammar and then to attempt to
compare them. The results of such descriptions will be incompatible and
incomparable.
Not all linguistic models are equally well suited as foundations of
cross-language comparisons. It seems that those models which make
36 Chapter IV

explicit references to universal categories are more suitable than those

which are connected with language isolationism, inherent in many vari-
ants of structuralism. In Chapter VI, we shall discuss some problems
which arise in various descriptive models as related to contrastive studies.

2. Juxtaposition

This step is crucial in deciding what is to be compared with what. In

classical contrastive studies, this step was based on intuitive judgments
of competent bilingual informants, who determined the material to be
compared. This sort of "bilingual competence", i.e., the knowledge of
two languages, enables one to make decisions about whether or not
element X in one language is equivalent with element Y in another
language. If the two given elements are equivalent, they are eo ipso
comparable. For example, anyone competent in English and in Polish
intuitively knows, on the basis of his "bilingual competence" that such
words as ktory, ktora in Polish and which/who in English are equivalent,
given appropriate contexts. Likewise, "bilingual competence" manifests
itself in judging the following pair of sentences as equivalent:

(1) I want John to come.

(2) Chc§ zeby Jan przyszedl.

Such judgments are taken for granted in classical contrastive studies,

so that elements recognized as equivalent are intuitively deemed to be
comparable. One of the obvious weaknesses of this approach consisted
in the lack of clearly stated principles underlying decisions about what
to compare and why. As was pointed out in Chapter II, formal resem-
blance and semantic resemblance were resorted to, but both, as we have
seen, led to circularities: similarity was presupposed before comparisons
yielded results allowing to ascertain it. In many instances, formal resem-
blance (at least at the level of surface structures) so drastically contrasts
with the disparity of meaning that comparisons based on formal criteria
alone are reduced ad absurdum. Consider one of the early examples quoted
by Stockwell et al. (1965: 40):

(3) English: I said to be sure.

(4) Spanish: Dije es tar seguro. Ί said I was sure'.
 Three steps in "classical" contrastive studies 37

Although formally very similar, (3) and (4) are semantically very
different. Therefore, they are incomparable since they do not share a
semanto-syntactic tertium comparationis. This example shows again that
formal considerations alone do not suffice in establishing comparability.
Therefore, juxtapositions based on formal criteria alone, though naturally
possible, are ill-conceived and must be discarded in contrastive studies.
In classical contrastive studies, the investigator himself often acts as
the bilingual informant and decides what to compare on the basis of his
own knowledge of the two languages. Unless more explicit criteria con-
straining the data are applied, such a procedure often leads to arbitrary
decisions, which seriously undermine the rigour required in scientific
investigations. Contrastive Generative Grammar, presented in Chapter
VIII, attempts to make explicit "bilingual competence" underlying in-
tuitive judgments of bilingual informants.

3. Comparison proper

We distinguish three basic areas of comparisons:

1. Comparisons of various equivalent systems across languages, such as
pronouns, articles, verbs, and in phonology consonants, vowels, as
well as subsystems, such as nasals, laterals, etc. depending on the
degree of "delicacy" of the grammar.
2. Comparisons of equivalent constructions, for example, interrogative,
relative, negative, nominal phrase, etc., and in phonology clusters,
syllables, diphthongs, and various distributions of sounds.
3. Comparisons of equivalent rules (in those models where the concept
of rule appears), for example, subject raising from the embedded
sentence, adjective placement, interrogative inversion, passivization,
etc., and in phonology assimilation, dissimilation, metathesis, etc. (cf.
Sussex 1976: 7 - 1 1 ) .

In each area of comparison one of three possible situations may arise:

(a) X Li = X Lj

when item X in Lj may be identical in some respects with an equivalent

item in Lj.

(b) X Li Φ X y
38 Chapter IV

when item X in L; may be different in some respects from an equivalent

item in Lj.

(c) XLi = 0y
when item X in Lj has no equivalent in Lj (Krzeszowski 1976: 90).
The words "in some respects" are very important. In cross-language
comparisons, the relative character of identity must be remembered.
Compared items can only be identical with respect to some selected
property or properties which they share. For example, the systems of
number of nouns in English, French, Polish, and many other European
languages are in one respect identical, viz. they are all based on the
dichotomy "oneness" vs. "more-than-oneness" Other, more subtle dis-
tinctions can also be made by means of numerals and quantifiers, but
the grammatical systems of those languages provide morphological means
to express just this dichotomy. In many other languages, the system of
number is in the same respect different. In Classical Greek, Sanskrit,
Lithuanian, and some other languages there is also dual number, which
is employed with reference to things coming in twos, usually various
body parts coming in pairs. In such languages the system consists of
three elements, expressing "oneness", "twoness" and "more-than-two-
ness" (cf. Hockett 1958: 234). The system of number is sometimes based
on the distinction "oneness" "fewness" "manyness", as in most
Melanesian languages, the West Indonesian language Sanir and, in the
case of personal pronouns, also in Fijian (cf. Hjelmslev 1961: 53; Hockett
1958: 234).
In Chinese, nouns have no plural inflection at all, and any concept of
plurality is expressed, if necessary, by means of quantifiers and numerals.
In contrast with any language in which nouns are inflected for number,
Chinese represents the third possibility, i.e., situation (c), distinguished
above in which no equivalent form can be attested.
With regard to comparisons of constructions we deal with the same
three possibilities. An equivalent construction may be identical in some
respects and/or different in some respects, or there may be no equivalent
construction at all. To illustrate, let us consider passive constructions in
English, Polish, German, French, and a number of other languages.
These constructions in all these languages are identical in that they consist
of an auxiliary verb, usually an equivalent of 'be', followed by past
participle forms of transitive verbs. In all these instances the passive
constructions are analytic:
 Three steps in "classical" contrastive studies 39

(5) English: The song was first recorded by Maria Callas.

(6) Polish: Piesn zostala nagrana po raz pierwszy przez Mari§ Callas.
(7) French: Pour la premiere fois la chanson a ete enregistree par
Maria Callas.
(8) German: Zum ersten Mal wurde das Lied von Maria Callas auf
Schallplatte gesungen.

In other respects, passive constructions differ in all these languages.

For example, French has a complex system of conjugation of verbs where
two verbs correspond to one auxiliary verb in English. German requires
verb-final word order, etc.
New Testament Greek and Japanese passives are different from English
passives. They do not involve any auxiliaries but are formed synthetically:

(9) Japanese: Kinoo zyunsa ga sanni korosaremasita.

'Yesterday three policemen were killed'

New Testament Greek and Japanese passives, in comparison with

English passives, represent situation (b) where an item in one language
is different in some respects from an equivalent item in another language.
Situation (c) occurs when we compare English passive constructions
with equivalent sentences in such languages as Sumerian and Basque. In
those languages, in which no formal distinction is made between transitive
and intransitive sentences, there is, consequently, no distinction between
active and passive constructions. These languages may be said to have
no passive at all (cf. Milewski 1965: 240). 1 In such instances, contrastive
studies must state what other means, if any, are employed to convey the
semantic content of passive sentences.
Within theoretical linguistic frameworks in which the concept of rule
is employed, one can also compare rules. Again, as in the case of systems
and constructions, three possibilities can be distinguished:
A rule in one language may be in some respects identical with an
equivalent rule in another language, which means that both operate on
identical inputs and that they have identical outputs. For example, the
rule of subject raising from embedded object clauses operates identically
in English and Latin, rendering the construction which is known as
accusativus cum infinitivo:

(10) Latin: Video magistrum ambulare.

(11) English: I see the teacher walk.
40 Chapter IV

In some respects, however, these rules are different in the two languages
since they operate on certain verbs in Latin but not on their English
lexical equivalents. In Polish the rule does not operate at all.
Two rules are equivalent if they operate on identical inputs. If equiv-
alent rules also yield identical outputs, they may be said to be congruent
(cf. Marton 1968 b). For example, extraposition of the subordinate clause
from the subject operates identically in English and in French:
(12) P R O N O M (itIii) (S) Pred => P R O N O M (itjil) Pred (S)
as in

(13) That the wine turned to vinegar is no surprise. => It is no surprise

that the wine turned to vinegar.
(14) Que le vin se soit transforme en vinaigre n'est pas surprenant. =>
II n'est pas surprenant que le vin se soit transforme en vinaigre.
In Polish and Italian, the respective rules are non-congruent with their
English and French equivalents as they yield different results. Notably
in Polish and Italian the underlying pronouns do not appear in the
extraposed sentences:

(15) Ν on e sorprendente che il vino sia diventato aceto.

(16) Nie jest niczym dziwnym, ze wino skwasnialo.
Due to the lack of explicitly stated tertia comparationis expressed in
some universal terms (such as a semantic representation of the compared
items), typical classical contrastive studies were directional: depending on
the aims of a particular contrastive analysis, one could start with a
description of linguistic forms in L, and match them for comparison with
equivalent items in L 2 . Or alternatively, one could begin with a description
of linguistic forms in L 2 a n d look for their equivalents in L u Presumably
an exhaustive contrastive study covering the entire grammars of both
languages has to be bi-directional. Typical directional contrastive studies
would bear such titles as {System XI Construction Y) in L, and itsj their
equivalents in Lh for example, "English modal auxiliaries and their equiv-
alent constructions in Polish" (Kakietek 1980), "On some properties of
action nominals in Polish and their English equivalents" (Lewandowska
1975), "Tags in English and equivalent constructions in Polish" (Oleksy
1977), "The equivalents of the Finnish passive voice in English" (Kart-
tunen 1977), etc.
Establishing semantic tertia comparationis creates the possibility of
adopting an alternative approach, which consists in selecting a concept
 Three steps in "classical" contrastive studies 41

and examining the ways in which it is realized through various gram-

matical means in the compared languages. Typically, such contrastive
studies have titles of the form Ways of expressing a category X in L, and
Li. For example, "The expression of future in English and Serbo-Croa-
tian" (Kalogjera 1971), "Ways of expressing cause in English and Polish"
(Danilewicz 1982) or "Directives in English and Finnish" (Markkanen
1985).
Either organization of contrastive studies can be adopted in both
"theoretical" and "applied" studies. There are cases of non-directional,
applied (pedagogical) contrastive studies (Sharwood-Smith 1975) as well
as non-directional theoretical contrastive studies (Kryk 1987). Likewise,
there are examples of directional applied contrastive studies (Karttunen
1977) and of directional theoretical contrastive studies (Kakietek 1980).
Thus, contrary to what Fisiak et al. (1978) claim, the distinction between
"theoretical" and "applied" contrastive studies does not seem to be
correlated with the distinction between directional (horizontal) and adi-
rectional (vertical) contrastive studies.
Having outlined the scope of grammatical contrastive studies cov-
ering comparisons of systems, constructions, and rules, we are now in
a position to suggest a set of operational procedures for conducting
such studies. Ideally, the procedures should be algorithmic, i.e., they
should be possible to follow automatically. The present state of the art
does not lend itself to working out such an algorithm. The theoretical
requirements for such an algorithm will be described in detail in Chap-
ter VIII. At this point, we wish to sketch a more practical procedure,
in which a number of steps have to be taken intuitively, which means
that the contrastive analyst will have to employ his knowledge of the
compared languages in order to make certain decisions. An overall
contrastive theory would have to provide explanation for such choices
in terms of explicit rules.
Beginning with comparisons of systems, we isolate a system in L, and,
having described it, we look for an equivalent system in L2, providing
there is an available suitable description of the system. Suppose we set
about comparing the systems of personal pronouns in English with the
equivalent system in Polish. The English system consists of the following
items:

/ we
you you
heI she I it they
42 Chapter IV

The equivalent Polish system looks as follows:

ja my
ty wy
on/ona/ono onijone

Comparing the two systems, we immediately notice that in some

respects they are identical; namely, in both, distinctions are made between
the first, second, and third person pronouns. These grammatical distinc-
tions are based on the semantic distinctions between speaker, hearer, and
the rest of the world. Furthermore, in both, distinctions in the systems
are made between singular and plural pronouns, although here we also
notice some differences (see below). Finally, we also note that in the third
person singular, distinctions are made between masculine, feminine, and
neuter pronouns. This is where the similarities between the two systems
end. We then proceed to look for differences, which are also quite
conspicuous. They involve the lack of distinctions in English between
singular and plural second person pronoun you in contrast to the dis-
tinction made in Polish between the singular ty and the plural wy. Another
difference consists in the distinction between virile and non-virile gender
in the third person plural in Polish, which contrasts with the lack of the
parallel distinction in English. The Polish grammatical distinction oni vs.
one reflects the semantic distinction: oni refers to groups of people in
which there is at least one male person, while one substitutes for all other
nouns in the plural.
Comparisons of systems are only of limited utility; they do not yield
any information about the ways in which elements of these systems
function in well-formed sentences and about their appropriateness in
particular communicative situations. Without information pertaining to
the semantics and pragmatics of the compared systems, contrastive studies
remain drastically incomplete.
For one thing systematic correspondences, established on the basis of
comparisons of paradigms, are seldom paralleled by syntagmatic corre-
spondences. Thus, although he/she/it systematically correspond to oni
ona/ono, respectively, in actual sentences correspondences are often quite
different due to the fact that the English system of gender is based on
different semantic principles and also due to the existence of various
idiosyncratic properties of particular nouns. The gender of nouns in
Polish is "grammatical", and every noun belongs to one of the three
genders on the basis of morphological properties. In English, the gender
is "natural", which means that it is determined by the sex of the objects
 Three steps in "classical" contrastive studies 43

to which particular English nouns refer. The situation is complicated by

numerous exceptions in both languages. These exceptions constitute a
numerous and diversified class of idiosyncracies (for some analytic details
see Fisiak et al. 1978: 60 — 63). Since in sentences the choice of a particular
pronoun is determined by the gender of the antecedent noun, the apparent
similarity of the systems of pronouns in Polish and English is misleading
and often leads to learners' errors. For example, in many cases the English
pronoun it corresponds in actual texts to the Polish pronouns on 'he' and
ona 'she' rather than to ono 'it' Such is the case in
(17) Zgubilem swoj stary portfel. On byl juz dose zniszczony.
(18) I lost my old wallet. It was already rather worn out.
Polish learners of English may, and actually often do, use 'he' rather
than 'it' in such contexts as (18). Since, with some marginal exceptions,
the selection of the pronoun is completely determined by the grammatical
gender of the antecedent nominal, it is impossible to determine which
Polish pronoun should be used as the equivalent of the corresponding
English pronoun without examining the larger context containing the
antecedent. In many instances, English personal pronouns appearing in
actual sentences have no direct match in Polish, where person is expressed
through verbal inflections alone. This in turn leads to many cases of
redundant use of pronouns in Polish by foreign learners, i.e. to the so-
called "errors of abundance" (see Chapter X).
Moreover, comparisons of systems isolated from contexts cannot, of
course, account for differences in the use of various elements of those
systems in actual communicative situations. We cannot go into all the
details at this point, but it should suffice to note that the circumstances
in which you in English is used are not at all parallel to those in which
ty or wy is used in Polish. Very often sentences with you in English
correspond to Polish third person constructions with Pan 'sir', Pani
'madam', Panstwo 'ladies and gentlemen' The switch to second person
forms requires the fulfillment of rigorously defined sociolinguistic con-
ditions, such as obtain between familiars, friends, peers, etc. Only when
one describes the particulars concerning differences in the contextual
distribution of the compared forms and the extralinguistic factors deter-
mining their usage, can one say that a contrastive analysis is reasonably
complete and has pedagogical relevance. 2
From the methodological point of view, situation (c) described above,
in which an item X in Lj has no equivalent in Lj presents a problem: if
there is no equivalent to compare, is it still possible to compare? The
44 Chapter IV

problem arises most sharply in the comparison of systems. Such is the

case with English articles, which cannot be juxtaposed with any single
system in a number of languages. In order to see what articles can be
compared with, we have to resort to the examination of construction
equivalents to see through what other means, if any, the semantic content
of articles is expressed. Without going into detail, let us assume that the
basic semantic distinction that the English articles express is that between
definiteness and indefiniteness. (In fact the problem is much more com-
plex, but for the sake of illustration of the methodological problem in
contrastive studies, we will take this simplified view of the semantics of
English articles). In many instances, Polish constructions contain no
indication of any distinction between definite and indefinite noun phrases:
(25) Kupilem ksigzkg.
(26) I bought a book.j I bought the book.
Sometimes, however, definiteness and indefiniteness in Polish is ex-
pressed by means of definite and indefinite pronouns appearing in the
function of noun determiners:
(27) Nawet polubilem tego cziowieka.
(28) I even began to like the man.
(29) Jakis czlowiek tu chce sig ζ tob@ widziec.
(30) There is a man here to see you.
(31) Raz zyl pewien krol w odleglej krainie.
(32) Once there lived a king in a far-away land.
Word order, too, may signal definiteness and indefiniteness in Polish.
Szwedek (1981: 53 — 54) convincingly demonstrates that
(33) W pokoju siedzial chlopiec.
'In room was sitting boy'
Chlopiec wyszedl
'Boy went out'
can be rendered in English as
(34) There was a boy sitting in the room.
The boy went out.
Indefinite and definite articles in English express what in Polish is
expressed by means of the word order in (33) in contrast to, for example,
(35) Widzialem na ulicy siedzqcego chlopca. Ρο chwili wstal i podszedl
do mnie chlopiec.
 Three steps in "classical" contrastive studies 45

which is either incoherent or could possibly mean that a different boy is

referred to in the second sentence of (35), in which case (35) could be
rendered in English as

(36) I saw a boy sitting in the street. After a while a (different) boy
got up and walked up to me.
In the same work, Szwedek explores some ways in which sentence
stress is involved in the expression of coreference connected with defi-
niteness.
Thus, we see that in order to find Polish equivalents of English articles
we have to go beyond a contrastive analysis of systems and broaden the
scope of investigations to cover the territory of text grammars and
discourse analysis. Our subsequent discussion will provide more examples
to support the, thus far rather tentative, conclusion that an adequate
contrastive analysis must reach beyond the domain of sentence grammars
towards text grammars and pragmatics.
Chapter V

Contrastive studies at various levels of linguistic

analysis

Most theoretical orientations, notwithstanding some terminological dif-

ferences and disagreements in matters of detail, distinguish the following
components of language, which correspond to various levels of analysis:
phonetic/phonological, morphological, syntactic, lexical, semantic and
pragmatic. The terms mentioned here do not constitute a coherent system
in a definite metalanguage. They can be found in various textbooks on
linguistics which refer to various components of language, and they are
not always compatible with other systems. However, regardless of their
theoretical orientation, these levels must be recognized and considered
by an adequate theory of language, even if some theories seem to favour
a particular component over other components (for example, transfor-
mational-generative grammars focus on syntax). Particular theories differ
with respect to how they subdivide these levels as well as to how, if at
all, they relate the levels to the world at large, i.e., to extralinguistic
reality. There is also a rather fundamental disagreement as to whether
these components are autonomous, as most structuralist and transfor-
mational theories claim, or whether the boundaries between them are
fuzzy, as the proponents of the so-called cognitive grammars claim (e.g.,
Langacker 1987: 7). In some theoretical frameworks, the syntactic com-
ponent of language is subdivided into phrase-structure rules and trans-
formations, while other theories do not recognize the transformational
component at all (Chomsky 1965 vs. Hudson 1976).
In some theories, a distinction is drawn between sentence semantics
and word semantics (e.g., Bartsch — Vennemann 1972), while in other
theories the semantic component is integrated with the syntactic com-
ponent to form the so-called semanto-syntax (cf., for example, generative
semantics as formulated by McCawley 1968; Lakoff 1971; Postal 1970).
In this book, we shall assume there is a distinction between sentence
semantics and word semantics. Consequently, problems connected with
sentence semantics will be dealt with in connection with lexicon. Thus,
48 Chapter V

we are left with three basic components of language (and corresponding

"levels" of analysis): phonological (including morphophonology), gram-
matical (including sentence semantics) and lexical (including word se-
mantics).
It is evident that an analysis of these three components represents a
drastically limited view of language since it does not treat language as a
multifunctional system, and therefore does not offer any insights into
how language actually functions in extralinguistic settings. In other words,
the approach to linguistic analysis through the components distinguished
above provides a framework to analyse but does not provide insights
into how this structure is used to perform its numerous functions. Lin-
guists do not always agree on whether or not they should include the
functional aspects of language in their analyses (cf. different versions of
structuralism discussed in Apresyan 1966, 1971). Many structuralist mod-
els of language, as well as Chomsky's standard theory, treat language in
complete abstraction, divorced from concrete situations in which the
language systems are realized. As a matter of fact, the long-standing
tradition originating with de Saussure and labelled as structuralism (in
most versions) explicitly abstracts language from its concrete realization
and thus adopts de Saussure's sharp distinction between la langue and la
parole, claiming that linguistics should deal only with the former. One
can conduct contrastive analyses of such abstract systems, and this
approach basically characterizes "theoretical" or pure contrastive studies.
Practical validity of such studies, however, is seriously limited. Moreover,
many linguists, opposing the structuralist tradition, claim that such lim-
ited approaches to language overlook the very essence of what language
is all about: its communicative function (cf. Hymes 1972: 292; Searle
1971: 29—40). Linguists of this persuasion include in their investigations
the relations between language as a system of signs and situations in
which it is used. This area of investigation is usually called pragmatics
(cf. Morris 1938). The importance of pragmatics will become particularly
clear in the chapter devoted to pedagogically oriented contrastive studies,
but in order to make the discussion of various aspects of language
reasonably complete, we shall also consider pragmatics in the present
section. Therefore, we add pragmatics to the previously distinguished
areas phonology, grammar, and lexicon.
An adequate view of language must consider the mutual interdepend-
ence among these components and possibly the lack of clear-cut bound-
aries between them. On the other hand, whatever rules operate at these
levels must be focused on and stated separately for each level. Linguists
 Various levels of linguistic analysis 49

have a fairly clear idea of the nature and form of rules operating at the
phonological and syntactic level. It is not equally clear what kind of rules
can be formulated for the lexical level and even less clear at the pragmatic
level. All the same, there is evidence that at all these levels there are
certain facts which can be formulated in terms of rules, while other facts
simply have to be listed. To the extent to which such regular facts do
exist, the respective rules can be described and compared at each level of
linguistic analysis. That rules operate in all four components and that
these rules are to some extent independent can be best seen if we examine
some deviant utterances and try to identify the source of deviance in
terms of rules which have possibly been violated:
(1) My father distrusts psychology/psai'kohd^i/. — phonological de-
viance
(2) Distrust my father psychology? — syntactic deviance
(3) My father distrusts psychologies. lexical deviance
(4) My father fucks up psychology. possible pragmatic deviance
in a specific situation
In (1), the phonological rule which bars initial clusters consisting of a
stop followed by a sibilant has been violated. Such a violation is likely
to occur in the speech of foreigners attempting to speak English, partic-
ularly, if there is no analogous constraint in their native language. In (2),
the transformational rules of question formation have been misapplied.
The entire verb instead of just the tense has been inverted and neither
the agreement transformation nor the Jö-support transformation have
been applied. Whether or not these failures may be due to the interference
of the native language is not a relevant issue at this point, but whatever
rules have been violated in (2) belong to the syntactic component of the
language. In (3), there appears a lexical item which is not part of the
English lexicon. It is not clear whether this case can be handled in terms
of a rule that has been violated. One could surmise, however, that false
analogy is at work here. Due to this sort of overgeneralization psychol-
ogies is formed by analogy with physics, mathematics, etc. In (4), the
deviation is not overtly detectable in the structure of the sentence itself
or in the ill-formedness of its constituents. The sentence in (4) can only
be seen as inappropriate with reference to a specific situation in which it
is uttered, even if phonologically, syntactically and lexically it is perfectly
well formed.
In the sections which follow, we shall present an overview of topics
which have to be dealt with within various components of language in a
reasonably complete contrastive analysis.
50 Chapter V

1. Topics in phonological contrastive studies

Contrastive studies at the phonological level are strictly determined by

the theoretical framework adopted in the description of the compared
languages. The specific linguistic model determines not only what to
compare: segments, features, rules, or representations; it also delimits the
upper and the lower bounds of such studies, i.e., morphophonemic
alternations and phonetics, respectively. The model, furthermore, deter-
mines the degree of refinement of the analysis or the number of details
covered by the analysis. Finally, the model determines the level of ab-
straction at which the contrastive analysis is most usefully performed.
For example, Gussmann, within the generative framework of abstract
phonology, claims that the level of underlying representation
is the only level where features and segments may be legitimately
compared (Gussmann 1978: 158)
He, consequently, maintains that comparisons on the level of phonetic
representation seem to be superfluous since phonological rules will pro-
vide the necessary specifications of phonetic properties of morphemes
and the ways in which phonetic representations differ from phonological
ones. Thus, the aspiration of /1/ in team will be described by an appro-
priate rule in the same way as the fact that /t/ is realized phonetically as
[s] in partial and as [c] in departure·, the fact that the Polish /t/ in trzeba
'it is necessary' is always alveolar and that it is realized phonetically as
[6] in locie 'flight' (dat. sing.) and as [c] in lec§ Ί fly' will be stated
similarly. Therefore, Gussmann concludes:
We believe that there is absolutely no need for separate comparison
of phonetic representations since at best it would merely repeat infor-
mation provided by the comparison of rules We conclude that a
contrastive phonological study of two languages should comprise a
comparison of features and segments at the level of underlying rep-
resentation, morpheme structure conditions and of phonological rules
(Gussmann 1978: 159)
Gussmann's radical view is not shared by a number of outstanding
contrastive linguists. Notably, Fisiak (1975) stresses the relevance of
contrastive studies at the level of phonetic representations, particularly
for pedagogical purposes:
applied pedagogically oriented PCS (phonological contrastive studies)
will contain maximum information about the low phonetic rules (e. g.
 Various levels of linguistic analysis 51

voice assimilation) and phonetic features and segments with very little
(if at all) abstract ("deep") phonological information (Fisiak 1975 a:
344).
Scepticism about psycholinguistic reality of abstract phonology and
about possible pedagogical uses of contrastive studies limited to under-
lying representations and rules is also voiced by Awedyk (1976: 65 ff.
1979: 125 ff., 131).
It seems that the reductionism advocated by Gussmann, though pos-
sibly admissible in some theoretical frameworks of limited psycholin-
guistic and pedagogical relevance, cannot be accepted in any descriptions
of a natural language aspiring to the status of adequate renderings of
psycholinguistic reality. Particularly unwelcome is the reductionism which
breaks language down into rules, their inputs, and/or their outputs; the
devastating consequences of such approaches are best seen in contrast
with cognitive approaches to language. Since language is a complex
organization of symbols at a number of hierarchically arranged levels of
analysis, with each level performing a distinct communicative function,
one must carefully describe each level in the search for what contributes
to the overall functional mechanism of language. It will not do to reduce
the description of one level to properties of another level and to the rules
of combination since such a procedure deprives the object of description
of its most relevant properties.
In the following paragraphs, we shall survey the problems which arise
at various levels of linguistic analysis, particularly in the context of
contrastive studies. In doing so, we shall highlight the shortcomings of
reductionism in order to justify the claim that an adequate contrastive
analysis must be conducted independently at every level of language
organization since it is possible and necessary to focus on what each level
uniquely contributes to the organization of linguistic communication.
In the description of speech sounds, it is essential to focus as much
on their mutual relationships and the functions which they perform in
the phonological organization of language as on their substantial prop-
erties, insofar as these properties are involved in the language symbolism
and hence in communication. Naturally, an adequate contrastive analysis
will embrace all these aspects of speech sounds.
In what follows, we provide an overview of topics which have to be
dealt with within various components of language in a reasonably com-
plete contrastive analysis. It must be noted, however, that due to the
limitations of current linguistic frameworks, as well as due to the limi-
52 Chapter V

tations of contrastive studies themselves, only some of these topics have

been worked out so far; many are still waiting to be dealt with, and some
cannot be handled before more complete linguistic descriptions become
available.
Contrastive studies concerning speech sounds and their arrangements
embrace segmental and suprasegmental phenomena.

1.1. Segmental phenomena

In classical contrastive studies, phonemes and their variants can be

compared across languages either in isolation from other sounds or as
patterns of linear arrangements. According to Lado, three types of ques-
tions can be asked about individual sounds compared in isolation: (1)
Does the native language have a phonetically similar phoneme? (2) Are
the variants of the phonemes similar in both languages? (3) Are the
phonemes and their variants similarly distributed?
The first of these questions concerns a more general problem of
establishing equivalent sounds across languages. Phonetic similarity as
tertium comparationis usually means similarity of articulation in terms of
manner and place. For the purposes of contrastive studies, this kind of
similarity is usually the only relevant thing to consider. But it is also
possible to consider acoustic similarity, which could be described with
reference to the structure of spectrographic images of sounds used in the
compared languages. Similarity of sounds could also be described with
regard to the auditory aspects of sounds, i. e., the ways in which they are
heard by listeners. This latter aspect of phonetics may have relevance in
contrastive studies inasmuch as sounds articulated in a different manner
may have similar acoustic properties and may be perceived as alike by
foreigners (Lamminmäki 1979: 180 ff.). For example, some renderings of
English dental fricatives /Θ/ and /δ/ may be confused with labio-dental
fricatives /f/ and /v/, especially by those listeners in whose native language
the first two are missing. In spite of the different places of articulation
acoustic properties of /Θ/ and /f/ and of /δ/ and /v/, respectively, are
often not sufficiently distinct to ensure proper discrimination of these
sounds (cf. Jassem 1983):
The two fortis fricatives /f/ and /Θ/ are acoustically and perceptually
very similar. This also goes for the two lenis fricatives /v/ and /δ/
(Jassem 1983: 222).
 Various levels of linguistic analy. 53

In the majority of cases, phonological contrastive studies are confined

to the examination of the articulatory features of speech sounds. Con-
sequently, phonetic similarity usually means similarity of articulation,
one of the possible substantial tertia comparationis. Therefore, one can
identify equivalents of the English stops /p, t, k/ in various languages on
the basis of the identity of place and manner of articulation, notwith-
standing the fact that certain variants of these sounds have different
articulatory features in various languages. For example, the English /p,
t, k/ are aspirated in word and syllable initial positions in front of stressed
vowels. Their Polish equivalents are not thus aspirated. In some lan-
guages, such as Sanskrit, there is a set of aspirated stops contrasting with
unaspirated ones. In a language in which aspirated [p] contrasts with
unaspirated [p], the two belong to two different phonemes, even if in
some sense they represent one sound, described as a bilabial stop. The
perception of the two sounds as belonging to one common type seems
to be connected with the importance of particular features in determining
category membership. Place of articulation seems to be the most impor-
tant in the articulation of stops. Typically, the areas involved are: labial,
dental-alveolar, and velar. They are respectively associated with the ar-
ticulation of three basic types of stops [p], [t] and [k], and their voiced,
or perhaps weak, counterparts [b], [d] and [g]. It appears that no other
articulatory property, like aspiration, palatalization, or labialization,
which may accompany the articulation of these consonants, is capable
of changing their status as some type of p, t, k, etc. Whether aspirated,
palatalized or labialized they remain ρ, t, k, etc., respectively, even if
these secondary features bring about phonemic contrasts. To some extent
such intuitive, "naive" and "natural" judgments concerning the catego-
rization of the sounds in question, regardless of the language, may be
supported and reinforced by the alphabets and the spelling conventions
in various languages, whereby various kinds of labial stops are spelled
as " p " and " b " no matter what other features they have. Thus, at least
in the case of stops, the spelling conventions seem to highlight whatever
psychological reality can be attributed to this sort of categorization.
To some extent what was said above is also true of fricatives. Barring
problems with sound discrimination, as in the case of /f/ and /Θ/, etc.,
discussed above, in a number of cases fricatives are identified as /f/, /Θ/,
and /x/, as long as they are respectively articulated in the major articu-
latory areas, viz., labial, dental, and velar. Minor articulatory differences,
accounting for such sounds as the Spanish bilabial [φ], differing from the
English labio-dental [f], or for intradental versions of [Θ], are not signi-
54 Chapter V

ficant enough to provide grounds for recognizing such articulatory var-

iants as belonging to other types of sounds. Consequently, labial fricatives,
whether they are bilabial as in Spanish or labio-dental, as they usually
are in English, will be recognized as equivalent on the basis of the identity
of their primary articulatory features, i. e., the labial area and the friction
accompanying the articulation (cf. Kopczynski 1973).
The lack of phonetic equivalents of some foreign sounds in one's native
language is said to be a source of major difficulties in learning a foreign
pronunciation. Phonological contrastive studies immediately reveal these
differences. For example, in comparing Polish and English speech sounds,
one finds that in English there are no equivalents of such Polish sounds
as /te/ in cienki 'thin', jdzj in dzien 'day', /e/ in sien 'hall', jzj in ziemia
'earth' or /x/ in chata 'cottage' Likewise, in Polish there are no equivalents
of the English /Θ/ as in thin, /δ/ as in though, or /h/ as in ham. Such lists
of differences in the phonemic inventories can be found in any contrastive
studies at the phonetic level (cf. Moulton 1962: 27; Delattre 1965:
155-203; A g a r d - D i Pietro 1965: 2 8 - 3 4 ; Stockwell - Bowen 1965:
1 6 - 1 9 ; Kopczynski 1973; Krzeszowski 1970: 54).
It is important to note that lists of non-equivalent phonemes are
determined by the specific phonological analyses of the data in the
compared languages. For example, there exist two different analyses of
Polish consonants, resulting from two different inventories not only of
consonants but also of vowels. Under one analysis most palatal conso-
nants are recognized as phonemically distinct from non-palatal ones. For
example, Stieber (1966: 117) recognizes the following 35 consonant pho-
nemes in Polish:
ρ b f ν m
p' b ' f v'm'
t d c* 3 s ζ η r I I
c 3 s ζ
c 3 s ζ ή**
k g χ
k'g'
j

* In Stieber's transcription [c] stands for the dental voiceless

affricate rather than for the IPA's palatal voiceless plosive.
** In Stieber's transcription ' marks both palatal and palatalized
sounds.
 Various levels of linguistic analysis 55

The alternative analysis classifies [p1, b', f , v', m1] as positional variants
of the corresponding non-palatal consonants, i.e., [p, b, f, v, m], respectively,
automatically reducing by five the number of non-equivalent phonemes in
English (cf. Jassem 1962: 7 - 1 5 , 1 9 6 6 : 8 7 - 8 8 ; Biedrzycki 1964: 26; Puppel
et al. 1977: 90 ff.). U n d e r this analysis, the five palatal, or rather palatalized,
consonants appear only in f r o n t of the high f r o n t vowel jij and the semivowel
/j/, while the non-palatalized variants appear everywhere else. This analysis
of consonants demands a re-analysis of vowels. U n d e r the first analysis we
deal with a five-vowel system, consisting of /i, e, a, o, u/, where j\j has two
variants (allophones): a higher, tenser and more fronted [i], as in piwo /
p'ivo/ 'beer' and a more retracted, relaxed and lower [i], as i n p y t a c j p i t a t e /
The first variant appears after palatal consonants while the second variant
appears after other consonants.
The second analysis, which recognizes some palatal consonants as
positional variants of the non-palatal ones, requires the recognition of [i]
and [i] as phonemically distinct to provide phonemic contexts for the
mutually exclusive variants of the consonants. Thus, the palatalized
variants appear in f r o n t of j'xj and the semivowel /j/, while the non-
palatalized variants appear elsewhere. Consequently, under the two anal-
yses we will have the following different transcriptions of words in which
the relevant sounds appear:
Analysis I
bicjb 'itej vs. bycjbitel
'beat' 'be'
Analysis II
bicjbite! vs. byc\bite\
U n d e r the first analysis, the English vowel phoneme /i/ will have no
phonemic equivalent in Polish, since the Polish vowel phonemes constitute
the set:
u
e ο
a
in which there is only one high non-back vowel.
U n d e r the second analysis there are two high non-back vowels in
Polish, i. e., j\j and /i/:
i u
e a ο
56 Chapter V

Following this analysis, it is possible to match the Polish /i/ of bic

/bite/ with the English /i:/ of beat as equivalent, and similarly the Polish
j\j of bycjbitQj with the English /i/ of big.
A contrastive study based on the first analysis will reveal a larger
number of Polish consonant phonemes which have no phonemic equiv-
alents in English than a contrastive study based on the second analysis,
inasmuch as the first analysis results in the recognition of five palatal
consonants as distinct phonemes, which the second analysis interprets as
positional variants (allophones). On the other hand, a Polish-English
contrastive study based on the second analysis will reveal a smaller
number of non-equivalent Polish vowel phonemes in comparison with
their English counterparts, since both the English f\:j and /[/ will now
have their phonemic equivalents in Polish. Thus, under the second ana-
lysis, the Polish and English phonemic systems will appear more similar
than under the second analysis, since the first analysis reveals more
differences in the inventories of both vowels and consonants.
Alternative analyses are also possible for English. For example, some
phonologists would claim that the vowel sound in beat is actually a
diphthong consisting of a variant of j\j followed by the semivowel j]j
(e.g., Jassem 1954: 74, 1983: 184). Consistently, the vowel sound in food
is phonemically interpreted as /uw/. This approach allows us to reduce
the number of English vowel phonemes to the following seven

u
e a ο
ε a

Obviously, if this sort of analysis of the English data is adopted as a

framework for contrastive studies, the number of non-equivalent Polish
phonemes will be smaller than if an alternative analysis, recognizing
twelve simple vowel phonemes in English is adopted (see, e.g., Gimson
1962: 86 ff.).
The fact that such divergent results of competing analyses are possible
is the sign of weakness in classical phonological theories. The deficiency
consists in the lack of theoretically motivated and empirically verifiable
criteria permitting one to decide which of the descriptively adequate
models is to be preferred over the alternative descriptions, assuming that
the models in question are descriptively adequate. (For a discussion of
weaknesses of classical phonology, see Postal 1968: 3 — 18). In other
words, classical phonological theories do not achieve the level of explan-
 Various levels of linguistic analysis 57

atory adequacy. Hence, one or another mode of description can be given

priority only on either intuitive or purely arbitrary grounds.
The foregoing discussion leads to the conclusion that Lado's first
question pertaining to phonological contrastive study, i.e. "Does the
native language have a phonetically similar phoneme?", cannot, in many
cases, be answered in unequivocal terms due to the lack of clearly stated
criteria determining what constitutes phonetic similarity and due to the
existence of alternative phonemic analyses resulting in alternative inven-
tories of phonemes, which in turn leads to alternative inventories of
differences and similarities across languages.
The failure to answer the first question satisfactorily is naturally
connected with similar difficulties in supplying answers to the next two
questions. Failure to identify equivalent phonemes across languages au-
tomatically rules out any possibility of comparing equivalent allophones
of those phonemes as well as the distribution of allophones.
A phonological contrastive study seems to be better off if the concept
of phoneme is bypassed, at least in those cases when exact matching of
equivalents is impossible. In such cases, the contrastive study involves
comparisons of articulatory (and possibly acoustic) properties of sounds
regardless of their place in the overall system. This is presumably what
Lado himself anticipated when he wrote:

In comparing the sound system of a foreign language and native

language I find it good safe practice to take up each phoneme sepa-
rately regardless of any general patterns of difference I may have
observed (Lado 1957: 13).

We shall take up the problem of equivalence at the phonetic level and

of the ensuing problems in the chapter devoted to structural contrastive
studies. At this point, let us only observe that contrastive studies limited
to a comparison of equivalent phonemes across languages are inadequate
as they often obscure conspicuous differences in the realization of these
phonemes in particular phonetic contexts. This problem will also be taken
up in due course.
Segmental phonological studies are not confined to paradigmatic list-
ings of sounds and their equivalents in the compared languages. A
complete account also includes comparisons of syntagmatic relations
between sounds, i.e. various combinations of sounds called diphthongs
and consonantal clusters. Contrastive studies must list similarities and
differences between specific constraints operating in the compared lan-
58 Chapter V

guages with respect to admissible combinations of sounds as they con-

stitute syllables, morphemes, and words.
Let us consider clusters as a characteristic phenomenon in syntagmatic
relations between speech sounds. Insofar as the number of sound types
in specific languages is finite and, insofar as there are limitations on the
number of consonants in a cluster, the number of clusters is finite in a
given language. Hence, one may find it superfluous to look for any rules
governing the structure of clusters since all clusters can be listed. Yet, it
is tempting to formulate generalizations (and consequently rules) gener-
ating possible clusters and ruling out impossible ones. (See, for example,
Gussmann 1980: 54.)
Syntagmatic phonological contrastive studies will also include com-
parisons of syllabic structures. For example, Awedyk (1974: 94) analyzes
structures of syllables in Polish and English and draws the following
conclusions from his contrastive studies:
1. Concerning the structures of nuclei. In English the nucleus of the
syllable may be either simple or complex. In Polish it is always simple.
The English nucleus may consist of the resonants /r, 1, n, m/. In Polish
the nucleus must consist of a vocalic phoneme.
2. Concerning the structure of the onset. English onsets may consist of
from zero to three positions. Polish onsets may contain from zero to
four positions. In English only /s/ can occupy position 0 3 (the first
one in a sequence of three consonants). Different constraints affect
other positions.
3. Concerning the structure of the coda. In both English and Polish codas
may consist of from zero to four positions. In English /s, t/ occur only
after the morpheme boundary.
4. Concerning the structure of the interlude. In English the interlude may
consist of from zero to four positions. In Polish the interlude consists
of from zero to five positions.
It is worth noting that Awedyk's results are influenced by the kind of
descriptions that he adopts for the two languages. Whereas for Polish he
adopts Jassem's analysis referred to above (Jassem 1954), for English he
adopts Gimson's description recognizing /i:/ and /u:/ as monophthongs
rather than as diphthongs /ij/ and /uw/, respectively, as in Jassem (1954,
1983). If Awedyk had adopted this latter analysis, he would have come
out with different numerical results concerning the number of complex
and simple nuclei in English (cf. Awedyk 1974: 91). Likewise, his claim
 Various levels of linguistic analysis 59

that "in Polish it [the nucleus] is always simple" (Awedyk 1974: 94) would
have proved to be false if he had adopted Biedrzycki's view that the
Polish nasal vowels [?] and [q] are, in phonological terms, nasal diph-
thongs /eg/ and /οη/, respectively (Biedrzycki 1964). As diphthongs, they
constitute complex nuclei in such Polish words as k§s/kegs/, wpz/v053/
and konski/kojiskij.
Awedyk's analysis is another example of the extent to which the results
of contrastive studies are determined by the descriptive frameworks
adopted in the analysis of the materials in the languages that undergo
comparison. This sort of dependence also has a bearing on the results of
contrastive studies vis-ä-vis pedagogical applications. Both these aspects
will be taken up in the appropriate parts of the book, i. e. in Chapters
VI and XII.

1.2. Suprasegmental phenomena

Phonological contrastive studies must also embrace suprasegmental phe-

nomena, stress and intonation. There exist numerous descriptions of
English stress, rhythm and intonation employing a great variety of no-
tations, which use numerals, strokes, accents, dots, and other graphic
devices. The situation is further complicated by the lack of parallel
descriptions for other languages. For these reasons any contrastive studies
of these phenomena are still very much a pioneering endeavour. (But see
the studies by Ozga 1974; Mackiewicz-Krassowska 1974; Woloszyk 1974;
Marek 1974; Mieszek 1974; Varga 1975; Stockwell - Bowen 1965;
A g a r d - D i Pietro 1971).
The lack of rigorous criteria in the description of suprasegmentals is
even more evident than it is in the case of segmental phonemes. Therefore,
the consequences in the realm of contrastive studies are even more
conspicuous: there is a lack of extensive and reliable contrastive studies
of suprasegmental phenomena. The available descriptions and compari-
sons suffer from lack of uniformity, rigour, and reliability. The problems
are compounded by the multiplicity of notational conventions, which still
only conceal more deeply grounded differences in the description of
linguistic data. For example, Stockwell — Bowen (1965: 21), following
Trager —Smith (1957: 37), distinguish three levels of English stress, which
they call "strong", "medial", and "weak" Gleason (1955: 4 0 - 4 2 ) , Fran-
cis (1954: 152), and many others distinguish four levels of stress in English.
60 Chapter V

Following the latter group, Fisiak (1978) speaks of four levels of stress
in English and distinguishes also four levels of stress in Polish:

Both in English and Polish at least four degrees [of stress] can be
recognized (Fisiak et al. 1978: 242).

Lado distinguishes as many as five levels of stress in English (Lado

1957: 28). Opinions concerning other languages are less diversified, per-
haps because less attention has been paid to matters of stress in terms of
structural phonemic analysis, widely practised in the United States. Still,
in one of very few works devoted to suprasegmentale in Polish only two
levels of stress are mentioned, viz., "main" and "subsidiary" (Dtuska
1976: 2 0 - 2 8 ) . Thus, Dluska tacitly contradicts both Fisiak et al. (1978)
and Ozga (1974: 133). In the latter paper, reference is made to at least
three levels of stress, though the author does not explicitly mention any
definite number.
In this situation no genuine contrastive studies of stress can be at-
tempted. Whatever has been done so far consists in juxtaposing inde-
pendent descriptions of the data in two languages, without the slightest
attempt to find any tertia comparationis nor to formulate any generali-
zations about differences and similarities of the compared phenomena.
Fisiak's two-page contrastive study of stress in English and Polish ends
with the significant sentence:

Because the stress in English may fall in so many different places in

the word, naturally the rules of stress assignment have to be more
complicated there than in Polish (Fisiak et al. 1978: 243).

Future contrastive studies in this domain will have to discover both

the rules in the two languages and the degree of their "complication"
For the time being, it seems, it is best to refrain from any further
comments on the subject.
The state of the art in the domain of intonation is no better, which is
not surprising, since all suprasegmental phenomena, i.e., intonation, and
pauses, are closely interrelated (Ozga 1974: 127). Again there is no
agreement as to the number of basic patterns. Whereas Mackiewicz-
Krassowska (1974: 138 — 139) assumes six basic tunes in both Polish and
English, Fisiak et al. (1978: 244) speak of twelve patterns in the two
languages. Fisiak's analysis is confined to listing examples of the twelve
patterns in the two languages with the indication of their meanings in
terms of such qualifying epithets as "cool", "reserved", "dispassionate",
 Various levels of linguistic analysis 61

"calm" "serious", "querulous" etc. The review ends with a rather ob-
vious conclusion that
there is no one-to-one correspondence between an intonation pattern
and an emotional attitude either in English or in Polish (Fisiak 1978:
250).
He immediately adds that
Intonation is assigned to particular types of syntactic constructions by
well defined rules which use both semantic and syntactic information
(Fisiak 1978: 250).
Unfortunately, the analysis ends there and not a single rule is presented.
More detailed analyses of intonation in Polish and English are pre-
sented in a series of papers devoted to declarative sentences, interrogative
sentences, imperative sentences, and compound sentences, respectively,
by Mackiewicz-Krassowska (1974), Woloszyk (1974), Marek (1974), and
Mieszek (1974). Although all these papers contain appeals for more
research, the authors are able to draw some specific and definite conclu-
sions from their contrastive studies.
Thus Mackiewicz-Krassowska (1974) concludes her contrastive study
of declarative sentences with the following statement:
As this brief analysis of Polish and English declarative sentence into-
nation shows, the falling intonation is used in Polish for the majority
of cases. In Polish more words are uttered on a level tune which evokes
the effect of monotony to some non-Polish speakers. Moreover, an
average Pole's pitch of voice does not go as high as that of an
Englishman is. This also accounts for the relative paucity of those
characteristic ups and downs heard in an English utterance. As we
have seen from the above examples, the pitch patterns of English and
Polish statements are on the whole different. The only similarity in
the intonational pattern can be noticed in straightforward statements
and the patterns used for enumeration. Even then, the similarity is
only apparent, as the Polish tunes, the high fall especially, are lower
than those of English (Mackiewicz-Krassowska 1974: 143 — 144).

With respect to interrogative sentences Woloszyk has discovered that

the intonation of questions in English and Polish has some certain
common features. The accentual function of intonation may be quoted
as the example, since there are certain common rules, one of which may
be the change of pitch on the stressed syllable (Woloszyk 1974: 157).
62 Chapter V

This is again where firm conclusions end. Everything else appears to

be rather muddy:

English intonation, however, in its non-accentual function is to a

certain degree [emphasis my own] fixed. Polish intonation depends
largely [emphasis my own] on the speaker's emotional attitude and
sometimes on his origin (Woloszyk 1974: 157).

The comparative problem which arises here, but which Woloszyk does
not tackle, is this: since in both English and Polish intonation is to some
extent fixed (and consequently to some extent unstable or varied), what
exactly is the extent of the variation in both languages and what exactly
determines the degree of stability of intonation patterns in both lan-
guages? Further contrastive studies must provide answers to these ques-
tions. Since there are still many unresolved problems in the description
of intonation in both languages, the answers may not be readily forth-
coming. For example, opinions still vary as to whether English has a
special tune characterizing questions, as is claimed in Gimson (1962:
256 ff.). Yet many questions have a falling intonation, in spite of theo-
retical predictions but in keeping with emotional attitudes of the speakers.
In Polish, too,

all types of questions which usually end in a rise may end in a fall
when there is a great emotional load, and vice versa those usually
ending in a fall may end in a rise when the speaker wants to express
his feelings fully (Woloszyk 1974: 158).

Marek's and Mieszek's papers end with similar notes of uncertainty.

Marek states that:

Intonation is not an independent system which can modify or change

the meaning of a sentence with which it occurs. It reflects the under-
lying, intended [emphasis my own] meaning and is subordinate to it
There is no separate 'normal' intonation pattern for imperative sen-
tences. Each command will have that intonation which can bring about
the desired meaning (Marek 1974: 178).

Mieszek ends her contrastive study of compound sentences in English

and Polish with the following words:

It is evident, however, that various kinds of intonation patterns expose

different feelings and connotations that add considerably to the literal
 Various levels of linguistic analysis 63

contents of the sentence. Therefore the expressive function of intona-

tion is relevant in any case (Mieszek 1974: 187).1

It seems that the research in this area can advance substantially only
if pragmatic and cognitive functions of intonation are incorporated into
the description. Specifically, one needs to consider the basically meta-
phorical and iconic aspect inherent in the orientation U P and D O W N as
involved in the explication of such concepts as U N K N O W N ( = U P (in
the air)) vs. K N O W N ( = D O W N (within my reach)), U N C E R T A I N
( = UP) vs. CERTAIN ( = DOWN), G O O D ( = UP) vs. BAD
( = DOWN), etc. Systematic research into the iconic relationships be-
tween these metaphorical concepts and the pitch level in sentences may
provide answers to some of the hitherto mysterious problems and, con-
sequently, cause a breakthrough in both descriptive and contrastive lin-
guistics. Some preliminary suggestions along these lines and concerning
possible relationships between orientational metaphors and syntax have
already been made by Lakoff—Johnson (1980: 126 ff.).

2. Topics in syntactic contrastive studies

Regardless of the adopted model, syntactic contrastive studies must

embrace basic sentence patterns, major sentence constituents, i. e., noun
phrase and verb phrase, various functions of noun phrases as subjects,
objects, complements, and modifiers and of verb phrases as predicates.
The analysis of sentence constituents must embrace both their internal
and external structure. The internal structure of a constituent embraces
the hierarchical relations which occur between subconstituents of a par-
ticular constituent. For example, in English, but not in Polish, the con-
stituent called noun phrase must contain an obligatory subconstituent
called determiner, which in turn can be analysed into subsubconstituents,
etc. In Polish the presence of a determiner in the noun phrase is optional,
and, insofar as this is so, the internal structure of noun phrases in English
and Polish can be said to be different. Another difference relating to the
internal structure of constituents may concern word order. The order of
subconstituents of both noun phrases and verb phrases in English is
comparatively more rigid than in Polish. Thus, whereas in English only
my first son is grammatical, while *first my son is not, in Polish both moj
pierwszy syn and pierwszy moj syn are grammatical. Yet another difference
64 Chapter V

concerning internal structure of constituents may involve co-occurrence

restrictions on subconstituents. For example, within the English deter-
miner, possessives and demonstratives, in most dialects of English rec-
ognized as standard, are mutually exclusive, which means that they cannot
both occur as determiners in the same noun phrase. N o such restriction
exists in Polish. Therefore, in English * these his children is ill-formed,
while in Polish te jego dzieci is well-formed.
The external structure of a constituent refers to the hierarchical rela-
tionship that it enters with other constituents to combine into more
complex linguistic forms. For example, nominal subordinate clauses can
appear as constituents of main clauses, namely as subjects, direct, indirect
and prepositional objects, subject and object complements, modifiers and
appositives. Such nominal clauses can also be conjoined with other
nominal clauses to form coordinated nominal clauses. What is relevant
from the contrastive point of view is the fact that there may be different
restrictions on the occurrence of nominal clauses in some of these func-
tions across languages. For example, in English, unlike in Polish, certain
types of nominal subordinate clauses can function as objects of prepo-
sitions. Compare:

She was right in what she said.

*Ona miala racjg w czym powiedziala.

They couldn't agree on whether they should go.

*Oni nie mogli zgodzic si§ na czy powinni tam isc.

In such cases, Polish (and many other languages) requires a supporting

pronoun or a noun phrase into which the nominal clause is embedded
as a relative clause:

Ona miala racj§ w tym, co powiedziala.

Oni nie mogli sie zgodzic na to, czy powinni tam isc.

The distinction between internal and external structures suggests one

possible strategy in looking for syntactic contrasts across languages.
Fisiak et al. (1978) distinguish the following types of differences in
comparing equivalent sentences (constructions) across languages: (a)
structural, (b) categorial, and (c) functional.
Structural differences occur when in L; there exists a syntactic structure
which has no congruent counterpart in Lj. In transformational terms (see
Chapter VI), these differences are due to differences in the application of
rules generating the equivalent structures. A structural difference occurs
 Various levels of linguistic analysis 65

whenever a particular rule of grammar is not shared by the grammars of

Lj and Lj. For instance, in Polish (and also in Spanish and other lan-
guages), but not in English (and German), the subject need not be
specified as a personal pronoun but may be expressed only by means of
the verb inflection:

On napisai list. Napisal list.

He wrote a letter. * Wrote a letter.
Ona napisala list. Napisala list.
She wrote a letter. * Wrote a letter.

Oni napisali list. Napisali list.

They wrote a letter. * Wrote a letter.
In addition Polish has a number of overtly subjectless constructions,
which as a rule correspond to English constructions with overt subjects:

Padalo wczoraj.
It rained yesterday. But not: * Rained yesterday.
Bylo gorgco.
It was hot. But not: *Was hot.
Slyszy sig ο tym duzo.
One hears about it a lot. But not: * Hears about it a lot.
The statement of structural differences of this sort must be enhanced
by the statement of categorial contrasts, which occur whenever in equiv-
alent sentences corresponding elements belong to different grammatical
categories. Such is the case in the following pair of equivalent sentences:
Czujp sip zle. (Adverb)
I feel bad. (Adjective)

Fisiak et al. (1978: 21 ff.) also note the following categorial contrasts:

Main verb vs. modal verb

Mog? tam pojechac. I can go there.
Noun phrase vs. prepositional phrase
wierzcholek gory the peak of the mountain
I often play harp. Czgsto gram na harfie.
Categorical contrasts are inherently connected with functional con-
trasts, which occur whenever equivalent categories perform different
66 Chapter V

functions in equivalent sentences. 2 Noun phrases and verb phrases are

the two major sentence constituents in a large number of languages. Verb
phrases typically function only as predicates, but noun phrases can
perform a variety of functions. Therefore, a large number of functional
contrasts can be expected if various functions of noun phrases are com-
pared across languages. Considering Polish and English and restricting
the illustrations to subjects and objects, we find the following contrasts:
Subject in Polish vs. direct object in English
Markowi (dat.) zgingiy pien- Mark lost his money.
i^dze (nom.).
Ksi^zka (nom.) jest mi (dat.) I need the book.
potrzebna.
Indirect object in Polish vs. subject in English
Chce sig jej spac. She is sleepy.
Brakuje mi czasu. I lack time.
Indirect object in Polish vs. direct object in English
On jej ufa. He trusts her.
Jan im wierzy. John believes them.
Indirect object in Polish vs. possessive attribute in English
Markowi popsul si§ samo- Mark's car broke down.
chod.
'To Mark broke down car'
direct, usually cognate object
Adverbial phrase in Polish vs.
in English
Umarl lekkg. smiercig.. He died an easy death.
'died-he light death-with'
The three types of syntactic contrasts: structural, categorical, and
functional, varied as they are, can be stated regardless of the theoretical
framework adopted in the description of the compared languages, as long
as the description is conceptually consistent in both languages. In this
respect, contrastive analysts rely on their intuition, which prompts deci-
sions not only about which sentences and constructions are equivalent in
the compared languages but also which elements (constituents) of the
compared constructions are structurally, categorically, and functionally
equivalent. Necessarily, since intuition is the guiding principle, these
decisions are often arbitrary and unmotivated on other, more objective
 Various levels of linguistic analysis 67

grounds. For instance, among the examples discussed by Fisiak et al.

(1978: 27 ff.) and quoted above, there are some which could not be
recognized as semanto-syntactically equivalent if more rigorous criteria
were applied and, consequently, if a more constrained view of semanto-
syntactic equivalence were to be employed. For example,

(1) John lost his keys.

would not be recognized as equivalent to
(2) Klucze zgingly Janowi.
since the former is not the closest approximation to an aceptable
word-for-word translation of the latter. In fact
(3) John's keys vanished.
is a more likely candidate, despite a subtle difference in the meaning: in
the Polish sentence the subject seems to be more directly affected by the
disappearance of the keys than the subject in the English sentence (3).
But one also notes a difference in the meaning between (1) and (2): in
the former the subject seems to be more directly responsible for the
disappearance of the keys than the subject in (2), a more passive exper-
iencer of the event. Considering these differences in the meaning, which
occur in both cases, the fact that no other translations are readily
available, and the fact that (3) is a closer approximation to (2) than is
(1) in terms of word-for-word rendering, (3) rather than (1) would have
to be recognized as semanto-syntactically equivalent with (2). By the
same reasoning, (4) would be the Polish equivalent of (1):

(4) Jan zgubil klucze.

It is worth noting that (4) does not differ semantically from (1) in the
same way in which (2) does, which provides another argument for why
only (4) rather than (2) can be recognized as the semanto-syntactic
equivalent of (1).
Decisions concerning equivalence must be based on rigorous criteria,
if equivalence is to be defined in terms of an explicit device, enumerating
semanto-syntactically equivalent sentences. A device of this sort is dis-
cussed in Chapter VIII. For pedagogical and other practical purposes, it
is perfectly possible to compare translationally or perhaps statistically
equivalent sentences, such as (1) and (2). The fact that (1) is often found
as a translation of (2) cannot fail to have pedagogical relevance. All the
same, the translational equivalence of (1) and (2) cannot serve as a
foundation for drawing syntactic generalizations about functional con-
trasts across languages since translational equivalence does not guarantee
68 Chapter V

semanto-syntactic equivalence (see Chapter III), hence it cannot provide

reliable data for contrastive generalizations of a semanto-syntactic nature.
Syntactic contrastive studies occasion certain theoretical and meth-
odological problems. The crucial problem consists in the question whether
syntactic contrastive studies can be projective, i.e., whether they can be
generalized in terms of rules, or whether they must remain reduced to
listing odd and unpredictable contrasts. The problem is connected with
linguistic universale insofar as syntactic contrasts across languages begin
in exactly that place where syntactic universale end. Thus, the scope of
syntactic contrastive studies is negatively delimited by the scope of syn-
tactic universals. This relationship is aptly perceived and discussed by
Preston (1975), who argues within Chomsky's theoretical framework that
universal grammar will eventually leave only idiosyncratic odds and
ends and irregularity behind in particular grammars better gram-
mars, in the generative sense, provide less and less detail for contrastive
analysis (Preston 1975: 69).
According to Preston, "contrastive analysis does not constitute a part
of theoretical linguistics" (Preston 1976: 69) and has no contribution to
make to a general theory of language:

Contrastive linguistics is not, then, a handmaiden to theoretical gram-

mar, providing insights into universal characteristics of human lan-
guage (Preston 1975: 67).
These words are drastically at variance with Zabrocki's view on the
subject. Though Zabrocki also believes that contrastive study "does not
aim at creating any original explanatory theory" (Zabrocki 1976: 108),
he still claims that it
can and should contribute to a) linguistic typology, b) general linguistic
theory, c) grammatical descriptions of particular languages (Zabrocki
1976: 108).
The two conflicting claims are typical of the state of uncertainty
beclouding the area and its theoretical status vis-ä-vis other domains of
linguistics. It is also worthy of note that controversies of this sort should
arise in connection with syntactic contrastive studies, since syntax is the
focus of interest in transformational-generative accounts of language,
which promote and attempt to answer all kinds of questions about
universal grammar. In subsequent chapters we shall attempt to reconcile
these conflicting views on the nature of syntactic contrastive studies.
 Various levels of linguistic analysis 69

3. Topics in lexical semantic contrastive studies3

Earlier in this chapter, we drew the distinction between sentence semantics

and word semantics. The former deals with that part of the meaning of
a sentence which remains constant regardless of the specific lexical items
which appear in the sentence. This aspect of the meaning of sentences
serves as tertium comparationis for syntactic contrastive studies. Word
semantic deals with the specific meanings which particular lexical items
contribute to the overall meaning of sentences. The categorial and sub-
categorial status of sentence constituents, morphological markers (inflec-
tions), word order, function words, and intonation, all contribute to
sentence semantics in that they are more or less systematically and
predictably correlated with such phenomena as declarative, interrogative,
and imperative forms of sentences, tense, sentence connection, embedding,
topic/comment relations, focus formation, and presuppositions.
For example, a prototypical statement has the form of a declarative
sentence, a prototypical question has the form of an interrogative sen-
tence, a prototypical expression of the past time assumes the form of the
past tense, etc. All these phenomena, insofar as they deal with form-
meaning relations, fall within the domain of sentence semantics.
Word semantics deals with idiosyncratic properties of articular lexical
items appearing in sentences and with various systematic relationships
between various categories and subcategories of words insofar as these
words constitute paradigmatic sets associated through some common
properties (such as kinship terms, colour terms), which have no bearing
on the syntactic well-formedness of sentences. The latter properties of
words are also considered as part of dictionary information, usually in
the form of information concerning syntactic contexts in which particular
lexical items can appear. This sort of information does not, strictly
speaking, fall within the domain of word semantics.
Let us note that the word dictionary has at least two senses: (a)
practical, in which it denotes "a reference book on the living-room or
library shelf' (Leech 1974: 202); (b) theoretical, in which it denotes "a
hypothesis about, or, 'the inbuilt dictionary' which every one of us carries
around as part of his mental equipment as a speaker of a language"
(Leech 1974: 203). Let us observe that this second sense, in fact, embraces
two senses, viz., (1) "a hypothesis" itself, which must have the form of a
set of statements in some metalanguage, and (2) the mental objects to
which the "hypothesis" refers. In the present work, whenever we use the
70 Chapter V

word dictionary, we are referring to the first two of the three senses, i.e.,
dictionary as a reference book and dictionary as a set of linguistic
statements, i. e. "a hypothesis"
The branch of linguistics which deals with words, called lexicology,
offers a large number of definitions of "word", none of which seems to
be obviously better than the next. In view of this, for the purposes of the
present work, we shall assume the following: just as in the case of the
sentence, which cannot be given a satisfactory one-sentence definition
but rather must be defined in terms of a device called grammar, which
generates sentences in a particular language, "word" cannot be defined
without constructing a dictionary of that language. Thus, while sentences
are objects generated by the grammar of a language, words are objects
which are listed in the dictionary to the left of each lexical entry. All the
same, before compiling a dictionary, the investigator must delimit the
scope of his data in such a way as to conform to a native speaker's
intuitions (unfortunately very misleading!) about what constitutes a word
in a given language, i.e., about what qualifies as a potential lexical entry
in the dictionary that he is compiling. Therefore, some tentative, working
definition is necessary to ensure a certain degree of consistency in com-
piling lexical entries. It follows that the notion "word" can be described
from two points of view. From the point of view of a linguistic theory
the word is a linguistic unit which is listed in the dictionary of a particular
language as the first element of each lexical entry. From the diagnostic
point of view, while providing criteria for identifying words in texts (both
written and spoken), it is possible to describe the word in a variety of
ways. In the present work we "define" the word after Arnold (1973: 30)
and Lyons (1968: 203) in a definition which is a synthesis of the definitions
by the two authors:

a word is the smallest significant unit of a given language, which is

internally stable (in terms of the order of component morphemes), but
potentially mobile (permutable with other words in the same sentence).
This definition makes it possible to distinguish between the word and
the phrase (not the smallest significant unit), the word and the morpheme
(not positionally mobile within a word), as well as the word and the
phoneme (not significant). In this way, the definition isolates lexicology
from syntax (phraseology), morphology, and phonology. It is needless to
say that all these areas are mutually interrelated, and that in actual
analytic practice it is often difficult to draw clear-cut boundaries. Espe-
 Various levels of linguistic analysis 71

daily, the boundary between lexicology and syntax is fuzzy, which creates
certain problems to which we shall presently return.
The crude definition of lexicology given above defines it as the study
of words. Considering the definition of words as significant units of
language, it is inevitable that what lexicology must be concerned with is,
among other things, meaning of words. Now, the study of meanings of
words is impossible without examining linguistic and extralinguistic con-
texts in which words appear and in which they assume various senses
within the limits characterizing a particular word. Therefore, any pro-
ductive approach to meanings of words must be through the contexts in
which words appear.
The definition of the word given above allows it to include among
words not only compounds such as blackboard or typewriter, in which
constituent morphemes cannot be permuted within a given sentence, but
also set phrases (fixed expressions) of various degrees of conventionality,
ranging from such non-motivated phraseological fusions as red tape
(bureaucracy) or kick the bucket (die), which are equivalents of words,
to highly motivated phraseological collocations such as alarm clock and
night-school, all of which express certain integrated notions. Some of
these expressions happen to have one-word equivalents in another lan-
guage (cf. Polish biurokracja, budzik, etc.). All such combinations of
linguistic units, which also function as words (red, tape, night, school,
alarm, clock, kick, the bucket), and which express fixed integrated notions
naturally fall within the scope of lexicology, and it would be useless and
impractical to pretend that they do not (see also Nowakowski 1977: 38).
However, if one accepts the view that lexicology has to deal with
compounds and fixed expressions of the type exemplified above, one
faces a formidable task of delimiting the upper bound of lexicology,
separating it from syntax. The basic problem is to what extent constraints
on collocations of particular lexical items in syntactic constructions are
subject to listing in a dictionary and to what extent they can be stated in
terms of rules. This in turn is connected with a more general problem of
what may be called "precision" of grammars. Early transformational
generative grammars, written in the framework of Chomsky's Syntactic
structures (1957), were extremely crude in that they imposed no con-
straints on the co-occurrence of various content words (nouns, verbs,
adjectives, adverbs) in syntactically well-formed preterminal strings. These
models generated both, semantically acceptable sentences, such as John
admires sincerity, and semantically unacceptable Sincerity admires John.
However, Chomsky realized that an adequate grammar should be en-
72 Chapter V

dowed with means for eliminating such semantic anomalies. One way to
do this was to subcategorize verbs and nouns in such a way as to allow
only some verbs to occupy positions after abstract nouns functioning as
subjects of these verbs, and only some verbs in positions before human
nouns, etc. This led to the formulation of the theory of selection restric-
tions expounded in Chomsky (1965). The elimination of all possible
semantic anomalies through the operation of selection restrictions re-
quires the introduction of an appallingly large number of theoretical
concepts called "semantic markers" This fact is noted by Bolinger, who
writes:

If we are to account for the fluent speaker's ability to recognize an

anomaly as well as an ambiguity through the markers at his
command, then the number is indeed legion (Bolinger 1965: 564).
All the same, there is no doubt that contextual features, strictly
subcategorizing verbs in terms of syntactic categories with which verbs
can co-occur in sentences, are syntactically relevant. Strict subcategori-
zation rules eliminate such obvious syntactic anomalies as *John found
sad, *John elapsed that Bill would come, or *John compelled (Chomsky
1965: 148 — 149). However, neither contextual strict subcategorization
features nor markers (in whatever number) are helpful in accounting for
ill-formedness of a large number of collocations such as * carry a grudge
as opposed to bear a grudge or *strong rain as opposed to heavy rain, or
*take hatred as opposed to take a liking. On the other hand, McCawley
claims that markers have no syntactic relevance at all, since
the matter of selectional restrictions should be totally separate from
the base component and the base component thus be a device which
generates a class of deep structures without regard to whether the
items in them violate any selectional restrictions (McCawley 1968:
135).
This claim leaves collocations totally outside the scope of any gram-
mar: lexical anomalies, such as *take hatred or *carry a grudge are
syntactically well-formed, and the theory does not provide any framework
for considering them semantically ill-formed, either, unless lists of col-
locations are systematically incorporated into the theory to be drawn
from at the same level at which other lexical items are inserted. We shall
take up this problem in a little while. Let us state, at this point, that
notwithstanding theoretical problems with finding a place for collocations
in an overall account of language, they do fall within the scope of
 Various levels of linguistic analysis 73

lexicology, constituting its upper-bound and being an area which shades

into syntax in a non-discrete (fuzzy) way.
Thus lexicology is concerned with formal and semantic properties of
the following phenomena: one-morpheme words, constituting the lower
bound of lexicology (e.g., table, man, dog, radio, etc.), complex words
(e.g., engulfj writer, disagreeable, etc.), complex-compound words (e.g.,
blackboard, armchair), compound-complex words (e. g., typewriter, radio-
announcer), phraseological fusions or idioms, i.e., poorly motivated or
apparently quite unmotivated combinations of words which are seman-
tically integrated (e. g., red tape, kick the bucket), phraseological unities,
i.e. partially motivated combinations of words (e.g., show one's teeth,
to wash one's dirty linen in public), and phraseological collocations, i.e.
combinations of words characterized by lexical valency but highly mo-
tivated from the semantic point of view, constituting the upper bound of
lexicology (e. g., bear a grudge, bear malice, take a liking, make a decision,
take a break, have breakfast).
Both formal and semantic properties of these units can be the object
of contrastive studies. Let us look at some of these properties from the
contrastive perspective under the following headings: (1) simple words;
(2) word formation; (3) phraseological fusions, unities, and collocations;
(4) semantic relations; (5) emotive, axiological and stylistic charge. 4

3.1. A simple word

A simple word is a word which consists of a single base with or without

inflections. A large number of simple words in English correspond to
simple words in other languages, at least in some of their senses, for
example English dog — Polish pies, E. bed, P. lozko, E. room — P. pokoj,
etc. In some cases, a simple word in English corresponds to a polymorphic
word in Polish and vice versa, for example handle u-chwyt, floor
podloga. Such comparative statements are theoretically uninteresting,
though it is by no means obvious on what grounds particular words in
one language should be matched with their equivalents in another lan-
guage. We take up this problem towards the end of this section.

3.2. Word formation

Word formation involves derivation by means of affixation, back-for-

mation, word-composition, shortening, acronymy, and some minor types,
such as sound interchange, distinctive stress, and sound imitation (see
74 Chapter V

Arnold 1973: 93 ff.). The investigator faces a vast area here, since lexical
equivalents across languages may display a considerable range of differ-
ences with respect to particular means employed in the formation of
words. No systematic correspondences seem to be in view, since a one-
morpheme word in one language may correspond to a compound word
in another language or a complex word in one language may correspond
to a simple word in another language, and so on, in all possible combi-
nations. Below are some examples of such correspondences:
English Polish
simple word complex word
seat siedzenie
bitter gorzki
darn cerowac
moral moralny

simple word compound word

hedge zywoplot
porcupine jezozwierz

complex word simple word

rubber guma
whiteness biel
flight lot
poster afisz
gift dar

compound word simple word

rainbow tecza
armchair fotel
father-in-law tesc
man-of-war okrgt
hedge-hog jez
moon-calf kretyn

compound word complex word

telltale plotkarz
sightseeing zwiedzanie
book-keeping ksi§gowosc
ear-ring kolczyk
ash-tray popielniczka
knee-cap rzepka
 Various levels of linguistic analysis 75

English Polish
complex word phrase
perlocator maszynka do kawy
Christmas Boze Narodzenie
ratter czlowiekjpies lapigcy szczury
complex word compound word
foreigner cudzoziemiec
Easter Wielkanoc
Since formal comparisons of individual lexical items do not seem to
lend themselves to any significant generalizations, contrastive studies of
word formation are better off if they are based on some conceptual
framework. Such an analysis could involve a comparison of various means
employed in the derivation of nomina actions, nomina agentis, nomina
loci, and also of adjectives of intensity, inclination, possibility, ability, or
of verbs of process, causation, instrument, and so on (cf. Seh-
nert —Sharwood-Smith 1974, as an example). As a matter of fact, any
aspect of the meaning can serve as a basis for cross-linguistic comparisons.
We thus obtain contrastive studies of various words expressing such
concepts as colours (Duczmal 1979; Schmitz 1983), kinship (Kalisz 1976),
modality (Aarts — Wekker 1982), or vision (Pasanen 1978).

3.2.1. Six possibilities for formal and semantic similarity

All types of words and their equivalents may exhibit various degrees of
formal and semantic similarity. Lado distinguishes six possibilities:
1. Similar in form and meaning — there are a number of words having
reasonably similar pronunciation and/or spelling in the compared lan-
guages. In some cases, they are borrowings, in other cases they can be
traced back to the same etymological source. Among examples from
English and Polish, one finds such words as map — mapa, lamp — lampa,
hotel hotel, nose nos, pilot pilot, minute minuta, second
Sekunda, son syn, and many others. Naturally enough, the spelling
conventions often conceal considerable differences in the pronunciation,
as in the case of such graphically similar words as psychiatrist
[sai'kaiatrist] psychiatra [psixi'atra] or pilot ['paibt] pilot ['pilot].
Conversely, differences in the spelling may conceal similarities in the
pronunciation, as in the case of such words as combine ['kambain]
kombajn [kombajn] or Shakespeare [Jeiksipis] — Szekspir [Jekspir], etc.
76 Chapter V

2. Similar in form but different in meaning — formally similar words

may be only partially similar in meaning or may have quite different
meanings. For example, the English word dog and the Polish dog are
partly similar in meaning, but the Polish word dog has a narrower range
as it denotes only one breed of dog; the word has undergone the semantic
process called specialization. The same process has affected such words
as toast [toast] from toast [taust], rekord [rekord] from record ['rekord],
prezent [prezent] from present ['prezant], and many others. Some words
belonging to this group are similar in form but totally different in
meaning, e. g., actually (really) — aktualnie (at present), desk (pulpit) —
deska (board), rower (one who rows) — rower (bicycle). In his numerous
contrastive analyses of German, French, and Italian, Schwarze calls such
pairs "false friends" {i falsi amici) since they often lead to errors in foreign
language usage (Schwarze 1979: 230). A detailed study of "false friends"
in Polish and English, or as the author, less metaphorically, calls them,
"deceptive words", can be found in Welna (1977).

3. Similar in meaning but different in form — a large bulk of equiv-

alents across many languages are of this type, at least as regards the
primary meaning of particular words. Among them one finds such pairs
as table — stol, chair — krzeslo, pen — pioro, tree — drzewo, etc. Although
the primary and perhaps some further senses of particular words in two
languages correspond, it is not necessarily the case with other meanings
of particular equivalents. Consider as an example the word tree and its
Polish equivalent drzewo. Their primary meanings correspond in that
both words refer to "perennial plant with single woody self-supporting
stem or trunk usually unbranched for some distance above ground" (The
Concise Oxford Dictionary). In both languages, however, the two words
have some further meanings which do not correspond. For example, in
Polish drzewo may also refer to wood or timber. Such "differential
contrasts" may also lead to serious learning problems (cf. Krzeszowski
1970: 81; Schwarze 1979: 227 ff.).

4. Different in form and in meaning — they are words with "strange"

meanings since, as Lado puts it, they "represent a different grasp of
reality" (Lado 1964: 217). The American English first floor is different
in form from apparently similar phrases in many languages including the
Spanish primer piso and the Polish pierwsze pi§tro, which are also different
in what they refer to as "first" The Spanish primer 'first' and the Polish
pierwszy 'first' do not refer to floor number one at the ground level but
 Various levels of linguistic analysis 77

to floor number one above the ground level so that the Spanish and the
Polish expressions refer to what American English calls second floor.
5. Different in their type of construction. Differences in this area
concern the morphological structure of words. For example, English has
lexical items consisting of a verb and a particle modifying it. Such verbs
often carry idiomatic meanings, which cannot be induced from the mean-
ings of the constituents. They are such lexical items as call up (telephone),
call on (visit), give up (resign), take in (cheat), take off (start), and so on
(see Hill 1968). Such two-word verbs have no formal counterparts in
most languages, including Polish. In some, less idiomatic instances the
verbal particles correspond to Polish prefixes as in

put vs. put on — klasc vs. na-kladac

write vs. write out — pisac vs. wy-pisac
call vs. call off — wolac vs. od-wolac
go vs. go out — chodzic vs. wy-chodzic
(Krzeszowski 1970: 82)
6. Similar in primary meaning but different in connotation. A word
may be harmless in connotation in one language but its equivalent in
another language may be offensive or taboo. For example, the Polish
word krwawy has no offensive overtones while its English equivalent
bloody is a swear word. Names of various body parts connected with sex
and defecation in many languages assume additional functions as swear
words. Apparently, however, some languages, such as Finnish, seem to
be fairly insensitive to this sort of semantic reorientation. Sometimes a
seemingly innocent word, such as kurczg 'chicken', assumes the additional
function of a swear word, in this case by phonetic analogy with the much
more effective kurwa 'whore, bitch' In any case, such words may be the
source of learning problems as their improper use constitutes sociolin-
guistic goofs. 5

3.3. Phraseological fusions, unities, and collocations

Another area of investigation covers what we have called phraseological

fusions, unities and collocations. Any differences between the three kinds
of units can be attributed mainly to varying degrees of motivation in the
combination of words within particular units: presumably, fusions are
semantically unmotivated, unities are partially motivated, and colloca-
tions are highly motivated. 6 Since clear-cut boundaries between these
78 Chapter V

types of phrases are impossible to draw, we shall discuss them jointly

under the general term phraseological units (cf. Ginzburg et al. 1966:
100 ff.).
Phraseological units are intermediate between compound words and
free phrases (loose phrases) in that like compound words they are severely
constrained with regard to the co-occurrence of their constituents, but
from the syntactic point of view they are phrases (cf. Arnold 1973: 148 ff.).
In many cases, phraseological units are semantically and functionally
equivalent to single words in more than one language, which is another
reason why they contribute to the domain of lexicology rather than
syntax. Phraseological units may exhibit considerable differences across
languages, both with respect to lexical congruity and with respect to
syntactic congruity. The following examples illustrate the lack of the two
types of congruity:

a. Lack of lexical congruity

English Polish
small hours wczesne godziny
'early hours'
fountain pen wieczne pioro
'eternal pen'
man and wife m^z i zona
'husband and wife'
heads or tails orzel czy reszka
'eagle or tails'
heavy storm gwaltowna burza
'violent storm'

The lack of lexical congruity in those pairs of equivalent phraseological

units resembles the lack of congruity of constituents of compounds and
their phraseological equivalents in Polish:

scare crow strach na wroble

(sparrows)
wrist-watch zegarek na
(hand)
night-school szkola wieczorowa
(evening)
 Various levels of linguistic analysis 79

b. Lack of syntactic congruity

Ν + Ν Ν + Adj
window curtains zaslony okienne
Ν + Ν Ν + Ν + genitive
brain trust trust mozgow
N's + Ν Adj 4- Ν
cat's paw kocia lapa
VP + NP V + Prep + NP
influence somebody wplyngc na kogos
In addition to comparisons of various syntactic patterns characterizing
equivalent phraseological units across languages, it is also possible to
examine and compare various degrees of valency of lexical items in
compounds and phraseological units. For example, the English phrase
calf love appears to be more lexically stable than its Polish equivalent
cielgca milosc inasmuch as in Polish one also finds cielgce lata (calf years),
cielgcy rozum (calf mind), cielgcy mozg (calf brain), and cielgcy zachwyt
(calf delight) as conventional phrases, which suggests that the adjective
cielgcy in this particular metaphorical sense collocates more freely with
other lexical items than its English equivalent calf does.

3.4. Semantic relations between lexical items

The next area of investigation embraces various types of semantic rela-

tions between lexical items and provides a convenient dimension for
cross-language comparisons. Among others the following types of se-
mantic relations can be described and compared: (a) polysemy; (b) ho-
monymy; (c) synonymity; (d) antonymity; (e) semantic fields. Let us look
briefly at each of these types of relations.

3.4.1. Polysemy

Most content words in everyday use in all natural languages have more
than one sense. Any specification of senses of a word in any dictionary
is a recognition of the polysemic character of that word. Any specification
of various equivalents in L2 of a lexical item in L, is not only a recognition
of the polysemic nature of words but also an exercise in lexical contrastive
studies. Insofar as bilingual dictionaries must contain such specifications,
the phenomenon of polysemy is among the most crucial in lexical con-
80 Chapter V

trastive studies. This is reflected in the fact that polysemous words

constitute networks of interconnections in any sizeable dictionary: for
every polysemous word in L^ there exists a set of equivalent items in L2,
with each such item having a set of equivalent items in L,, etc. For
example, the English word table has the following twelve equivalents in
Polish: 1. stol; 2. towarzystwo przy stole, biesiadnicy,; 3. plyta; 4. napis
wyryty na kamiennej tablicy, ~ s prawa, ustawy; the ten ~ s dziesigcioro
przykazan; 5. plaskowyz; 6. karnisz; 7. plaszczyzna drogiego kamienia,
klejnotu; 8. dlon; 9. blaszka kostna; 10. tablica; tabela; spis; wykaz; 11. ~
s tryktrak; 12. stol przewodniczpcego parlamentu angielskiego (Jan
Stanislawski, The Great English-Polish Dictionary).
The Polish word plyta, which is one of the equivalents of table, has
the following equivalents in English: 1. slab·, plate; sheet·, board·, 2. record;
3. table-land; 4. wrest-block; wrest-plank (np. fortepianu); plaque (pamiat-
kowa) (Jan Stanislawski, The Great Polish-English Dictionary).
In turn, the English word board, one of the equivalents of plyta, has
eight equivalents in Polish, and so on.
To a large extent the same is true of

3.4.2. Homonymy

Homonymy is the relation between lexical items which are formally

identical as in the case of the Polish zamek\ 'castle', zamek2 'lock', 'zip',
zamekj 'halving' (archaic) and zamek4 'hinge ligament' (in muscles), where
lock has eight equivalents in Polish.
The presentation of lexical stocks in two languages in terms of such
networks of interconnections is possible for a given finite set of lexical
items in both languages, but it is a futile procedure if textual settings in
which particular lexical items appear are not considered. For the same
reason, paradigms of grammatical forms are meaningless and do not
constitute a grammar of a language, if they are not seen as elements of
structures into which they can be inserted.

3.4.3. Synonyms

Like almost every notion in lexicology, the term "synonym" has no

clearly defined scope. Roughly speaking, synonyms are the words which
are different in phonic/graphic form but similar in denotational (refer-
ential) meaning and interchangeable at least in some contexts. This
 Various levels of linguistic analysis 81

definition of synonyms makes use of the well-known distinction between

denotational and connotational meaning, which itself is not clear, as we
shall see later. It also relates to the notion of context. This is so because,
firstly, it is impossible to talk about synonyms of individual words as
such. Usually a particular sense of a word is synonymous with one or
more senses of another word. Moreover, it is impossible to match words
as synonyms if they are examined in isolation from the contexts in which
they can appear. For example, the verb to read in the context read dreams
is a synonym of to interpret while in the context read one a lesson it is a
synonym of to give. The matching of synonyms is thus basically no
different from the matching of lexical equivalents across languages insofar
as in both cases the selection of the appropriate sense, essential in choosing
the appropriate synonym or the appropriate equivalent, is determined by
the context in which the investigated word appears. Secondly, the simi-
larity of meaning involves only that part of the overall lexical meaning
which is called denotation (reference) in contrast to connotation in the
well-known distinction. Presumably, words like live, dwell, and reside in
their relevant senses are denotationally similar, but differ with respect to
their stylistic values, i. e. with respect to their connotations: live is said
to be neutral, dwell is poetic, while reside is formal. Thirdly, synonyms
are interchangeable only in some contexts, while in other contexts they
are not only not interchangeable but also sometimes turn out to be
antonyms. Such is the case with the words exceptional and abnormal. In
the context of the weather is , the two words are interchange-
able, but in the context my son is , the two words are near
antonyms (Quirk 1962: 120).

3.4.4. Antonyms

Antonyms involve similar problems. As in the case of synonyms, one can

only talk about a specific sense of a particular word being antonymous
to a specific sense of another word. For example, the word good in most
senses and in most contexts is antonymous to bad as in a good girl vs. a
bad girl, to be good at something vs. to be bad at something, etc. Yet, good
in the sense 'not less than a certain amount' as in a good three miles or
a good way is not an antonym of bad since in a bad three miles and in a
bad way the word bad is an antonym of good in its primary sense, i.e.,
'having the right quality' The antonymous polarity in the second sense
is in fact impossible to express by means of a single word, which, if it
82 Chapter V

existed, would have to express the notion 'less than a certain amount'
The nearest possible candidate would be the word little as in a little three
miles and a little way.
Contrastive studies of synonyms and antonyms do not basically differ
from such studies of other words, since in all cases the essential problems
are connected with the selection of appropriate senses on the basis of
relevant contexts.

3.4.5. Semantic fields

Semantic fields provide grounds for yet another way of grouping words.
Those words which share a common concept are said to constitute
semantic fields (Trier 1931). The basis for grouping is always extralin-
guistic since words are grouped in semantic fields because things which
they refer to are connected with extralinguistic reality (Arnold 1973: 206).
Some examples of semantic fields are: colours, kinship terms, pleasant
and unpleasant emotions, military ranks, educational terms, gastronom-
ical terms, vehicles and their parts, being at a place, leaving, sports,
games, etc. Contrastive studies of words constituting semantic fields in
various languages are a rewarding activity and are so often performed
that they are sometimes identified with lexical contrastive studies in
general (e.g., Duczmal 1979; Sehnert 1975; Kalisz 1976; Tomaszczyk
1976; Schmitz 1983).
The fifth type of comparison involves stylistic-emotive charge of lexical
items, i. e., the connotational component of their meaning. Synonyms or
groups of words with similar denotational meaning usually differ with
respect to their connotations. The stylistic charge of lexical items can be
described in terms of features falling into keys — oratorical, deliberative,
consultative, casual, and intimate — and in terms of parameters super-
imposed on keys — status, technicality, dignity, and conformity (Gleason
1965: 358 ff.). The emotive charge of lexical items can be described in
terms of markedness with respect to such features as "appreciative" or
"positive", "depreciative" or "negative" and "neutral" The number of
parameters along which lexical items can be described has never been
definitely established, 7 which obviously creates problems for lexical con-
trastive studies. Contrastive studies of synonyms and semantic fields,
enhanced by the distinctions suggested above, are a rewarding as well as
amusing area of investigations (cf. Lawendowski 1974).
 Various levels of linguistic analysis 83

The area of lexical contrastive studies bristles with problems, both

theoretical and practical. By definition, lexicology deals with those ele-
ments of language which cannot be generalized in terms of rules within
the domain of grammar. However, as has been repeatedly pointed out,
no strict boundaries between lexicology and grammar (syntax) can be
drawn. This fact creates a challenge for linguists, who, so far, have been
unable to deal satisfactorily with the uncertain linguistic status of phra-
seological units, which have syntactic properties of loose phrases but, like
words, are much more severely constrained with respect to the selection
and order of their constituents. Since these constraints are not dependent
on the syntactic structure of the units in question, and since they are not
systematic, like other unpredictable phenomena they are subject to listing
and thus fall within the domain of lexicology.
Yet it is impossible to ignore the systematic and predictable character
of some words as regards their inner structure. As Halle suggests, a
linguistic theory should account for the native speaker's ability to distin-
guish between well-formed combinations of morphemes constituting
words and ill-formed combinations of morphemes. Halle argues that
lexical insertion transformations take items from the dictionary rather
than from the list of morphemes a large part of the dictionary is
stored in the speaker's permanent memory and he needs to involve
the word-formation component only when he hears an unfamiliar
word or uses a word freely invented (Halle 1973: 16).
Being systematic in nature, word formation can be handled in terms
of rules generating words (cf. Saumjan's (1968) concept of 'word-gener-
ator'). Such rules can be contrasted with analogous rules in another
language according to principles which are not different from those that
govern syntactic comparisons. Comparisons of this sort may turn out to
be theoretically rewarding. Still one must bear in mind that in a large
number of cases, language users treat derived lexical items as indivisible
(though potentially analysable) entities and thus subject to listing rather
than to generating in terms of rules. Therefore, it appears that the area
of word-formation is a border area between those linguistic phenomena
which are subject to listing and those that are subject to enumerating in
terms of a finite set of rules generating an infinite set of lexical items.
Also "vertical" and "horizontal" relations, as grasped by Nowakows-
ki's lexicon (Nowakowski 1977: 27), may turn out to have a more
systematic character than has been previously ascribed to them. However,
as in the case of word-formation, any rules which might underlie such
84 Chapter V

relations are probably dormant in language users and are activated only
in those special instances when the need to employ those rules arises.
Among "vertical" relations, Nowakowski distinguishes between hypon-
ymy and hyperonymy (see also Lyons 1968: 453 ff.), and among "hori-
zontal" relations those of paraphrase, synonymy, oppositeness, reference,
presupposition, and metaphorization. As we have suggested earlier on,
all these phenomena can be investigated in the contrastive dimension,
regardless of whether in particular languages they are handled in terms
of rules or in terms of lists.
The overlapping of grammar and lexicology is also evident in the
description of lexical items themselves. Each lexical item is described in
terms of a quadruplet of features: LE = </, s, m, r>, in which LE stands
for "lexical entry", / is a specification of phonological features, s is a
specification of syntactic features, m is a specification of semantic features,
and r is a specification of relational features.
Phonological features are specified as distinctive features of segments
(cf. Fisiak 1976: 120) or abbreviated as a phonological transcription.
Syntactic features specify syntactic contexts into which particular lex-
ical items can be inserted. Each word is categorized as a noun, verb,
adjective, adverb, preposition, pronoun, etc. Certain words are further
subcategorized either in terms of inherent features, such as [ +countable],
[ + common], [ +abstract], etc. or in terms of contextual features, which
specify actual syntactic contexts into which particular subcategories of
words fit. For example, certain verbs (traditionally called "transitive")
can occupy positions immediately preceding noun phrases while other
verbs (traditionally called "intransitive") do not fit such contexts. The
information conveyed in terms of inherent and contextual features can
reach a considerable level of "delicacy" and shades into information of
semantic rather than syntactic sort. Thus, whereas in
(1) *John elapsed the time
the verb elapsed has been inserted in defiance of the feature subcatego-
rizing it as intransitive, i. e. one which does not tolerate an object noun
phrase, in
(2) *John drank the table
the verb appears in the context of the object noun phrase with the feature
[ — liquid], thus violating a selection restriction of semantic rather than
syntactic nature (cf. McCawley 1968: 133 ff.). Therefore, information of
this sort ought to be contained in the specification of semantic features
 Various levels of linguistic analysis 85

which characterize particular lexical items within subcategories specified

independently by syntactic features <s>.
Finally, relational features, relate a specific lexical item to other mem-
bers of the same subcategory with respect to synonymity, hyponymy,
entailment, etc.
Thus, lexicon overlaps with syntax by virtue of <s>; but it also overlaps
with the outer reality by virtue of <m>, i. e., semantic features which refer
to properties of things in the world at large and which eventually are
responsible for the distinctions in the meaning of particular words. But
such distinctions, in most cases, are based on our knowledge of the world
and thus constitute encyclopedic rather than linguistic knowledge. One
can expect a great deal of fuzziness in this area as well. But encyclopedic
information turns out to be as much a part of the meaning of words as
linguistic information. As Langacker puts it

The distinction between semantics and pragmatics (or between lin-

guistic and extra-linguistic knowledge) is largely artifactual, and the
only viable conception of linguistic semantics is one that avoids such
false dichotomies and is consequently encyclopedic in nature (Lang-
acker 1987: 154).
Practising lexicographers have been well aware of the problem when
they did not hesitate to include "pragmatic" information in their descrip-
tions of various lexical items. Without "pragmatic" knowledge, a great
number of words could not be defined in a sufficiently precise way to
warrant the necessary contrasts with other lexical items within the same
semantic sphere. This indeed happens on those occasions when the lexi-
cographer disregards knowledge of the world in attempting to define a
lexical item. Such is the case with cat and dog, respectively defined as "a
domesticated carnivore, bred in a number of varieties" and as "a domes-
ticated carnivore, bred in a great number of varieties" (The Random
House Dictionary of the English Language. College Edition). If one wishes
to distinguish between the two species of animals, as indeed one ought
to in a decent dictionary, one has to make a more liberal use of our
knowledge of the world as in the following definitions of cat and dog,
respectively: "a small, four-footed, flesh-eating mammal, often kept as a
pet or for hunting rats and mice ..., a domesticated mammal, related to
wolves, foxes, and jackals, that is kept as a pet, for hunting, and for
guarding property" {The World Book Encyclopedia Dictionary).
From the contrastive point of view, one of the thorniest problems is how
to establish equivalence of lexical items across languages. One of the meth-
86 Chapter V

ods involves the decomposition of lexical items into features and considering
as equivalent those lexical items which can be decomposed into identical
sets of features. Some difficulties connected with this method transpire from
Di Pietro's analysis of flesh and meat in English and their equivalents in
various languages. One rather obvious difficulty consists in the fact that
each feature resulting from the decomposition must be regarded as a "the-
oretical" construct with little, if any, independent motivation.
Decomposition of lexical items into "atomic" features in fact means a
reduction of the concepts associated with these lexical items into some
more fundamental concepts, which are presumably universal and can
thus serve as tertium comparationis for establishing equivalence between
lexical items across languages. Presumably, then, equivalent words would
have identical matrices of "features", i. e., the atomic concepts obtained
from the decomposition. Adopting this procedure results in the necessity
of constructing a special "theory" for every pair of equivalent items
across languages. Every feature resulting from the decomposition is a
"theoretical" construct with little, if any, independent motivation. The
problem which arises is analogous to the one encountered in connection
with the recognition of semantic markers, discussed earlier on, whose
"number is indeed legion" (Bolinger 1965: 564). What is even worse,
there seems to be no upper limit to this number since the process of
decomposition of words into components of meaning appears to have no
end, and it, moreover, often leads to circularities. The sort of problems
that one encounters when adopting the decompositional procedure can
be easily seen when a standard example involving the words man, boy,
woman, and girl is considered. These words are usually decomposed into
the features [male], [female], [young], and [adult], according to the fol-
lowing matrix:

[male] [female]
[adult] man woman
[young] boy girl
This sort of framework can hardly serve as tertium comparationis for
lexical contrastive studies for the following reasons:
Firstly, none of the features resulting from the decomposition is a
primitive concept: they can all be decomposed further. For example, The
Concise Oxford Dictionary defines young as "that has lived a relatively
short time; not mature [italics my own] or fully developed" Further
decompositions usually lead to circularities: mature is defined as "com-
 Various levels of linguistic analysis 87

plete in natural development or growth; fully developed in body and mind"

and finally, to complete the circle, develop is defined as "to grow into a
fuller, higher, or maturer condition" Likewise, the definition of adult
resorts to the same circularity: "grown up, having reached the age of
maturity'''' Such examples are a matter of rule rather than exception.
Since the features resulting from decomposition are not primitive con-
cepts, they are usually ambiguous, which renders them useless as potential
tertia comparationis. For example, young, in addition to the sense described,
has some further senses, such as 'newly begun or formed; not advanced:
The evening is young', 'pertaining to or suggestive of youth or early life:
young for her age\ 'vigorous or fresh, youthful' Although these senses are
related through "family resemblance", they do not necessarily correspond
to a set of parallel equivalents in another language. For example, young
corresponds to mlody in Polish in its most prototypical sense "that has lived
a relatively short time', but not always in the sense 'being in the early period
of growth' when it refers to plants or feelings. In this sense, young may
correspond to the Polish adjective swiezy. The polysemous character of such
"features" and the various connotations, inseparable from their basically
lexical character make them very questionable as possible tertia compara-
tionis for lexical contrastive studies.
Secondly, the reductions inherent in decompositional analyses of this
sort do not seem to reveal accurately the meaning of the analysed lexical
items insofar as they leave some area of the meaning unaccounted for.
That boy is not merely 'young male' is well seen in the collocation young
boy, which is not tautologous even if the componential analysis suggests
that the adjective young, when placed in front of a noun containing the
feature (young) as one of its components, should produce tautology.
Similarly, the expression young man would be impossible since the analysis
suggests the incompatibility of the adjective young with a noun containing
the feature (adult) as one of its components. That fact that such ex-
pressions as young man are semantically well-formed demonstrates the
presence of some aspects of meaning which escape description in terms
of the componential analysis of this sort. What is even worse, such
analyses lead to faulty predictions.
Thirdly, because of the presence of areas of meaning which escape com-
ponential analysis, the matching of feature matrices will also lead to faulty
predictions in the domain of lexical contrastive studies. There is no doubt,
for example, that the features (young) and ( m a n ) are somehow present
in such lexical items as boy, lad, youth, urchin, and master. The most ap-
propriate ways in which these words can be matched with their equivalents
88 Chapter V

in a foreign language cannot depend on such features as <young> and

<male> alone. A great deal more is at stake and a host of other features
would have to be postulated to account for contrasts between these words.
The nature of these additional features, especially with reference to the
emotional aspects of the meaning of such words has never been made suf-
ficiently clear to provide a basis for contrastive studies.
Fourthly, the existence of various fuzzy concepts casts further doubts
on the componential analysis of the sort described above. The compo-
nential analysis rests on the classical view of sufficient and necessary
conditions, determining category membership, whereby an element is a
member of a category if and only if it is characterized by a given set of
features which define a given category. The presence of these features
creates both necessary and sufficient conditions of category membership.
This approach may work reasonably well when categories with clear-cut
boundaries are considered. For example, the category "bird" can be
defined in terms of two features (feathered) and (vertebrate). Anything
to which these two features can be attributed is a bird. Thus, the two
features are sufficient and necessary as conditions of membership in the
"bird" category.
This procedure cannot be applied to various concepts which have no
clear-cut boundaries and which shade into other concepts in a non-discrete
way. For example, such English words as forest, wood, woodland, grove,
copse and spinney express non-discrete concepts. The differences between
them are a matter of degree. The American Heritage Dictionary defines
spinney as "a small grove", and the same dictionary defines grove as "a
small wood", while wood is defined as "a dense growth of trees; forest"
Clearly, then, since the adjective small expresses an inherently relative no-
tion, the differences between these concepts are a matter of degree and as
such cannot be stated in terms of necessary and sufficient conditions.
One faces the same sort of problem with certain polysemous lexical items
such as game. There is no set of "precise necessary and sufficient conditions
for something to be game" (Lakoff 1982: 14). Various games are related
through "family resemblance" which allows us to identify such diverse
activities as ring-around-the-rosie, solitaire, Monopoly, and football as be-
longing to the category "game" (Wittgenstein 1953 [1972]: 50 ff.).
In view of these problems, componential analysis cannot be adopted
either as a method of analysis of fuzzy concepts or as a provider of tertia
comparationis for lexical contrastive studies.
Bilingual dictionaries are practical results of lexical contrastive studies.
Such dictionaries consist of lexical items in one language and their
 Various levels of linguistic analysis 89

equivalents in another language, usually selected on the basis of linguistic

intuitions of the compilers. A more rational procedure would be to match
lexical items in two languages on the basis of detailed analyses of nu-
merous equivalent texts written or spoken in both languages. This sort
of analysis would provide a rigorous way of determining the number of
senses of polysemous words: a careful comparative analysis of transla-
tionally equivalent texts would reveal the number of ways in which a
particular item in one language could be rendered in another language.
The number of these renderings would correspond to the number of
senses of that lexical item (Manczak 1970). Such an analysis would
provide statistical grounds not only for matching equivalents but also for
listing them in the order from the most frequent to the least frequent in
all those numerous instances when a lexical item in one language has
more than one equivalent in another language. In any case, no decom-
position of lexical items into semantic features would be required to
guarantee correct matching of equivalents. (For more arguments against
decomposition see Wierzbicka 1980: 70 ff.).
Thus, a large bulk of any bilingual dictionary is a juxtaposition of lexical
equivalents across languages on the basis of the identity or at least similarity
of referential meanings, such as cat — hot, table — stol, wireless — radio,
walk — chodzic, yellow — zolty, small — maly, often — czgsto, diet — sejm,
primary school — szkola podstawowa, M.A. thesis — praca magisterska,
power plant — elektrownia, etc. Any person using a bilingual dictionary
certainly expects to find this sort of information in the first place. Yet, he
is seldom happy if his dictionary does not go beyond listing equivalents,
even if, or perhaps especially if, more than one equivalent is given for a
particular item, as indeed is the case in the vast majority of cases. Most
dictionary users inevitably wish to be able to find information about the
use of particular lexical items in both situational and linguistic contexts.
They also welcome information about the possibilities of forming derived
lexical items on the basis of those listed in the dictionary. A dictionary which
fails to provide this sort of information is considered to be inferior to a
dictionary in which such information can be found (Tomaszczyk 1979:
103-119).
Here we arrive at a difficult problem: how much information about
contexts in which particular words appear should be included in a dic-
tionary? The answer partly depends on the needs of the prospective users,
different in the case of native and non-native users. Ideally, every bilingual
dictionary should consist of four parts: 1. L, *-*• L2, 2. L 2 <-»· L, for users
of L, as a native language and 3. L, <-• L 2 , 4. L 2 «-* L t for users of L 2 as
90 Chapter V

a native language. Generally speaking, however, a non-native user of a

target language will welcome as much information about linguistic and
extralinguistic settings of lexical items as is possible to contain in a
dictionary. This is to say that, other things being equal, the quality of a
dictionary is in proportion to its size seen as a volume of the material to
the right of each entry. It also means that there is no such thing as the
best dictionary since a better (larger) one can always be compiled, and
that the lexicographer's job is never done.
What has been said so far suggests that all types of lexical contrastive
studies, regardless of their questionable theoretical reputation, will be of
value in compiling dictionaries since such studies will increase the inform-
ative part of the dictionary by just the sort of information which is so often
sought by the learners: the contexts. It does not matter that some of these
comparisons, for example, of thematic groups, involve encyclopedic rather
than linguistic knowledge. This sort of knowledge must be included liberally
in any dictionary which is intended to help the learner to communicate in
real life situations. "Encyclopedic" descriptions of those lexical items in L,
that have no conceptual equivalents in L2 is often a must. Such words as
vicar — in the Church of England, the priest of a parish who receives a
stipend or salary from a layman or religious corporation responsible for
appropriating the principal revenues; villadom — suburban society; thrasher
— any of various New World songbirds of the genus Toxostoma, having a
long tail, a long curved beak, and, in several species, a spotted breast; kilt
a knee-length skirt with deep pleats, usually of a tartan wool, worn
especially as part of the dress for men in the Scottish Highlands, and many
other lexical items have no equivalents in other languages, so that in bilin-
gual dictionaries they must be rendered as descriptions making a direct
appeal to the reader's knowledge of the world. Such situations often result
in borrowings from one language to another.

4. Topics in pragmatic contrastive studies

The status of pragmatics as a level of linguistic analysis is by no means
certain. Fillmore thus expresses his doubts in this matter:
there is no characterization of linguistic pragmatics on which linguists
are in agreement, nor is there, in fact, a universally convincing case
that such a field exists (Fillmore 1984 a: 119).
There are at least two reasons which cause this situation. The first has
to do with the lack of agreement about the scope of linguistics, while the
 Various levels of linguistic analysis 91

other has to do with the lack of agreement about the scope of pragmatics,
whether treated as part of linguistics or not.
Linguistics can be considered in a broader or in a narrower sense. In
the broader sense, linguistics deals with the description and possible
explanation of facts connected with human communication. This sort of
linguistics can be called, after Yngve (1975: 47 — 62), "human linguistics"
in contrast to what may be called "code linguistics", which deals with
language as an abstract system in isolation from the psychosocial setting
in which it is used. Depending on whether one adopts the broader or the
narrower view of linguistics, the area of pragmatics will, respectively, fall
within or outside the domain of linguistics proper. Whereas "code lin-
guistics" in the de Saussurian and Chomskyan tradition is mainly pre-
occupied with grammatical competence (Chomsky 1965: 9), "human
linguistics" deals with the process of communication more than with just
the code, instrumental in this process. "Human linguistics" deals with
what Hymes (1972: 291) calls "communicative competence", i.e. briefly
speaking, with who says what to whom, where, when, how, and why (cf.
James 1980: 101).
The advocates of linguistics conceived in the broader human perspec-
tive are usually agreed in the rejection of the traditional clear-cut dis-
tinctions between various components of language, rejecting what is often
called modular approaches to language (Chomsky 1984). Instead, they
advocate holistic interpretations of language, which are to replace those
approaches which distinguish autonomous, well-defined, and clearly sep-
arated components of language, such as syntax, semantics, phonology,
and lexicon. By the same principle, the traditionally assumed separation
of semantics and pragmatics has to vanish.
Unfortunately, pragmatics itself is not a well-defined or a well-delim-
ited area. For example, Leech (1983) distinguishes between general prag-
matics and socio-pragmatics. It is the latter that deals with the various
specific "local" conditions on language use (Leech 1983: 10). But such
"local" conditions constitute what Fillmore (1984 a) considers as the
central part of his pragmatic contrastive studies, while Leech places them
on the margin of his own pragmatic considerations.
Although Leech admits that pragmatics deals with how language is
used in communication, he still excludes various "local" conditions which
deal with the "cooperative principle" and the "politeness principle" from
the field of what he calls "general pragmatics" by relegating them to the
field of "socio-pragmatics" This latter field is very much in the centre
of Fillmore's contrastive pragmatics (Fillmore 1984a: 127 ff.).
92 Chapter V

For Gazdar, who in this respect, by his own admission, follows such
scholars as Karttunen, Stalnaker, and Thomason, pragmatics deals with
those aspects of the meaning of utterances which cannot be accounted
for by straightforward reference to the truth conditions of the sentence
uttered (Gazdar 1979: 2).
Consequently, semantics would deal with those aspects of the meaning
which can be described in terms of truth conditions (cf. Montague 1974).
Gazdar himself admits, however, that the distinction between semantics
and pragmatics is often difficult to draw in a clear-cut way, mainly due
to the fact that Grecian pragmatic implicatures also rely on truth-con-
ditional properties. On the other hand, various indexical expressions,
which are naturally connected with relations between linguistic forms and
elements of the world at large (and as such would fall within the scope
of pragmatics), are "standardly and naturally handled with truth-condi-
tional apparatus" (Gazdar 1979: 2). Therefore, they must be assigned to
the domain of semantics.
Such uncertainties and fuzzy areas are to be expected in natural
languages, and at this stage we are not going to concern ourselves with
precise or clear-cut definitions of pragmatics since the task clearly goes
beyond the scope of the present work. Instead, we shall take a descriptive
look at the work of three scholars who have contributed most to the
development of contrastive pragmatics. A brief summary of their research
will, on the one hand, show the diversity and the vast scope of the field;
and, on the other hand, it will provide a survey of topics in the cross-
linguistic dimension.
Riley (1981: 123) maintains that the main concern of pragmatic lin-
guistics is the act of communication, of which the speech act is one
possible realization, since one can communicate agreement with a nod of
the head just as effectively as with the word yes. In Riley's view one
cannot claim to provide an adequate account of meaning without con-
sidering such vital questions as: Who is speaking to whom? When? Where?
What is the nature of their relationship? Of the circumstances? What
activity are they involved in? What is its purpose and that of the com-
munication? (Riley 1981: 123). He further claims that the study of com-
municative acts within pragmalinguistics rests crucially on the theory of
illocution which deals with such communicative acts (including speech
acts!) as accepting, agreeing, disagreeing, explaining, denying, suggesting,
hypothesizing, promising, offering, etc. All these can be expressed verbally
but also paralinguistically or non-verbally by grunts and noises, gestures,
 Various levels of linguistic analysis 93

mimicry, body movements, etc. The illocutionary values of communica-

tive acts, which Riley briefly calls functions, have no one-to-one corre-
spondence to grammatical structures. Thus, a linguistic form like You are
not going out, which is an affirmative sentence, can perform a variety of
pragmatic functions, such as prohibiting, confirming, threatening, ex-
pressing surprise, and stating. On the other hand, one function, such as
agreeing, can be expressed through a variety of linguistic and paralin-
guistic means, such as Yes, sure, right, fine, Ο. K. repetition of part of
the utterance (You're leaving? I'm leaving), nod of the head, overt ex-
pression of agreement, such as I agree, I accept your point, I see what
you mean, etc.
Riley proceeds to suggest a contrastive procedure whereby one com-
pares the discourse structure of different languages in terms of patterns
of codings, i.e. the communicative content of utterances, change of
address, accounting for the selection of the addressee by the speaker, and
change of the first speaker. Such descriptions of discourse structure
provide information about social roles, participant states, formality, and
situation. Briefly speaking, it provides information about what Riley
considers to be non-semantic parameters of meaning, essential in inter-
active discourse.
Riley describes discourse as a sequence of illocutionary acts and a
series of interactive acts parallel to a series of formal structures called
realization. There is no one-to-one correspondence between these parallel
structures constituting discourse, which Riley describes in the following
words:

By Formal Structure here we mean realisation: the set of message-

bearing elements (verbal, paralinguistic, non-verbal) in a situation.
These elements have substance and are realisations of various systems
and structures whose organisation can be described in terms such as
class, unit, structure and distribution. Illocutionary Structure: here
we deal with sequences of illocutionary acts (e. g. Inviting, Accepting,
Confirming, Thanking). Interactive Structure: here we describe lin-
guistic organisation in terms of interactional tactics: turns, (opening,
reply, closing) address, relative distribution of utterances (exchange
transaction) (Riley 1981: 129).

Riley illustrates these distinctions with the following example:

Mr and Mrs A approach Mr Β to ask about the way in London.

94 Chapter V

(1) Mr A: Sorry, but can you tell us the way to St James' Park, please?
(2) Mr B: Are you on foot?
(3) Mrs A: Yes, we are. Is it far?
(4) Mr B: Then you just go down those steps there and turn right.
(5) Mr A: Thank you very much.
(Riley 1981: 129-130)
This dialogue has the following illocutionary structure:
(1) Requesting information
(2) Requesting information
(3) Informing, Requesting information
(4) Informing
(5) Thanking
and the following interactive structure:

(1) Opening Exchange

(2) Response
(3) Opening
(4) Response Exchange
(5) Closing
Mr A is Speaker in (1) and (5), Hearer in (3) and (4), Address in (2),
Mrs A is Hearer in (1), (2) and (5), Speaker in (3) and Address in (4),
while Mr Β is Address in (1) and (3), Speaker in (2) and (4) and Hearer
in (5).
Adopting this framework, one can conduct pragmatic contrastive
studies by
(a) comparing what functions a particular structure in one language
can realize in contrast to the range of functions of a similar structure in
another language. Riley's example involves three languages: French,
English and Swedish. Now, the construction If (si, om) + 'conditional
clause' can be used in at least three different functions, i. e., to hypothesize,
to request information, and to suggest, as in

(i) S'il arrive, je le lui dirai.

Om han kommer, ska' jag berätta dat for honom.
(ii) Si je sui pret? (C'est bien ce que tu viens de me demander?)
Om jag är färdig? (jo, jo!)
(iii) Om vi sku' gä pä bio i kväll!
Et si on allait au cinema ce soir!
 Various levels of linguistic analysis 95

In English (and we might add, also in Polish), the corresponding

structures do not function as requests for confirmation or as suggestions;
they can also function as expressions of hypothesizing:

(i) If he comes, I'll tell him.

Jezeli on przyjdzie, powiem mu.
(ii) * If I am ready?
* Jezeli jestem gotow?
(iii) *If we go to the cinema this evening.
* Jezeli pojdziemy do kina dzis wieczorem.

(b) looking for various realizations of one particular function in the

compared languages. For example, suggesting can be realized in a variety
of ways in French, Swedish, English, and (we add) Polish:

French Et si on allait au cinema ce soir.

On pourrait peut-etre aller au cinema ce soir.
Vous n'auriez pas envie d'aller au cinema ce soir.
One possibilite serait d'aller au cinema ce soir.

Swedish Jag tänkte att vi künde gä pä bio i kväll.

Hör skulle det vara att gä pä bio i kväll?
Vi künde gä pä bio i kväll, eller hur?
Om vi skulle gä pä bio i kväll!

English How about us going to the cinema this evening?

I tell you what, let's go to the cinema this evening.
Why don't we go to the cinema this evening?
I wouldn't mind going to the cinema this evening.

Polish A moze bysmy poszli do kina dzis wieczorem.

Moznaby pojsc do kina dzis wieczorem.
Sluchajcie, chodzmy do kina dzis swieczorem.
Kto idzie do kina dzis wieczorem?

(c) considering sentence-function relations in sequences, i.e. a con-

trastive look at illocutionary structure. Here, as Riley points out, an
enormous a m o u n t of actual language data in various languages could be
investigated. Riley's example involves an exchange used for illustrative
purposes some time earlier by the present a u t h o r (cf. Krzeszowski 1970:
23):
96 Chapter V

English (1) That's a very pretty dress you're wearing.

(2) Oh, thank you very much.
Illocutionary structure: compliment + thanks
Swedish (1) En sä vacker klänning du hart
(2) Tack sä mycket.
Illocutionary structure: compliment + thanks
French (1) Que c'est jolie la robe que tu portes!
(2) ?
Illocutionary structure: compliment + ?

Polish (1) Ladnq masz sukienkg.

(2) E, nie takiego. Kupilam jg w Modzie Polskiej.
(3) Niemozliwe! lie dalas?
(4) ?
Illocutionary structure: compliment + downplaying + explanatory
comment + doubting + request for information + ?

One can see that the response to a compliment in Polish in contrast

to the other languages evokes a fairly complex sequence of illocutionary
structures. However, like in French, in Polish a compliment is not fol-
lowed by an expression of thanks, but rather by an expression of down-
playing one's own contribution to whatever is being complimented.
Various other situations and the linguistic forms associated with them
can be thus compared. Riley mentions such situations as business nego-
tiations, telephone calls, casual encounters, etc. (Riley 1981: 135).
Riley's proposal concerning pragmatic contrastive studies is valuable
in that it extends the traditional scope of contrastive studies, which is
restricted to language in isolation from the communicative settings. Al-
though not altogether new as a practical procedure, it explicitly draws
the analysts' attention to a somewhat neglected area of contrastive studies
and suggests specific topics for investigation.
A major inadequacy of Riley's proposal is that it does not distinguish
between more and less prototypical functions of various linguistic forms,
and, conversely, more and less prototypical expressions of various func-
tions. It may be the case that there are various possible ways of expressing
a certain function in a given context, but some of them are more likely,
or more appropriate, or less marked in a given context than others. A
contrastive analysis must consider these possible differences in the status
— more or less privileged in a given context — of the alternative forms.
 Various levels of linguistic analysis 97

Although structures and functions are not in one-to-one correspondence,

as Riley repeatedly and correctly points out, they still may have more or
less prototypical matches. For instance, although declarative sentences
perform a number of functions, such as prohibiting, confirming, threat-
ening, etc. their most prototypical function is stating; this is the kind of
fact which cannot be overlooked in either an adequate analysis of prag-
matic functions of various linguistic forms or in an adequate contrastive
analysis, which may or may not reveal some universal tendencies. (For
more details on this issue see Chapter XI.)
Fillmore's contrastive pragmatics embraces a much more modest scope
than does Riley's. Having rejected the commonly held view that the
"meaning" of words must be kept sharply distinct from our knowledge
of what words are associated with in the world at large (e.g., what
carpenters actually do as opposed to the meaning of the word "carpenter"
itself), Fillmore proceeds to restrict the scope of his pragmatic analyses
to what he calls special pragmatic practices or "small facts" Fillmore
distinguishes between "large facts" and "small facts" in pragmatics.
Among the "large facts", he includes politeness systems, patterns of
indirectness, repertories of registral differences, patterns in the rhetorical
organization of discourse, and the special devices languages use for
constructing narrative texts. Among the "small facts", he includes things
"that need to be learned one at a time" (Fillmore 1984 a: 127). Each of
these topics constitutes a field broad enough to merit special investigation
and each of these topics can be studied contrastively.
"Small facts" include linguistic phenomena at various levels of analysis,
which cannot be adequately described without recourse to their use, i.e.
to their pragmatics. "Small facts", too, can be studied contrastively since
they may pertain to different situations across languages. When translated
into another language, such linguistic forms do not always fit well into
analogous situations, usually due to the violation of some "large scale
principles" (Fillmore 1984 a: 127 — 128). One of Fillmore's examples in-
volves English negative w/ry-questions, which have the illocutionary force
of suggestions as in Why don't we go the opera tonight? Semanto-syntactic
equivalent questions in many languages do not have a similar illocution-
ary force, i.e., they cannot be just as naturally used as suggestions.
Examples of this sort can be quoted endlessly and are usually given in
pedagogical grammars of the "communicative" type (e.g.,
Leech — Svartvik 1975). Ideally, this sort of information should be pre-
sented contrastively, since here, perhaps more than in other areas, there
exists the danger of negative transfer. However, phenomena of this sort
98 Chapter V

are so idiosyncratic and unpredictable that, as Fillmore says, "They

appear to be among the things that, from a language learner's point of
view, are best learnt one piece at a time" (Fillmore 1984 a: 128).
Fillmore also discusses formulaic expressions, which he regards as
"The most striking kind of small issue in pragmatics" (Fillmore 1984 a:
128). Various languages have different patterns of such expressions with
different pragmatic functions. When translated into another language,
they do not fit the corresponding situational contexts. Thus, Japanese
has an elaborate set of formulaic utterances expressing apology, which
have no pragmatic equivalents in European languages. American English
has a number of formulaic expressions which do not convey information
but help to maintain conversation, the so-called conversational greasers,
while Yiddish appears to be particularly rich in curses, wishes, and
blessings as described by Matisoff (1979).
Fillmore provides a detailed pragmatic analysis of several English
formulaic expressions: It's not what you think. This hurts me more than it
hurts you. It takes one to know one. Present company excepted. I'll tell
you what, and You should talk. For each expression the analysis specifies
the setting, the anteceding event, speaker's attitude, function of utterance,
usage notes, prototype example, "creative" departures (if any), similar
formulas, and miscellaneous observations. For example, You should talk
is analysed in the following way:
Formula YOU SHOULD TALK
Setting A and Β are in conversation; Β is discussing C;
both A and Β understand that to say X of some-
body is to be critical
Antecedent event Β has just described C as being X
Speaker's attitude B's criticism is unjust, since Β too is X
Function of A chides Β for commenting on C's faults without
utterance recognizing his own
Usage notes common
Prototype example Β says, "C never has a kind word to say about
anyone." A says, "You should talk."
"Creative"
departure
Similar formulas YOU'RE A FINE ONE TO TALK/IT TAKES
ONE TO KNOW ONE/WHO ARE YOU TO
TALK7/LOOK WHO'S TALKING1/PEOPLE
WHO LIVE IN GLASS HOUSES SHOULDN'T
THROW STONES
 Various levels of linguistic analysis 99

Miscellaneous
observations
Use of this formula seems to include A's acknowledgment of the
aptness of B's description of C as X. "It takes one to know one" does
not include that.
(Fillmore 1984 a: 140)

Intricate and illuminating as they are, these analyses are not fully
contrastive since they do not provide explicit answers to such questions
as whether formulaic expressions in one language have any corresponding
formulas at all in another language, and whether in all these cases in
which such formulas are appropriate it would be possible to find an
equivalent situation in another culture. We shall return to this problem
and to some other problems which ensue from Fillmore's analysis at the
end of this chapter.
Oleksy's pragmatic contrastive study, like Riley's, hinges crucially on
the concept of speech act (Oleksy 1983). In order to perform a contrastive
analysis of pragmatically equivalent linguistic forms, Oleksy adopts a
definition of pragmatic equivalence, which he formulates as follows:

A linguistic expression Xt in is pragmatically equivalent to a linguistic

expression X 2 in L 2 if Xj and X 2 can be used in the performance of the
same SA (Speech Act) in L, and L 2 relative to the corresponding set
of pragmatic, contextual and socio-cultural factors (Oleksy 1983: 85).

Although it is not quite clear why Oleksy chooses to contrast pragmatic

factors with contextual and socio-cultural factors since one would have
thought that they all constitute extralinguistic settings in which linguistic
expressions are appropriately or inappropriately used, and hence that
they are all pragmatic, we take Oleksy's definition to be fundamentally
correct, particularly if we recall that the scope of pragmatics is by no
means a settled issue. However, correct as it is, Oleksy's definition,
consistent with the terminology that he adopts, defines pragmatic, con-
textual, and socio-cultural equivalence rather than merely pragmatic
equivalence. The issue seems to be of terminological nature, though it
does involve the scope of pragmatics.
Oleksy's analysis is based on two experiments involving native speakers
of Polish. In the first experiment the respondents (40 in all) were asked
to rank Polish sentences expressing request in the order of descending
deference. The results were compared with an analogous list of sentences
in English of which the Polish sentences were translations. The second
100 Chapter V

experiment, in which 44 persons participated, required decisions about

which of the eight expressions provided by the researcher they would use
to ask someone to close the window in various situational contexts (a
room in a dorm, classroom, living room in one's own appartment) and
with various addressees (colleague, pupil, parent). Both these experiments
supported Oleksy's basic claim that "Pragmatically equivalent expressions
across languages need not be equivalent formally" (Oleksy 1983: 91),
where the word "formally" is to be understood as referring to semanto-
syntax. Oleksy's experiment has thus confirmed our earlier observations
concerning semanto-syntactic dissimilarity of translational (or pragmatic)
equivalents (cf. Chapters II and VII).
Oleksy's contrastive pragmatic analysis lends further support to our
claim that pragmatically equivalent utterances are in fact optimum trans-
lations in a given linguistic and situational context.
The comparison of the three approaches to pragmatic contrastive
studies presented above shows that despite the common terminological
denominator, viz. the word pragmatic, they do not follow a common
contrastive procedure. Moreover, they deal with different fields, all three
outside the domain of traditional sentence-bound contrastive analyses.
Riley primarily deals with longer stretches of texts or with sequences of
sentences to discover how they are structured to form discourses in
appropriate situational contexts. Oleksy is mainly interested in cross-
language equivalence of utterances associated with identical speech acts.
Thus, they are both interested in what Fillmore calls "large facts" per-
taining to speech acts and the organization of discourse. Fillmore himself,
primarily dealing with "small facts", concentrates on the appropriateness
of particular linguistic forms in particular situations. Oleksy's analysis,
it might be added, also embraces "small facts" insofar as he examines
the appropriateness of linguistic forms expressing request relative to
pragmatic, contextual, and socio-cultural factors. It is evident that the
three approaches are by no means contradictory, but rather they are
complementary. Together they cover an extensive area of investigation
which constitutes pragmatic contrastive studies.
What does not ensue from the proposals is a common comparative
procedure. Oleksy's proposal is the most explicit one in this respect
inasmuch as he actually formulates certain claims concerning pragmatic
equivalence. Fillmore's proposal is the least explicit one in this respect
since comparative remarks are only made in passing in the presented
analyses. Fillmore takes for granted situational settings with which certain
linguistic forms are associated as a common platform of reference (tertium
 Various levels of linguistic analysis 101

comparationis) against which such forms are deemed to be equivalent: if

a linguistic expression in one language is associated with a situational
setting, and if a linguistic form in another language is associated with
the same setting, then these forms are presumably equivalent. Fillmore
does not consider cases in which a particular situational setting has no
match at all in another culture. Examples of such situations abound in
cross-language or rather cross-cultural studies. Linguistic forms associ-
ated with such situations constitute even "smaller facts" than those that
Fillmore considers. They are often very highly conventionalized and very
closely associated with specific cultural, ethnic, religious, etc. settings,
which may have no matches in the next culture. Consider such examples
as various formulaic expressions connected with greetings at Christmas
time, for which one finds no equivalents in either linguistic expressions
or situational settings in various non-Christian cultures. Consider, too,
various linguistic forms used to order various drinks in English pubs, for
which there are no equivalents in various "dry" states for want of
situational settings.
Fillmore's examples involve settings of a more universal character, not
culturally bound. Situations of this sort could be found in any culture,
at any time. Therefore, a viable and reasonably complete contrastive
study in this area ought to embrace the whole range of situations and
the related linguistic expressions, from the least to the most convention-
alized one. The following research procedure could be suggested:
I. Given a socio-cultural setting m (C m ) in a language i (CmLi), is
there an equivalent socio-cultural setting η (C n ) in language j (CnLj)? If
the answer is "no", note the socio-cultural contrast. If the answer is
"yes":
II. Is there a linguistic form (F q ) in Lj (FqLj) which is prototypically
associated with CnLj in a way similar to the way in which FpLi is
associated with CmLi? If the answer is "no", note the pragmatic contrast.
If the answer is "yes":
III. Is FqLj a semanto-syntactic equivalent of FpLi? If the answer is
"no", abandon the analysis. If the answer is "yes", proceed with semanto-
syntactic contrastive studies until you find contrast at some level of
analysis.
Before we provide some examples, let us consider the concept of
"prototypical association" It will be recalled that Riley's analysis revealed
the existence of various linguistic forms which can be associated with
various settings. Riley correctly observed that because there is no one-
102 Chapter V

to-one correspondence between forms and their functions, one can expect
a number of linguistic expressions to be associated with one setting/
function, and, conversely, the same function can have a number of
linguistic expressions. Oleksy's analysis has revealed that some subtle
factors, such as the degree of deference, determine which particular
linguistic expression of many possible ones is likely to occur in a specific
context rather than any possible other form. Incidentally, Oleksy's con-
clusions are in accord with my own earlier investigations concerning the
frequency of occurrence of various alternative forms in particular texts.
The issue was raised and discussed in connection with quantitative con-
trastive studies and will be resumed in Chapter X. Briefly speaking, those
forms which associate more readily with particular settings will be said
to be more prototypical than those that associate less readily with the
same settings. Generally, the most prototypical forms will be the most
frequent ones in particular texts and the most probable ones in specific
linguistic settings with which these forms can be associated.
These remarks pertain mainly to highly conventionalized linguistic
expressions for which fairly accurate predictions can be made as to the
contexts in which they appear. Less conventional forms, looser colloca-
tions of words and various innovations may be so unpredictable as to
the settings in which they appear that they will fall entirely outside the
scope of the present discussion and will have to be investigated in isolation
from these settings. Generally, most conventionalized forms are most
setting-dependent, i. e., given a form, it is possible to predict the setting
with which the form associates, and, conversely, given a setting, it is
possible, with a high degree of accuracy, to predict which linguistic forms
are appropriately associated with it. All linguistic forms seem to form a
cline from the most conventionalized ones to the least conventionalized
ones (the most free collocations). The least conventionalized ones are the
least setting-dependent, and hence they are the least predictable. They
can be successfully handled by a grammar alone, without recourse to
pragmatics or to socio-cultural conditions which attend their appropriate
use. In the case of such unconventionalized linguistic expressions, the
concept of prototypical association with a setting ceases to have any
relevance at all since the mutual relation between the two becomes
unpredictable. On the other hand, the relevance of the concept "proto-
typical association" increases with the increase in the conventionality of
a given linguistic form, particularly if other conventional expressions
which can be associated with the same setting are also available.
 Various levels of linguistic analysis 103

In view of all this, one can see that Fillmore's examples can be situated
somewhere in the middle of the scale of conventionality: while not being
totally novel (after all they are called "formulaic" expressions), they are
still not altogether situation/setting-bound, since they may, but do not
have to, be used in a given setting, and many of them have alternative
similar formulas. Further on the scale of conventionality, near its extreme,
are totally fixed and totally conventionalized formulaic expressions, per-
manently and inevitably associated with specific socio-cultural settings,
which are obligatorily used in certain situational contexts. It is mainly
among such expressions and such situations that one looks for examples
of socio-cultural contrasts. Linguistic expressions of this kind are so
highly conventionalized and so predictably associated with the appropri-
ate contexts that they deserve the epithet "ritualistic" Many of them,
though of course not all, are associated with situations which have no
match in other cultures. Here belong various greetings connected with
religious holidays and feasts which are not universally celebrated through-
out the world, commands and cheers connected with various games and
sports activities uttered both by participants and supporters, as well as
various ritualistic expressions during religious services and lay ceremonies,
such as Ego te absolvo Ί absolve you' the formula used at the end of
a Catholic confession or Niech sig swi§ci Pierwszy Maja 'Let the
Mayday be celebrated', the formula used by communist leaders at the
end of their Mayday speeches. Situations in which such ritualistic ex-
pressions are used have no parallel in other cultures and do not render
themselves to regular contrastive studies. So from the contrastive point
of view, one simply notes the socio-cultural contrast, which may have
linguistic repercussions, should the need to translate the respective for-
mula arise.
If, however, a parallel situation does occur in another culture, i. e., if
the answer to Query I is 'yes', one has to look for forms which are
prototypically associated with a particular situation. Here the range of
various typical and less typical situations, general and less general settings,
is very broad indeed. In many instances, an answer to Query II is not
available without thorough pragmatic research, preferably based on sta-
tistical data, since there is no a priori way of determining what, if anything,
is prototypically said in a given situation entrenched in a given culture.
Impressionistic observations suggest that Finns are much more reluctant
to open their mouths in a number of situations than, say, Italians in
analogous situations (a train compartment, a queue at a check-out
counter, etc.). The situations quoted by Fillmore, though less specific,
104 Chapter V

may likewise evoke different linguistic responses across cultures, including

no linguistic reaction of any sort.
Possible research along these lines could not ignore various socio-
linguistic factors which determine who says what to whom in a given
situation (Janicki 1985). Given a situation, a prototypical linguistic re-
sponse or its lack would be a resultant of such parameters as age, sex,
temperament, verbal skills, social conditions, and quite a few other traits
(cf. Preston 1979: 2ff.). Briefly speaking, in many instances, especially
those situated further away from the extreme end of the scale of conven-
tionality, whether and what one says in a given situation is to a large
extent determined by individual, idiosyncratic characteristics of the
speaker. But one can talk about certain tendencies towards using certain
forms in specific situations, and some of these tendencies may point to
using certain forms rather than others. Such forms can be considered to
be prototypical in those situations in which they are most frequently
used. In those cases we face the task of performing a routine contrastive
analysis of the pragmatically equivalent linguistic forms to establish
differences and similarities at various levels of linguistic structure and in
this way provide an answer to Query III. In the preponderant majority
of cases pragmatically equivalent linguistic forms will be semanto-syn-
tactically non-equivalent as the following list of Fillmore's formulaic
expressions and their pragmatic equivalents shows. In this list Fillmore's
formulaic expressions are followed first by their Polish semanto-syntactic
equivalent (cf. Chapters II and VII) and next by their pragmatic equiv-
alents as attested by a competent bilingual informant: 8

IT'S NOT WHAT YOU THINK.

To nie jest to, co myslisz.
TO ΝΙΕ ΤΑΚ.
THIS HURTS ME MORE THAN IT HURTS YOU.
To boli mnie bardziej niz ciebie.
(NO, NO, JUZ, JUZ.) TO TYLKO TROCHg POBOLI.
IT TAKES ONE TO KNOW ONE.
Potrzebny jest ktos (taki sam), by poznac kogos.
SW0J DO SWEGO PO SWOJE. (TRAFIL SW0J NA
SWEGO.)
- PRESENT COMPANY EXCEPTED -
- Ζ wylgczeniem obecnego tu towarzystwa —
- Ο Ο ΒECNYC Η ΝΙΕ Μ 0 WIM Υ -
 Various levels of linguistic analysis 105

I'LL TELL YOU WHAT (TELL YOU WHAT).

Powiem ci tak.
JUZ WIEM, CO ZROBIMY. (WIESZ CO?)
YOU SHOULD TALK.
Ty (wlasnie) powinienes mowic.
PATRZ, KTO M0WI. (TY SIE LEPIEJ NIE ODZYWAJ).
Even a cursory look at the pragmatic equivalents of the English
formulaic expressions reveals their semanto-syntactic non-equivalence: all
the Polish formulaic expressions represent different grammatical construc-
tions, though in some of them there is a certain degree of lexical similarity,
viz. hurts — poboli, present — obecni, talk — mowi. This kind of resem-
blance is, however, remote and highly unpredictable.
Chapter VI

Linguistic models and contrastive studies

Contrastive studies are performed in various linguistic models, ranging

from traditional grammars (Stanislawski 1950 — 1951; Kielski 1957;
Wojcik 1975; Gniadek 1979) through structural ( A g a r d - D i Pietro 1965;
Kufner 1962), generative-transformational (Stockwell et al. 1965; Fisiak
et al. 1978), relational grammar (de Geest 1979) to rather esoteric models,
such as stratificational grammar (Snook 1971), to mention just a small
fraction of contrastive studies conducted in various frameworks and
involving a large number of languages throughout the linguistic world.
For some purposes, especially when pedagogical practice is concerned,
the "traditional" approach is likely to be very rewarding:

Traditional foreign-language grammars make use of the contrastive

approach when they compare constructions or functions of the target
language with those of the source language. Such comparisons within
the framework of traditional grammar have turned out to be quite
useful for pedagogical purposes; however, they suffer from a general
defect: the traditional grammar on which they are based is not suffi-
ciently explicit to permit exact analyses (Nickel 1971: 4).

Structural contrastive studies, which flourished in the United States in

the early 1960s, from the theoretical point of view, are poorly motivated,
while generative transformational contrastive studies, theoretically better
founded, have limited practical uses.
Various authors have tried to restrict the class of linguistic models
providing a foundation for contrastive studies, notably Van Buren (1974),
Krzeszowski (1974), Lipmska (1975). The general conclusion that can be
drawn from these attempts is that not all models are equally suitable as
theoretical foundations of contrastive studies. The minimum requirement
that one can place on such models is that a model employed in the
description of LI should be compatible with the model employed in the
description of L2, since as Fisiak et al. have written:
108 Chapter VI

CS can be performed meaningfully only if the confronted languages

are presented in the framework of the same theory. One cannot take
Polish described within a structural theory and compare it with English
presented within a transformational theory. Such a contrastive com-
parison would be meaningless and, moreover, practically impossible
as each theory uses its own set of concepts and it would not be feasible
in most cases to establish similarities and differences holding between
linguistic elements. Furthermore, different theories vary as to the
number of facts they can account for (Fisiak et al. 1978: 13).
In the subsequent sections of this chapter, we shall take a closer look
at three approaches embracing various specific models, within which
most contrastive studies have been so far conducted, viz., "traditional",
structural, and generative-transformational, to see how they cope with
contrastive facts. We shall also try to evaluate their adequacy in terms
of the number of facts that they are capable of handling as well as in
terms of generalizations which they can grasp. In doing this, we shall
also note certain methodological problems which arise in the contrastive
practice and which are connected with the adoption of particular models.
In this way we shall provide grounds for the resolution of these problems
in the chapters which follow.

1. "Traditional" approaches

"Traditional" grammars are directly based on linguistic concepts worked

out by ancient Greek scholars. This tradition was later continued by
Roman scholars and throughout the following centuries provided a foun-
dation for all grammars of all languages connected with Western civili-
zation. 1 What characterizes these traditional approaches to the descrip-
tion of language is the assumed universality of linguistic concepts applied
in the description of specific languages. Consequently, any contrastive
study conducted within these traditional models would be at least im-
plicitly generalized (in Di Pietro's sense, cf. Chapter III) in that it would
relate specific phenomena in specific languages to some common set of
linguistic concepts, forming what is taken to be universal grammar. These
concepts employed in descriptive grammars of nearly all Indo-European
languages and many non-Indo-European ones, relate to ways of analysing
the basic grammatical unit, i. e., sentence. One type of analysis deals with
 Linguistic models and contrastive studies 109

the identification of parts of speech, such as nouns, verbs, adjectives, and

adverbs, as well as an assortment of function words, such as prepositions,
conjunctions, pronouns, particles, and interjections. Function words,
limited in number, are used in structural approaches to identify nouns,
verbs, adjectives, and adverbs, sometimes referred to jointly as parts of
speech (see Francis 1954: 231). However, in traditional grammars, these
words are defined in terms of a mixed batch of criteria, involving both
formal and semantic considerations. For example, nouns may be defined
as those words which typically denote persons, things and ideas (semantic
criteria) and typically assume such inflections as number and case. More-
over, they typically function as subjects and objects in sentences (formal
criteria). Similarly, verbs express actions and states and are typically
inflected for tense, mood, and person, functioning as predicates in sen-
tences. Adjectives express qualities and can be inflected for degree.
These definitions were not very precise and, in many instances, failed
as methods of identification. Many nouns express actions or states, for
example, race, birth, counting, proposal, knowledge, etc. On the other
hand, some nouns are not inflected for number, for example, sugar,
traffic, morality, etc. while in some languages, including English, they
are not inflected for case. 2
Similarly, not all verbs express actions or states, while in such languages
as Hopi or Chinese they are not inflected for tense. As a result of such
facts, the identification of various words as parts of speech, i.e., their
assignment to grammatical categories, was only partly motivated by
semantic and formal criteria, which often had to be supplemented by the
grammarian's intuition, presumably involving the knowledge of universal
grammar (cf. Wilkins' attempt at formulating the principles of universal
grammar dating back to 1668).
The fuzzy character of various language phenomena cannot be rec-
onciled with traditional, clear-cut categorization. Therefore, traditional
grammars necessarily contained lists of exceptions to numerous rules.
But while the rules could be formally and/or semantically motivated, the
exceptions remained a rather obscure area, for which motivation could
not be found. Recent advances in cognitive grammars and a new approach
to categorization (Lakoff 1986; Langacker 1987) offer a framework
whereby these fuzzy phenomena are accepted as a natural property of
language and treated as such in the description (see Chapter XI).
Another kind of analysis in the traditional framework concerned
"parsing" or identification of functions which various constituents per-
form in the sentence. Parsing is basically the recognition of the hier-
110 Chapter VI

archical structure of sentences. As a result of parsing, such functional

parts of sentences as subject, predicate, and various kinds of comple-
ments and modifiers are identified. This sort of analysis is sometimes
called "logical" inasmuch as subject and predicate were originally iden-
tified in terms of the question "What is being said about what?",
referring to the traditional distinction made by ancient logicians be-
tween the theme of the sentence, i.e., the person or thing being spoken
about, and the rheme, i.e. that part of the sentence which says or
predicates something about the theme. In spite of the fact that the
theme of the sentence cannot always be identified with its grammatical
subject, the analysis identifying various functional parts of the sentence
continued to be called "logical"
Contrastive studies based on traditional models were consistent with
the assumed universality of basic grammatical categories, which provided
a convenient tertium comparationis against which it was possible to state
differences in the realization of these universal categories in various
languages.
The comparisons themselves were also based on intuitions insofar as
grammarians depended on their knowledge of the compared languages
in deciding on what a particular item in LI should be compared with in
L2. The matching of equivalents was accomplished either by translating
the relevant constructions and sentences (in the case of comparisons of
sentences and constructions) or by identifying various categories of words
as equivalent across languages on the basis of their apparent functional
and semantic similarity. Although few grammars would be fully compar-
ative, a great number of grammars and textbooks of foreign languages
published both in Europe and in America would contain numerous
contrastive observations connected with an assortment of rather ran-
domly selected facts. 3
Below are some examples of such traditional contrastive studies. Let
us first look at a passage from French reference grammar by
J. E. Mansion, published in London and Edinburgh in 1928. On p. 53,
concerning the passive voice, the author writes:

English easily distinguishes between 'The door is shut' (completed

action, or state), and 'The door is being shut', by the use of the
continuous present, but as French lacks this construction, the present
and past descriptive passive of verbs denoting momentary action are
not often used unless the agent is mentioned. They are replaced by
 Linguistic models and contrastive studies 111

(a) The active voice with on as subject

On ouvre la porte. The door is (being) opened.
(b) The pronominal voice
La porte s'ouvre. The door opens.
It is permissible in English to make a dative or a prepositional object
the subject of the passive
John gave me an apple. I was given an apple by John.
People laughed at me. I was laughed at.
This construction is impossible in French. Note especially that indi-
rectly transitive verbs have no passive voice. Such sentences as
On repondit ä sa question. Son fils lui succedera.
On peut resister a cette tentation.
cannot be recast in the passive, except occasionally
One finds similar contrastive remarks throughout the entire grammar,
except, interestingly enough, in the section dealing with the French
sounds, which is consistently devoid of any contrastive observations.
Shorter German reading grammar by B. Q. Morgan and
F. W. Strothmann, published in New York and Boston in 1952, contains
a number of contrastive passages, among them the following one:
Most English nouns form the plural by adding -s to the singular.
(Modern English, however, still retains some exceptional plural forms
of a different kind: foot, feet; mouse, mice; sheep, sheep; ox, oxen;
child, children). In German only a few modern nouns of non-Germanic
origin form a plural in s, especially those which end in a vowel.
The authors proceed to give a detailed description of how plurals are
formed. The same book contains an interesting comparison on the use
of tenses in German and English, especially the formally similar "Perfect"
tenses, which in the two languages have contrasting uses (1952: 126 ff.).
Many traditional grammars contain passages devoted to comparisons
of speech sounds across languages. They are usually concerned with
differences and similarities in the articulation of particular speech sounds
with regard to place and manner of articulation, often containing addi-
tional information about phonetic contexts in which a specific articulatory
feature of a specific sound appears, or does not appear, in contrast with
a comparable situation in another language. Such studies anticipate later
subphonemic contrastive studies within structural phonology (see the
112 Chapter VI

next section) in that they deal with contextually determined variants of

phonemes rather than with comparisons of phonemic inventories alone.
For example, in The first French handbook for teachers in elementary
schools by Jack Kolbert and Harry Goldby, published in 1958 in Pittsburg,
pronunciation was introduced contrastively, and the appropriate chapter
begins with the following general statement:
French and English rarely correspond exactly. French has many sounds
which have no equivalents in English and vice versa (p. 4).
Talking about the articulation of vowels the authors observe that
French vowels are more sharply pronounced, more pure, more precise
than their English counterparts. They are free from a certain gliding
tendency that seems to elongate the sound of vowels in English (p. 5).
Such impressionistic and metaphorical descriptions of the differences
in the articulation seem to echo various comparative passages from early
English-French grammars as, for example, those by Guy Miege (1687)
or Peter Berault (1688). On the other hand, they also anticipate some
later analyses in that they provide various very detailed descriptions of
differences. The grammar by Kolbert and Goldby thus describes the
articulation of the French "a" in contrast to the English "a" (the authors
consider differences in which the letter "a" is articulated in the two
languages):
In most words "A" is a cross between the "A" sound of the English
word CAT, AS or CAN, and the A sound of English FATHER, closer,
however, to the "A" in Albert, Absent (Kolbert - Goldby 1958: 7).
They further observe that the French L "is the same as the L of the
English word LAKE" and advise the learner to "Avoid the English 'dark
Γ, especially in final position" (p. 23).
The examples which we have quoted continue a long tradition of
pedagogically oriented and motivated contrastive studies. Traditional
contrastive studies attempted to embrace a fairly large scope of linguistic
data from speech sounds through morphology, syntax, and lexicon to
some facts about the usage, i.e., the area which we would now call
pragmatics of language. The breadth of the scope must be considered as
one of the advantages of traditional contrastive studies. Among their
drawbacks is the lack of precise criteria of identifying various linguistic
categories and of comparability. These shortcomings are alleviated by the
fact that traditional contrastive studies employed a set of universal con-
 Linguistic models and contrastive studies 113

cepts in cross-language descriptions. On the one hand, the implicitly

assumed universality of grammar underlying linguistic descriptions made
such comparisons possible by providing the necessary tertium compara-
tionis, on the other hand, however, it provided grounds for criticism,
especially in structuralist quarters, for its lack of rigour and lack of
empirical motivation in justifying the introduction of specific linguistic
categories in specific languages (see the next section).
Both these objections have been overcome in the approach called
cognitive grammars, which seeks to reconcile the universality of certain
concepts with low-level idiosyncracies characterizing particular languages
(see Chapter XI).

2. Structural approaches
In this section, we shall follow the American version of structuralism,
primarily because American linguists rediscovered contrastive studies for
themselves in the times when structuralism was flourishing, i.e. in the
late 1940s and early 1950s, and secondly, because in comparison with
other versions, American structuralism, also called "distributionalism"
(Apresjan 1966: 64 ff.), was most radical in rejecting meaning as a possible
tool of linguistic analysis. This radicalism best reveals some weaknesses
of structural methods, both in the area of linguistic description and in
the area of contrastive studies.
American structuralism was moulded under the influence of two out-
standing linguists, Edward Sapir and Leonard Bloomfield, who in 1921
and 1933, respectively, published their major works, incidentally both
entitled Language. Their views were espoused by a generation of linguists
in America and some other parts of the world, including Europe. It was
Leonard Bloomfield who originally suggested that linguistic analysis
should be conducted in distributional terms (1933: 158 f f ) . This view
subsequently culminated in the works of Zellig Harris (1946, 1951) and
Charles C. Fries (1952).
One of the fundamental tenets of distributionalism was that every
language should be analysed and described in its own categories insofar
as every language employs different and unique grammatical means
(Fisiak 1975 b: 62). Consequently, according to structuralist principles, it
is impossible to describe all languages within the same universal frame-
work of grammatical categories, since every language has its own cate-
114 Chapter VI

gories, which can only be identified and described in terms of unique

structural signals operating in that language. According to Sapir

no logical scheme of the parts of speech — their number, nature, and

necessary confines — is of the slightest interest to the linguist. Each
language has its own scheme. Everything depends on the formal
demarcations which it recognizes (Sapir 1921: 119).

This point of view materialized in the structural methodology worked

out by Harris (1951) and was consistently implemented in the structural
description of American English (with the exclusion of phonology, to
which we shall return presently) by Fries (1952: 76 ff.). Fries attempted
to describe English solely in terms of structural properties of English
sentences. To emphasize the unique relevance of his description to English,
Fries disposed of the traditional grammatical terminology and replaced
it with numerical labels referring to specific classes of lexical words and
with letters of the English alphabet referring to specific groups of function
words. Thus, class 1 comprised those words which can fit the blank in
the following 'test-frame':

The was good

into which such words as concert, food, coffee, difference, privacy, com-
pany, clerk, husband, supervisor, woman, team, etc. could be inserted.

Words constituting class 2 fitted such frames as

The Class 1 good.

The Class 1 .Class 1.
The Class 1 .there.

as in

The concert was good.

The husband wanted coffee.
The boys walked there.

To class 2 belonged such words as be (in its various forms), seem,

sound, become, remember, suggest, etc.

Words of class 3 fitted the frames

The Class 1 Class 2.

The Class 1 Class 2 Class 3.
 Linguistic models and contrastive studies 115

as in
The concert was very good.
The good concert was good.
Here belonged such words as good, large, necessary, foreign, new,
empty, hard, best, lower, etc.
Words of class 4 were distinguished on the basis of their ability to fill
the appropriate positions in the frames
Class 3 Class 1 Class 2 Class 3 .
Class 1 Class 2 Class 1 .
Class 1 Class 2 .
as in
The good soup was rare there.
The man remembered the day clearly.
The boys went away.
To class 4 belonged such words as there, back, out, up, down, upward(s),
forward(s), upstairs, away, rapidly, eagerly, confidently, singly, safely, etc.
These classes cannot be fully equated with the traditional nouns, verbs,
adjectives and adverbs, respectively, since they do not always coincide
with what was included or excluded. For example, class 1 embraces such
words as father's and plains, traditionally recognized as parts of noun
phrases or nouns but also such words as red, traditionally recognized as
an adjective as well as a noun.
The procedure of identifying parts of speech through their structural
properties was further elaborated to include structural signals other than
the ability to occupy certain positions in "test-frames" W. Nelson Francis
(1954: 229) mentions five types of structural signals:
a) function word
b) word order
c) inflection
d) derivational contrast
e) suprasegmentale
Francis subdivides function words into Noun-determiners, Auxiliaries,
Qualifiers, Prepositions, Coordinators, Interrogators, and a few more.
They differ from parts of speech in that they constitute closed classes
and, as Francis maintains, "are largely devoid of lexical meaning" (Fran-
116 Chapter VI

cis 1954: 234). Although Francis retains traditional names of various

parts of speech, he also defines them in purely structural terms, involving
the five types of structural signals. Thus, for example, he defines nouns
as
a class of lexical words marked by their appearance following certain
noun-determining functions words, such as the, my, some, two; by their
use of two inflections, {-es} and {-'s}; by certain derivational suffixes;
by their appearance in certain positions; and occasionally by certain
superfixes of stress (Francis 1954: 252).
It is obvious that this sort of definition embraces only English nouns,
not nouns as a universal category. Therefore, in no way can it serve as a
platform of reference tertium comparationis for cross-language compari-
sons. Consequently, it turns out that although, according to structural
methods, parts of speech and sentence patterns are described indepen-
dently for every language and on the basis of form alone, these rigours
must be lifted if one is attempting to conduct a cross-language compar-
ison.
In his contrastive study of English and German, Kufner (1962) equates
nouns in English with nouns in German without providing any motivation
for juxtaposing these phenomena in the two languages, i.e. without
providing any tertia comparationis:
Nouns are marked by function in much the same way in both English
and German. They serve as subjects of verbs, as objects of verbs, as
objects of prepositions, as forms modified by adjectives and determin-
ers (Kufner 1962: 53).
Yet, the fact that nouns in English and nouns in German may function
as subjects, objects, etc. in both languages cannot be considered as a
reason for comparisons as long as those functional aspects remain un-
defined or as long as they are like other concepts in structural grammars,
defined in terms of some language-specific properties. Since Kufner does
not provide any general definitions, his comparative procedure is based
on intuitive juxtapositions. The structural definition of nouns in English
is completely irrelevant to German and vice versa; any structural defi-
nition of nouns in German would be useless when applied to the identi-
fication of English nouns. In fact, there is no guarantee that there will
be any structural resemblance between the compared languages, especially
if typologically remote languages are compared. If the structuralist prin-
ciples were consistently observed no comparison would be possible since
 Linguistic models and contrastive studies 117

both sentence patterns and parts of speech must be defined structurally

in particular languages in terms of language-specific structural signals
which operate only in the overall structure of the given language, or as
Fries puts it

the formal signals of structural meanings operate in a system that

is, that the items of form and arrangement have signalling significance
only [emphasis my own] as they are parts of patterns in a structural
whole (Fries 1952: 60).

It follows that if one defines nouns or, as Fries insists class 1 words,
in English in terms of such structural signals as plural inflections, deter-
miners the, a, some, the ability to occupy the position before and after
verbs, and possibly a set of derivational suffixes, such as -er, -ant, -encej
-ance, -ing, etc., a descriptive definition of this sort is valid only for
English and not other languages, where completely different signals
defining different classes of words may be at work. There is no way of
equating or comparing particular items thus defined insofar as the signals
involved in the respective definitions are language-specific and may be
radically different.
Ultimately, in structural contrastive studies, particular linguistic items
in various languages were matched for comparison not because they were
formally similar — the fact that they were and the extent to which they
were similar was to be determined by the contrastive studies but
because they meant more or less the same thing. Thus, meaning was
employed without ever being given proper recognition and without being
mentioned. 4 If one juxtaposes this knife and dieses Messer, one does so
not because of the form, which is obviously different in more than one
way, but because of the meaning which is common in both forms. This
commonness of meaning, intuitively recognised by anyone with the
knowledge of English and German, is due to the fact that the two
juxtaposed linguistic forms refer to the same sort of object, i.e. 'blade
with sharpened longitudinal edge fixed in handle, used as cutting instru-
ment or as weapon' In this way, meaning acquires critical importance in
establishing equivalence of linguistic forms across languages. This cannot
come as a surprise if one realizes that in order to compare any two items
it is first necessary to find tertia comparationis, i. e., those properties
which the compared items share, but which are outside the scope of
comparison itself. Juxtaposition of systems and constructions for com-
parisons on the basis of the common meaning was a concession in favour
118 Chapter VI

of semantics, which was never openly admitted but which was necessary
if similarities and differences in the form were to be found and described.
Let us look at such a contrastive study of the system of plural inflection
in English and German:
English and German nouns are marked as such by inflection for case
and number, but in drastically different ways. The English system can
be illustrated as follows:
Singular, general boy/'boi/ sheep wife man
possessive boy's/'boiz/ sheep's wife's man's
Plural, general boys/'boiz/ sheep wives men
possessive boys'/'boiz/ sheep's wives' men's
The vast majority of English nouns are inflected like boy and show
only two different phonemic shapes for these four grammatical forms.
A few words like sheep also show two different phonemic shapes, but
in a different distribution. Another small group, like wife, shows three
different shapes. Still another small group, like man, shows four dif-
ferent phonemic shapes. Only on the basis of this last group are we
justified in setting up four grammatically different forms.
The German system of inflection for case and number can be illustrated
by the following examples:
Sing. nom. Wagen Frau Knabe Name Mutter Vater Sohn
acc. Wagen Frau Knaben Namen Mutter Vater Sohn
dat. Wagen Frau Knaben Namen Mutter Vater Sohn(e)
gen. Wagens Frau Knaben Namens Mutter Vaters Sohn(e)s
Plural Wagen Frauen Knaben Namen Mütter Väter Söhne
dat. Wagen Frauen Knaben Namen Müttern Vätern Söhnen
Some nouns, like Wagen, Frau, Knabe, show only two different forms,
though in different distributions. Others, like Name and Mutter, show
three different forms, again in different distributions (the distribution
of the three forms is still different in Herz). Still others, like Vater,
show four different forms. The maximum number of different forms
is four or five, as illustrated by Sohn.
Students are quite familiar with inflection for number, since this
category is just as compulsory in English as it is in German. The
conflict which arises is not in the category as such, but rather in the
very different form it takes in the two languages. Aside from a few
forms like foot-feet, mouse-mice, man-men, ox-oxen and learned plurals
 Linguistic models and contrastive studies 119

like criterion-criteria, alumnus-alumni, English has only a single type

of noun plural formation, consisting of the following three shapes in
complementary distribution:
/-iz — s ~ -z/
In contrast to this, even aside from such forms as Doktor-Doktoren
(with stress shift) and such learned plurals as Atlas-Atlanten, Thema-
Themata, German has no less than six different plural types (Kufner
1962: 53-55).
The author proceeds to provide details about the distribution of the
plural inflections in the two languages.
This rather extensive quotation from Kufner shows how radically
different the plural forms are in such genetically (and in some respects
also typologically) close languages as English and German. Formal dif-
ferences are likely to be even more conspicuous in the case of languages
which are genetically and typologically more remote, such as Polish and
English, or English in contrast with other highly inflected languages. It
is evident that one could not possibly juxtapose English plurals with
German plurals (or plurals in any other language) on the basis of form,
which is so radically different. The juxtaposition is made on the tacit
assumption that one compares different forms across languages, because
they express the same meaning, i.e. plurality or "more-than-oneness",
and in doing so one looks for similarities and differences of form.
Meaning as tertium comparationis is no less essential in the comparison
of constructions or sentence patterns. Again they are juxtaposed by the
contrastive analyst due to the fact that he knows they mean more or less
the same in the two languages. Let us look at another example from
Kufner's contrastive study of English and German:
A speaker of German can negate any affirmative clause simply by the
insertion of nicht, e. g.
er wird mitgehen er wird nicht mitgehen
NEG
er arbeitet zuhause er arbeitet nicht zuhause
NEG
The difficulties which our students face when confronted with negative
clauses become apparent only when we contrast the German clauses
with their English equivalents (emphasis my own):
120 Chapter VI

he will come along he will not (won't) come along

NEG
he works at home he does not work at home
DO NEG
(Kufner 1962: 21)
The author then proceeds to discuss formal differences between Ger-
man and English negative sentences. Such comparisons of formal prop-
erties are possible only if the meaning of the compared constructions is
assumed to be identical. Unfortunately, this is not always the case since
meaning usually varies with form, and one cannot expect full cross-
language synonymity (semantic equivalence). Since in structural contras-
tive studies no principles of meaning-based juxtapositions were ever
formulated, the juxtapositions were based on intuitive judgments of
analysts, even though the comparisons proper purported to be rigorous.
As in the case of grammatical comparisons, structural contrastive
studies of speech sounds must include both paradigmatic and syntagmatic
aspects. The paradigmatic analysis results in finding systems and subsys-
tems of phonemes in both languages and in listing their relevant and
redundant features. The syntagmatic analysis embraces descriptions of
clusters, diphthongs, and other linear combinations of sounds.
According to Stockwell — Bowen (1965: 8), before comparing pronun-
ciation in two languages, the analyst must answer four basic questions
about each language: (1) What are the phonemic contrasts? (2) What are
the allophones of each phoneme, and to what environments are the
allophones restricted? (3) What is the distribution of each phoneme? (4)
What is the frequency of each phonemic contrast?
Structural phonological contrastive studies also raise some methodo-
logical problems, which mainly stem from inadequacies of structural
theories themselves. In structural phonologies two basic approaches to
the principal phonological unit, the phoneme, can be distinguished. In
one tradition, a phoneme is defined as a group of sounds which are
phonetically similar (by sharing at least one relevant feature) and are in
contrast with all other sounds in the system. Particular members of
phonemes are either mutually exclusive, which means that they never
appear in the same contexts, or mutually indifferent, which means that
they do appear in the same contexts without bringing about contrasts
(cf. Jassem 1954). Mutually exclusive variants of phonemes are sometimes
called allophones, while mutually indifferent ones are called free variants
or variphones (Krzeszowski 1968). For example, the English phoneme
 Linguistic models and contrastive studies 121

j\j lateral has at least four allophones which are contextually

conditioned:

1. The so-called "clear 1" [1], appearing in word and syllable initial
position, as in lip and belate.
2. The so-called "dark 1" [1], appearing after vowels at the end of
syllables and words, or after vowels before consonants, or as a syllabic
consonant, as in well, melt, and kettle.
3. The devoiced 1 — [1], appearing after voiceless consonants, as in play.
4. The dental 1 — [1], appearing before dental consonants as in health.

All these allophones of j\j are lateral, which is the relevant feature
they all share and which accounts for the 'phonetic similarity'

In the contexts in which the "dark 1" appears, there occur its free
variants: [L], pronounced with the tip of the tongue lowered to the lower
teeth and [ui], the back u n r o u n d e d vowel. The four allophones of /l/
appear in

/ like to play football for health

"clear" "devoiced" " d a r k " "dental"
(for further details see Krzeszowski 1968: 29).

The view which defined a phoneme as a group of sounds predominated

in American structuralism (Pike 1947; Twaddell 1963; Francis 1954;
Gleason 1955), and in Europe it was promulgated by Jones (1950) and
Jassem (1954).
The other tradition, going back to Jan Baudouin de Courtenay, the
founder of the concept of phoneme, interprets it as a bundle of distinctive
features, which do not carry meaning but are capable of bringing about
changes in meaning. Though based on articulatory parameters, distinctive
features can be correlated with acoustic parameters. This tradition flour-
ished in the Prague School of Linguistics (Trubetzkoy, Karcewski, Ja-
kobson) and was transferred to America by R o m a n Jakobson, where it
was a harbinger of numerous variants of generative phonology.
The difference between the two traditions is largely a matter of focus
either on segments or on features, but the two approaches are mutually
complementary, and no adequate phonological analysis is possible with-
out employing both concepts. As Fisiak writes, "Segments are composed
of features and have the form of matrices" (Fisiak et al. 1978: 226). Thus,
segment-based phonological contrastive studies will require at least the
following steps: (1) establishing the inventories of phonemes in the two
122 Chapter VI

languages; (2) establishing their equivalence (juxtaposition); (3) compar-

ison proper. The three steps are analogous to those in grammatical
contrastive studies. However, comparison proper of phonemes usually
leads to further, more delicate analyses at the subphonemic level.
Suppose that a structural analysis of the sound system of Modern
Standard Educated Southern English (often referred to as "Received
Standard" or "Received Pronunciation") reveals the presence of the
following 18 consonant phonemes and their relevant (distinctive)
features 5 :
/p/ stop labial strong
/bj stop labial weak
/t/ stop alveolar strong
jdj stop alveolar weak
jkj stop velar strong
jgj stop velar weak
/f/ fricative labial strong
/v/ fricative labial weak
/Θ/ fricative dental strong
16/ fricative dental weak
/s/ sibilant alveolar strong
jzj sibilant alveolar weak
/J/ sibilant palatal weak
/3/ sibilant palatal weak
/m/ nasal labial
/n/ nasal alveolar
/η/ nasal velar
j\j lateral

A structural analysis might establish the following consonant pho-

nemes in Polish, with the relevant features indicated 6 :
/p/ stop labial voiceless
/b/ stop labial voiced
/t/ stop dental voiceless
jdj stop dental voiced
jzj stop palatal voiceless
/j/ stop palatal voiced
jkj stop velar voiceless
jgj stop velar voiced
jtsj affricate dental voiceless
 Linguistic models and contrastive studies 123

/dz/ affricate dental voiced

/tf/ affricate alveolar voiceless
/<%/ affricate alveolar voiced
/te/ affricate palatal voiceless
/φ/ affricate palatal voiced
/f/ fricative labial voiceless
j\j fricative labial voiced
/x/ fricative velar
/s/ sibilant dental voiceless
jzj sibilant dental voiced
HI sibilant alveolar voiceless
/3/ sibilant alveolar voiced
/e/ sibilant palatal voiceless
jzj sibilant palatal voiced
/m/ nasal labial
/n/ nasal dental
/η/ nasal palatal
/ji/ nasal velar
/l/ lateral
/r/ trill

The material presented and described above must now undergo jux-
taposition. Looking at the two systems, one is struck by the quantitative
difference: the Polish system of consonants outnumbers the English
system by 11 phonemes, i.e., by about 30 percent. This finding itself,
when related to the number of vowel phonemes in the two languages and
to the ways in which vowels and consonants are distributed in English
and Polish texts, may lead to some interesting observations concerning
the overall sound effects evoked by texts uttered in the two languages. 7
On the Polish side, one observes a large number of affricates and sibilants
articulated in the dental and palatal area. These sounds have no match
in the English system. It is this group that mainly makes up the large
number of consonants in Polish and is responsible for the characteristic
hissing-hushing noises accompanying the articulation of Polish texts.
A comparison of particular groups of phonemes and individual pho-
nemes inevitably involves the problem of equivalence, which practically
manifests itself in the usual question "What to compare with what and
why?" There are numerous parameters (immediately relevant tertia corn-
par ationis, as we called them in Chapter II), on which one can base the
comparison. The ultimately relevant tertium comparationis is substantive
124 Chapter VI

in that it refers to material, phonetic substance in which particular

phonemes are moulded. For immediately relevant tertia comparationis,
one, most frequently applied option, is to base the comparison on artic-
ulatory (rather than acoustic or auditory) substance. The very fact that
consonants are juxtaposed has articulatory motivation since all conso-
nants in both Polish and English are articulated with at least occlusion
of the vocal tract. 8 However, comparing specific phonemes across lan-
guages, one requires more detailed descriptions, which involve relevant
features as more detailed tertia comparationis. Thus, among consonants
one can distinguish among stops, fricatives, sibilants, and affricates, each
group being characterized by a distinct way of constricting the vocal
tract. In Chapter II, we outlined a method of comparing sounds across
languages, whereby one evokes more and more specific immediately
relevant tertia comparationis in order to compare equivalent groups of
sounds. Eventually, one reaches individual sounds about which one can
make comparative statements. The procedure can be tabulated as in
Table 1.

Table 1. Comparing sounds across languages

Tertia comparationis Sounds

English consonants Polish consonants

narrowing:
at least occlusion
contact ρ t k ρ tc k
b d g b dj g
ts tj" te
dz d3 dz
release:
plosive ρ t k ρ tc k
b d g b dj g
strength/voicing:

strong/voiceless ρ t k ρ t c k

place:
labial ρ ρ
dental t
alveolar
palatal c
velar k k

Applying successive more and more specific immediately relevant tertia

comparationis and tabulating the results as in Table 1 (juxtaposition), it
 Linguistic models and contrastive studies 125

is possible to make the following contrastive observations (comparison

proper):
1. Polish has 14 stops, English has 6 stops.
2. Among the Polish stops, there are 6 stops with an affricative rather
than plosive release. English has no affricates, at least under the
analysis adopted here. 9
3. Polish has 2 stops articulated in the (post)palatal area. They have no
match in English.
4. The English /t/ is alveolar while the Polish /t/ is dental. Considering
this difference one can doubt whether these two phonemes should be
considered as equivalent.
5. The English jkj and /g/ and the Polish jkj and /g/ are articulated in
the same area.
6. The English stops /p, t, k/ contrast with /b, d, g/ by virtue of the
strength of articulation. The Polish stops /p, t, k/ contrast with /b, d,
g/ by virtue of the feature [ +voice].
The fact that certain groups of phonemes and certain individual
phonemes are comparable because they share certain relevant features
does not imply similarity of redundant features as they account for
allophonic variation. A complete contrastive study of two sound systems
must embrace these subphonemic differences. A subphonemic contrastive
study will reveal, among other things, the fact that English strong stops
are aspirated when they appear in front of stressed vowels, while the
Polish voiceless stops are not aspirated. On the other hand, the Polish
stops are palatalized before [i] and [j]. Various other redundant features
such as retraction, glottalization, devoicing, voicing, fronting, etc. have
to be considered and used as immediately relevant tertia comparationis
in a complete contrastive study of sound systems (for details see Krze-
szowski 1970; Kopczynski 1968, 1973; Suomi 1980).
The feature-focused approach in phonological analysis, developed in
the works of J a k o b s o n - H a l l e (1956), Halle (1964), Jassem (1962, 1966),
Anderson (1974), and many others, is based on the assumption that every
language draws a certain number of features from a universal stock of
phonetic features as components of sound segments. Some of these
features in a given language are distinctive, and they account for the fact
that a particular segment is different from all other segments in that
language. Such segments constitute representation at the phonological
level. Phonetic segments, however, have fully specified matrices of fea-
tures, not restricted to distinctive features alone. All these features con-
126 Chapter VI

stitute a convenient tertium comparationis since, as Fisiak et al. (1978)

rightly observe:
Features are easily comparable, because they come from the universal
stock, hence
E. [ + c o r ] = P. [ + cor].
(Fisiak et al. 1978: 225).
Consequently, they base their contrastive study of Polish and English
segments on 19 features constituting the tertium comparationis, which
allows them to compare vowels, consonants, diphthongs, and supraseg-
mental phenomena against a substantive tertium comparationis (for details
see Fisiak et al. 1978: 225 ff.). The tertium comparationis allows them to
perform a detailed phonological contrastive study of the two languages,
including the comparison of vowels, diphthongs, consonants, and supra-
segmental phenomena against one, unified tertium comparationis.
It seems that this approach in structural and generative phonology
occasions relatively fewer problems than the approach through the allo-
phonic analysis, as it does not force the analyst to make various awkward
decisions about how to categorize a particular sound and how to establish
equivalence. Specifically, the approach in question does not involve the
notorious problem of the so-called phonemic overlap, which results from
the fact that one speech sound may represent two phonemes, as is the
case with the English flap [L], either as an allophone of /r/ as in the word
right or as an allophone of /t/ in a possible rendering of the word water.
Such cases practically rule out the possibility of finding cross-language
equivalents on the phonetic level if a phonological tertium comparationis
is adopted. The feature-focused approach, by ignoring phoneme mem-
bership, handles such cases in terms of identity or non-identity of matrices
of features at the phonetic level. The difference between the English flap
/r/ and the Polish trill /r/ as captured by Fisiak's analysis is expressed in
terms of one feature from the universal stock: the English jxj is [ — ante-
rior], while the Polish /r/ is [ +anterior] because of its more distinctly
consonantal character. (Anterior sounds are articulated with at least
obstruction — cf. Chomsky— Halle 1968: 304.) True as this observation
may be, it is, unfortunately, inadequate since it does not point to the
most conspicuous articulatory difference between the compared sounds,
viz. that the English /r/ is a flap while the Polish jxj is a trill. Neither
feature is listed among the stock of "universal" features on which Fisiak
et al. base their analysis. This example demonstrates a fundamental
 Linguistic models and contrastive studies 127

weakness of the feature-based approach: it cannot adequately describe

language-specific idiosyncracies, and it cannot, in effect, be used to
describe all possible phonetic differences which exist across languages
insofar as it is constrained by the strait-jacket of the 19 (or whatever
number, for that matter) most typical features. Therefore, the picture of
phonetic differences that one gets from such an analysis is correct but
grossly oversimplified. Specifically, approaches of this sort, no matter
how refined (as, for example, Rubach 1983), are unable to describe with
sufficient detail either the articulatory basis, critically important in the
production of "foreign" sounds (cf. Ozga 1976: 61), nor perceptual
characteristics of sounds, essential in correct sound discrimination.
These limitations, which in equal measure concern both structural and
generative approaches to analyses of speech sounds, are extensively dis-
cussed by Kohler (1971), who concludes his critical survey with the
conclusion which anticipates later developments in cognitive linguistics:

To put contrastive studies and their practical applications in language

teaching on a better foundation than taxonomic phonology we cannot
substitute generative phonology, we do not need an ideal speaker-
hearer, but a real speaker and a real hearer. And to achieve this
linguists need the help of psychologists; purely linguistic operations
can only give a necessary and very useful first approximation; for
further work the concepts of phoneme, phonological system, langue,
parole can only be a hindrance (Kohler 1971: 87 — 88).

3. Generative transformational approaches

There are many versions of generative grammar, and contrastive studies

can be conducted in each version, even if criteria of comparability are
not equally clear in each of these versions. Most generative transforma-
tional grammars share the basic premises that (a) rules are the central
theoretical concept in the grammar and (b) transformational rules mediate
between abstract 'underlying' representations and their surface structure
representations. 10 Some versions claim that the representation underlying
the surface structure of various sentences in various languages is semantic
and universal. Such a universal "base" would provide an excellent tertium
comparationis for all kinds of grammatical contrastive studies. However,
other versions of generative transformational grammar, including the
128 Chapter VI

earliest model as described by Chomsky (1957), do not make such daring

claims. Nevertheless, the structures on which transformations operate in
this early model, i. e. basic sentence patterns, seem to be much more
similar than the structures which result from the operation of transfor-
mations, generally more diversified in many languages.
Before we discuss a few specific variants of generative transformational
grammar as they can be used as a background for contrastive studies, let
us briefly introduce the concept of "generative grammar" regardless of
whether it contains transformations among its rules.
A generative grammar is a device or a set of rules which explicitly
enumerates those combinations of linguistic forms which belong to a
particular language and at the same time assigns structural descriptions
to those forms. Most generative grammars enumerate sentences rather
than longer stretches of texts (see, however, Van Dijk 1972). Such gram-
mars do not produce linguistic forms of particular languages since they
are not speaker-oriented but neutral between speaker and hearer. They
are conceived as models of ideal speaker-hearer competence rather than
as grammars of performance as manifested in speaking or hearing. The
principles of generative grammars were made widely available in Chom-
sky (1957) and gained enormous acclaim in the first years following its
publication. Chomsky (1957) defines a natural language, such as English,
Chinese, etc., as "a set (finite or infinite) of sentences, each finite in length
and constructed out of a finite set of elements" (Chomsky 1957: 13). The
infiniteness of sentences in every natural language (in contrast with some
artificial languages of logic or mathematics) is connected with the fact
that there are no limits on the length of sentences: every sentence can be
lengthened, in principle, ad infinitum, as in Gejf knows that Mary knows
that..., etc., although every specific sentence is finite in length. However,
the set of rules required to generate the infinite set of sentences must be
finite. This is accomplished by incorporating the so-called recursive rules
into the grammar. These rules account for those elements of sentences
which are structurally identical (which recur). In the quoted example,
what can be added to the sentence could be described by the same rules
which are required to describe the first part of the sentence.
Chomsky further claimed that an optimal generative grammar of a
natural language must contain transformational rules (T-rules) to handle
certain linguistic phenomena which cannot be as effectively and as ele-
gantly handled by phrase-structure rules (PS-rules). In his early model,
PS-rules generated kernel sentences, i.e., simple, declarative, affirmative
sentences, consisting of no more than one finite verb, its subject, and
 Linguistic models and contrastive studies 129

possibly its object. Such kernel sentences were then transformed into all
kinds of sentences, including passive sentences, interrogative sentences,
negative sentences, as well as complex and compound sentences.
The early model of transformational grammar provided a framework
within which numerous contrastive studies were conducted (e.g., Bross
1962; Borkowski - Micklesen 1963; Dingwall 1964 b; Marton 1968 a and
b, 1970). The methodological problems inherent in such studies were
similar to those encountered in structural contrastive studies; the problems
were, however, compounded by the presence of T-rules in the grammars
of the compared languages, which necessitated the extension of the
concept of equivalence to cover rules.
In order to see what kind of problems arise when one adopts early
transformational grammar models as a framework for contrastive studies,
let us consider PS-rules suggested by Chomsky (1957) for English and
PS-rules suggested by Polanski (1966) for Polish:
English PS-rules (after Chomsky 1957: 111):
(1) Sentence —> N P + VP
(2) VP —> Verb + N P
iNPs,n g ]
(3) N P -
(.NPpiurJ
(4) NP sing —> Τ + Ν + 0
(5) NPplur—» Τ + Ν + S'
(6) Τ — the
(7) Ν — man, ball, etc.
(8) Verb — Aux + V
(9) V —> hit, take, walk, read, etc.
(10) Aux — C(M)(have 4- en)(be + ing)
(11) Μ —> will, can, may, shall, must
(12) fpres

Η [past
(Note: Rule (12) is not explicitly formulated on p. I l l , but it is referred
to throughout the book.)
Polish PS-rules (after Polahski 1971: 199):
(1) Zdanie —• GN n o m + G W
(2) GN n o m Przyrn nom + R n o m
(3) GW —• C + GN a c c
130 Chapter VI

(4) GN a c c —> R acc

(5) Przym —• znajomy, maly
(6) Rnom —*• chlopiec, uczen
(7) C —• widzi, wola, lubi
(8) Race —• kolegg, ojca.
In these rules
Zdanie = Sentence
GN n o m = Noun Phrase in the nominative
GW Verb Phrase
Przym nom = Adjective in the nominative
Rnom = Noun in the nominative
Racc = Noun in the accusative
The sets of rules such as those above cannot easily be compared if
only because they do not represent the same degree of accuracy. For
example, the Polish set does not provide any information about tenses,
nor does it mention modal verbs. Moreover, both these grammars can
be further refined to account for a greater number of sentence types. For
example, the English rules do not generate sentences without objects or
sentences with adverbs, such as John smiled, Mary danced yesterday, etc.
The Polish rules have analogous limitations. In addition, it is unable to
generate sentences with objects in cases other than the accusative as in
Jan pomaga dzieciom (dative) 'Jan helps (to) the children' Such trivial
inadequacies can be easily eliminated as new rules are added to both
grammars to make them generate whatever number of sentence types are
used in a particular language. As a result, both the sets of PS-rules can
be brought to an approximately comparable level of refinement.
Less trivially, even if reasonably complete PS-rules for the languages
to be compared were formulated, one would still confront formidable
problems connected with comparability. Since all the rules concern only
the syntactic structure of sentences, there is no principled way of estab-
lishing the necessary tertium comparationis, insofar as no semantic con-
siderations ever enter into play in the PS-rules formulated in Chomsky's
early model. It will be remembered that in Chapter III we argued that
only a semantic tertium comparationis in syntactic contrastive studies is
viable. For this reason, no contrastive studies of PS-rules have ever been
successfully performed (cf. Dingwall 1964 b). Instead contrastive analysts
compare basic sentence patterns in a fashion resembling structural con-
trastive studies (Stockwell et al. 1965; Lipinska 1972, 1973; Fisiak et al.
1978). Particular basic sentence patterns across languages are brought
 Linguistic models and contrastive studies 131

together for comparison (juxtaposition) in an intuitive manner on the

basis of the judgments of the analyst himself, who associates particular
patterns across languages on the basis of either their formal similarity
(again a vicious circle since similarity is to be revealed by the contrastive
studies and cannot be used as tertium comparationis) or by means of their
semantic similarity, never explicitly acknowledged.
Thus, the problems with comparability which arise in connection with
comparisons of basic sentence patterns are identical to those that are
inherent in structural contrastive studies. Stockwell et al. (1965: 32)
distinguish six basic sentence patterns in English and Spanish. They
compare them mainly on the basis of formal resemblance (committing
the error of circularity). They resort to translation equivalence — loosely
connected with semantic equivalence (cf. Chapter III) — only in those
instances when structural similarities are not matched by semantic simi-
larity or similarity of usage. Disparities of this kind occur mainly in the
area of verbal complementation. For example, English sentences:

Ε (a) I wanted him to come

(b) I asked him to come
containing verbs of communication can only be rendered in Spanish with
two clauses, each containing its own finite verb:
S (a) Queria que el viniera
(b) Le pedi que viniera
On the other hand, a sentence like
Ε (c) I beg you not to talk any more about politics,
which contains a verb of suasion, can be rendered in Spanish by means
of either a complement construction or a clausal one:
S (c) Les ruego que no discutan mäs de politica
Les ruego no discutir mas de politica
However, a construction such as:
Ε I said to be sure
and its Spanish equivalent:
S Dije estar seguro Ί said I was sure'
express completely different meanings. Stockwell et al. (1965: 40) quote
this example to show that although similar patterns generally express
132 Chapter VI

similar meanings, in specific instances structurally different constructions

are required.
The examples quoted above demonstrate the lack of consistency in
deciding about criteria of comparability. Like in structural syntactic
contrastive studies, many basic sentence patterns are juxtaposed for
comparison because they resemble each other formally and in this way
what should result from the comparison proper, i.e. the statement of
formal similarities and differences, is anticipated in juxtaposition, as it
illegitimately serves as tertium comparationis. The rare occasions when
the semantic tertium comparationis is referred to are confined only to
those instances when there is a discrepancy between meaning and form
across languages. Such discrepancies, too, should be stated as results of
comparison proper. Thus, what characterizes both structural contrastive
studies and generative transformational contrastive studies is the confu-
sion between the principles of juxtaposition and the results of comparison
proper. There are, however, further problems and limitations inherent in
transformationally based contrastive studies, to which we shall now turn
attention.
A typical contrastive study, within the generative transformational
framework, usually also contains a detailed description of particular
sentence constituents in the compared languages, often described in terms
of PS-rules. In Stockwell et al. (1965: 64 ff.) one finds descriptions and
comparisons of noun phrases and their constituents, verb phrases and
their constituents, determiners, etc. For instance, the rule expanding N P
as suggested by Stockwell et al. is identical in both languages:
N P —> D + Ν + N U M
where D — determiner, Ν — noun, N U M — number.
The rules rewriting N U M are also identical:

NUM

The rules rewriting determiners are different in the two languages:

S:
 Linguistic models and contrastive Studie. 133

This is as far as the analysis goes within the purely generative frame-
work. In fact, what we are dealing with here is not a comparison of rules
but rather constructions generated by those rules. The rules are merely a
convenient algebraic device to describe the constituent structure of de-
terminers in both languages. Unfortunately, such an analysis does not
reveal much. Therefore, the authors proceed to analyse the relevant
material in a more detailed way, but in doing so they abandon the
generative framework and adopt a combination of traditional and struc-
tural principles, involving comparisons of systems and constructions
rather than of rules. They juxtapose the articles in the two languages in
the following table:

Definite Articles

SPANISH ENGLISH

singular plural (singular or plural)

before before
a consonant a vowel

masc. el los [di] [diy]

fem. la (el) las
neut. lo

Some obvious differences concerning the gender and number distinc-

tions in Spanish and their absence in English can be seen immediately.
There are, however, other differences which can only be revealed when
constructions in which these systems are employed are compared. In
specific constructions, various articles in Spanish correspond to various
articles in English and vice versa. The comparison of systems proves to
be totally inadequate as one faces the following equivalence possibilities,
analogous to those encountered in the comparison of pronouns (see
Chapter IV):

Ε S

(1) the el
(2) a un
(3) 0 0
134 Chapter VI

Ε S
(4) the un
(5) the 0
(6) a el
(7) a 0
(8) 0 el
(9) 0 un

The table shows that in six out of nine cases there is a deviation from
"systematic" correspondence the — el, a — un, 0 — 0. In actual equivalent
sentences, considerable discrepancies occur, and they can be accounted
for only in terms of conditions of usage. Stockwell et al. state them in a
traditional, nonformal way. For example, (7) is described thus:

Pattern (7) above is that in which English has a matching Spanish 0.

This contrast occurs most frequently after a linking verb when the
following noun shows mere identification, as in the following examples:

He's a philosopher Es filosofo

He's a friend of mine Es amigo mio
He's a relative of the chief Es pariente del jefe, etc.
(Stockwell et al. 1965: 68).
The example concerning the determiner in English and Spanish dem-
onstrates the inadequacy of a contrastive study limited to a study of PS-
rules generating the constituent called Determiner. To be complete, a
contrastive study of determiners must be expanded to cover both the
study of the entire systems and, moreover, the study of ways in which
elements of those systems are used in actual utterances. This is so because
the PS-rules, no matter how detailed, do not provide access to the nature
of paradigmatic relations between particular elements of the systems.
Specifically, the rules provide no clues as to when elements of the systems
can be used and to what choices are appropriate in those cases in which
rules contain various options. Such is the case with the rule rewriting
Determiner in Spanish and English, where the indication "in specified
environments" cannot be made a working part of the rule without
reference to extralinguistic situations attending the specific utterances
rather than to intralinguistic (textual) factors.
In contrastive studies based on generative transformational grammars,
comparisons of various transformations constitute an essential part of
these studies. In contrastive studies based on early models of transfor-
 Linguistic models and contrastive studies 135

mational grammar (Chomsky 1957), T-rules were used to account for

structural relationships between kernel sentences and transforms. Kernel
sentences, i. e., sentences generated by PS-rules and obligatory transfor-
mations, were simple, declarative, affirmative sentences in the active voice.
Interrogative, negative, passive, complex, and compound sentences were
derived through T-rules from kernel sentences.
Expectedly, T-rules operating in various languages displayed differ-
ences and similarities and in this way yielded themselves to contrastive
studies. In short, they were assumed to be comparable (cf. Dingwall 1964:
158 — 159). According to Marton, two T-rules in two languages are
essentially identical if they "operate on two congruent structures in the
same way and consequently result in congruent transforms" (Marton
1968 b: 59). Congruent structures (strings) are those structures (strings)
which underlie congruent sentences. Congruent sentences, in turn, are
defined as sentences in two languages, which are semantically equivalent
and consist of equivalent items belonging to the same word classes and
having the same syntactic functions in the respective sentences. An ad-
ditional requirement is that in congruent sentences all constituents should
appear in the same linear order (cf. Marton 1968 b: 56 — 57; Krzeszowski
1967: 37). Although articles and auxiliaries are disregarded in this defi-
nition of congruence, there still remain considerable problems with other
function words and grammatical morphemes. Marton quotes the follow-
ing pair of strings as congruent in Polish and English:

Pol: # d z i e c k o + Neut + Sing + 3rd Pers + Past

4- Perf + isc — do — kino + Neut + Sing 4- G e n i t #
Eng: # the — child + 02 — Past + go — to — the — cinema + 02 #
which underlie two presumably congruent sentences:
Pol: Dziecko poszlo do kina.
Eng: The child went to the cinema.
The quoted strings do not consist of the same number of identical
elements. Specifically, in Polish, but not in English, there appear forma-
tives representing gender, third person, aspect, and case. On the other
hand, the English string contains articles, which have no equivalents in
Polish. Thus, the two strings can be considered as congruent only if all
function words and grammatical morphemes are disregarded. The defi-
nition of congruent strings would then have to be reformulated as follows:
Congruent strings are such strings which consist of the same number of
lexical words, representing equivalent grammatical categories, arranged
136 Chapter VI

in the same linear order and underlying semantically equivalent sentences.

By virtue of this reformulated definition, the two strings and the two
sentences quoted above could indeed be recognised as congruent.
Having defined congruent strings and sentences in the way suggested
above, one can accept Marton's definition of "identical" transformations
as those that "operate on two congruent structures in the same way and
consequently result in congruent transforms" (Marton 1970: 59). As an
example, consider one of Marton's cases, viz., the transformation which
derives attributive adjectives from nouns. In both Polish and English it
adds an adjective forming suffix to a noun. A variant of this transfor-
mation derives the adjectives in question from kernel strings Ν has N'
and Ν ma Νrespectively, and this transformation is identical (congru-
ent?) in both languages in that it derives congruent transforms from
congruent structures:

Eng: the peasant has a beard => the bearded peasant

Pol: chlop ma brodg => brodaty chlop
Marton also distinguishes "similar" transformations across languages.
Like identical transformations, similar T-rules operate on congruent
structures but do not yield congruent transforms. Thus, the Polish ex-
pressions niebieskooka dziewczyna, siwowlosy mgzczyzna, jednooki zebrak
and czteromotorowy samolot, according to Marton, are not fully congru-
ent with their English equivalents a blue-eyed girl, a white-haired man, a
one-eyed beggar, a four-engined aircraft because of the differences in the
distribution of stresses (one primary stress in Polish against two stresses
in English, compound word in Polish against a hyphenated word in
English).11 Still, because they are derived from congruent structures, they
involve what Marton calls "similar" transformations.
Marton's proposal is an important step in making such fundamental
concepts in contrastive studies as equivalence and congruence more
precise. Within the adopted framework of the early transformational
grammar it was impossible to make these notions fully explicit. Later
developments in linguistic theory led to the reformulation of the place of
semantics in the model and rendered it possible to couch the concepts of
equivalence and congruence in more rigorous terms (see Chapter VIII).
In 1965, Chomsky presented a new version of his theory, which became
known as the standard theory. In the standard theory, all transformations
were made to be obligatory and were triggered off by various abstract
symbols implanted in deep structures underlying particular sentences.
Such symbols as Q for questions or Neg for negative, generated by PS-
 Linguistic models and contrastive studies 137

rules, indicated that the appropriate T-rules were to apply to the strings
in which these symbols appeared to yield surface structures of interrog-
ative, negative, passive, etc., sentences. The standard theory incorporated
semantics as one of the components of the grammar. The semantic
component was interpretative in that it assigned semantic interpretations
to the deep structures generated by the central syntactic component, or
more strictly, the part of the syntactic component referred to as the Base
(PS-rules and the lexicon). All the necessary aspects of the meaning of
various types of sentences were potentially present in the form of symbols
indicating, on the one hand, what sort of meaning is to be assigned to
the sentence by the interpretative semantic component, and on the other,
what T-rules are to be applied to yield the surface structure of the
sentence. Thus, transformations did not, in principle, change the meaning
of strings on which they operated because all the relevant meanings were
determined at the level of deep structures, which constituted inputs to
semantic interpretations and to transformations.
In contrast to earlier models, in the standard theory PS-rules did not
account for only simple, declarative, affirmative sentences. They provided
deep structures of all types of sentences. As was said above interrogative,
negative, and passive sentences contained appropriate abstract symbols,
such as Q (Question), Neg (Negation), or Pass (Passive) in their deep
structures. These symbols triggered off the appropriate T-rules and se-
mantic rules. Complex sentences were generated by means of the recursive
S in the rule rewriting NP:

NP (Det) Ν (SO
(for details see Chomsky 1965: 107 ff.)
By way of example, let us see how various types of questions are
handled in a comparative study conducted within the framework of the
standard theory. All types of interrogative sentences are derived from
underlying basic strings containing the abstract element Q as a pre-
sentence morpheme, which triggers off the interrogative transformation.
Since there are several types of questions, including w/z-questions, tag
questions, and echo questions, and since many questions may occur in
either dependent or independent clauses, they require various T-rules to
account for them. These rules and details of their applications differ
across languages. The illustrative survey that follows concerns questions
in English and Polish and is based on Fisiak et al. (1978: 180 ff.).
The rule which ultimately accounts for the surface structure of English
interrogative sentences, the interrogative transformation, inverts the sub-
138 Chapter VI

ject NP and the auxiliary constituent called tense and the modal and/or
have and/or be, whatever the case may be. If no modal or have or be
follow tense, another transformation, called do-support, must be applied.
These rules yield surface structures of such sentences as

Has he gone there?

Will you do it?
Are we going to dance?
Do you speak English?

derived from the following respective strings

Q + NP + present + have + gone + there

Q + NP + present + will + do + it
Q + NP 4- present + be + going + to dance
Q + NP + present + speak + English

The interrogative transformation yields the following strings:

present + have + NP + gone + there

present + will + NP + do + it
present + be + NP + going + to dance
present + NP + speak + English

The last string illustrates the situation in which the transformation do-
support must be applied to eliminate the case of the "dangling" tense.
In Polish, inversion is optional, and when it does apply, it also moves
the main verb with tense attached to it. Therefore, Polish has no trans-
formation which would parallel the English do-support. In contrast to
English, the Polish rule which preposes auxiliary or the main verb is
ordered after the rule which attaches the tense affix to the auxiliary or
the main verb, with the result that in Polish sentence tense never "dan-
gles", and, consequently, do-support is not required. Moreover, since in
Polish, word order is generally more free, at least as with regard to
grammatical constraints, and the question word czy can be optionally
omitted, and a much larger number of arrangements of words in questions
is possible than in English:

Czy Piotr bgdzie jutro tutaj?

'Whether Peter be-future tomorrow here?'

Czy bgdzie Piotr jutro tutaj?

'Whether be-future Peter tomorrow here?'
 Linguistic models and contrastive studies 139

Gdzie b§dzie Janek spac?

'Where be-future Janek sleep?'

Gdzie Janek bgdzie spac?

'Where Janek be-future sleep?'
Czy on czytal t§ ksiqzkg?
'Whether he read that book?'
(Czy) czytal on t§ ksigzk§?
'(Whether) read he that book?'

The interrogative transformations in the two languages are formulated

as follows:
English: Q + N P + Τη + V => Tn + N P + V => i/o + Tn + N P + VP
through inversion through do-support
and Q-deletion
Polish: Q + NP, + Τη + V + NP 2 => Q + NP, + V Tn + NP 2
through Affix-placement
or => (Q) + V - Tn + NP, + NP 2
through inversion
and interrogative-deletion (optional)

Abstracting from the accuracy of these rules, they show that tense in
Polish does not dangle, and, consequently, ί/o-support is not necessary.
Furthermore, ß-deletion in Polish is optional. In those cases when Q is
not deleted, it is realized lexically as czy. In English Q is obligatorily
deleted in independent questions, and thus we get
Does he drink?
and not * Whether does he drink?
In dependent questions, the lexical marker of yesjno (general)-questions
is realized in both the languages, as czy in Polish and as if/whether in
English:

Zapytaj go, czy chce pic.

Ask him if/whether he wants to drink.
This sample contrastive study performed in the framework of the
standard theory, rather inconsistently applied and augmented by remarks
and observations couched in traditional terms, reveals some weaknesses
of the standard theory as a model founding contrastive study: On the
140 Chapter VI

one hand, its formal apparatus is rather complex, and it tends to obscure
rather than to highlight the relatively simple structural differences between
questions in the two languages (these differences can be stated more
clearly in traditional terms). On the other hand, the standard theory pays
insufficient heed to certain matters of detail, such as gender, number, or
case, especially in highly inflected languages, such as Polish. Details of
this kind may be trivial in the context of a theory which seeks to formulate
universal principles of language, but they may be of utmost importance
in establishing cross-language contrasts, and in fact they constitute the
main body of grammatical contrastive studies. One might say that cross-
language differences, and not linguistic universale, are what contrastive
studies are all about.
Fisiak et al. are well aware of these dilemmas inasmuch as they allow
their contrastive study to be eclectic rather than consistent within the
framework of the standard theory. In the Preface to their analysis, they
admit the following:
The authors, however, have tried to avoid formalizations as much as
possible, and have not hesitated to be eclectic and to use traditional
interpretations to explain the facts which have not yet been satisfac-
torily explained by transformationalists (e. g. gender, number, case and
aspect, to name but a few) or whose explanation, tentative as it is,
would require the introduction of a highly abstract and theoretical
apparatus, thus contributing to an already complicated picture of
language structure (Fisiak et al. 1978: 5).
The most serious limitation of the standard theory as a framework for
contrastive studies in the domain of syntax is that it does not provide a
clearly delimited semantic tertium comparationis against which such stud-
ies could be conducted. The semantic component in the standard theory,
being interpretative, is peripheral. The relevant semantic information,
though certainly available, is diffused in the device called the lexicon and
in the "projection" rules assigning semantic interpretations to sentences
and their constituents. On the other hand, the Base, generated by PS-
rules and lexical insertions, as formulated in the standard theory, cannot
serve as a tertium comparationis inasmuch as it is syntactic in nature. It
is required of a tertium comparationis that it should consist of properties
outside the compared material (Krzeszowski 1984: 305; Janicki 1985: 15).
Therefore, a syntactic base cannot serve as a tertium comparationis for
syntactic contrastive studies, if only because its shape is determined by
syntactic properties of specific languages. As a result, Bases in specific
 Linguistic models and contrastive studies 141

languages must be different in precisely those ways which contrastive

studies seek to discover and describe. Juxtaposing such Bases for com-
parison would illegitimately anticipate the stage of comparison proper
since such a juxtaposition would be founded on the assumption that the
Bases are in some respect similar, and this could only be a syntactic
similarity. Thus highlighting similarity and suppressing difference would,
on the one hand, defy the very purpose of contrastive studies, and, on
the other, more seriously, would entangle the procedure in the circularity
whereby syntactic similarity would be used as tertium comparationis for
a procedure which seeks to establish syntactic similarity and difference,
etc.
It is mainly due to these problems that the standard theory did not
enjoy popularity among contrastive analysts. 12 However, later develop-
ments in the transformational theory, such as Case Grammar and Gen-
erative Semantics, offered much better perspectives for solving these
thorny problems. Unfortunately, they also introduced a host of new
problems to cope with. Fillmore's Case Grammar (Fillmore 1968), not-
withstanding its obscurities and theoretical shortcomings (cf. Robinson
1970; Nilsen 1972), provided a very convenient framework for numerous
contrastive studies (e.g., König 1970; Krzeszowski 1971; Di Pietro 1971;
Konderski 1973; Boas 1977; and many others). In contrast with Chom-
sky's PS-rules generating deep structures of only English sentences, Fill-
more's rules were designed to generate universal structures, from which
surface structures of sentences in all languages could be derived. The
elements of these structures were unordered and involved nominal phrases
and verbs. The semantic functions of noun phrases were represented by
so-called cases (or more appropriately "roles"). Fillmore assumed that
the rules of the Base conceived in this way were universal and could be
stated in terms of modified predicate calculus:

In their deep structure the propositional nucleus of sentences in all

languages consists of a V and one or more NPs, each having a separate
case relationship to the Ρ (and hence to the V) (Fillmore 1968: 51).
Each proposition is preceded by Modality (M), so that the first rule
of the universal Base is formulated as
S —> Μ + Ρ (Fillmore 1968: 24)
This rule reflects the general assumption that every sentence, on the
semantic plane, consists of the propositional core, symbolized as Ρ and
non-propositional semantic material, symbolized as M. The concatena-
142 Chapter VI

tion symbol " + " appearing in the rule is somewhat misleading since the
two elements cannot be seen as appearing in linear order from left to
right in what purports to be a universal base where elements cannot be
linearly ordered. Neither does Μ stand in any one-to-one correspondence
to a particular sentence constituent or constituents since Μ can be realized
by a variety of syntactically heterogeneous means. In Fillmore's deep
structure, Μ represents such divergent categories as tense and aspect of
the verb appearing in P, sentence modality (declarative, interrogative, or
imperative), as well as modality expressed by various modal verbs, such
as English shall, will, can, may, etc. Fillmore does not provide any
indications as to how these various semanto-syntactic phenomena are
generated in the Base, nor what rules map the semantic structure of Μ
on surface realizations. 13
The propositional core Ρ states the event and its actors, i.e., nominal
concepts involved in the event and characterized as particular "cases":
Agentive (A), Dative (D), Factitive (F), Locative (L), Objective (O),
Instrumental (I), and perhaps a few more. The number and the nature
of these "cases" have never been determined or limited either by Fillmore
himself or by his followers, although various revisions of the original
proposal have been suggested. 14 Regardless of the number of postulated
cases, Ρ can be expanded according to the following schema:
P-^V + C,+ +cn
where particular expansions can be realized as V + A, V + Ο + A,
V + D, V + O + I + A, etc. In the dictionary, particular verbs are
represented in terms of "frame features" indicating sets of case frames
into which particular verbs can be inserted. A given verb can appear in
more than one frame. For example, open in the door opened appears in
the frame O, while in John opened the door the same verb
appears in the frame Ο + A. In the wind opened the door the
verb open is assumed to appear in the frame Ο + I, whereas
in John opened the door with a chisel, the frame is Ο + I + A.
Fillmore further argues that subject and object are notions which are
only relevant in the surface structure of sentences, and that, in most
cases, the NP which becomes the subject in the surface structure is not
marked as such in the deep structure. Instead the following general rule
applies:
If there is an A, it becomes the subject; otherwise, if there is an I, it
becomes the subject; otherwise, the subject is Ο (Fillmore 1968: 33).
 Linguistic models and contrastive studies 143

This rule can be augmented to cover other "cases" as in Langendoen

(1970: 59 ff.).
A modified version of Fillmore's model was used by Di Pietro in his
monograph Language structures in contrast (Di Pietro 1971: 54 ff.), which
was the first book entirely devoted to various theoretical and methodo-
logical issues in contrastive studies, rather than to a contrastive analysis
of a specific pair of languages. Di Pietro assumes that there is a deep
structure shared by all languages and that

the differences between languages can be found at any number of

stages between the deep and the surface structures It follows that
languages are similar to each other in proportion to the number and
hierarchical ordering of rules shared in the intermediate levels. Thus,
Spanish shares more rules with Italian than either does English or
German. The four languages all share more of such rules than any of
them does with Swahili. Finally, Spanish, Italian, English, German,
and Swahili all share the deepest level of structure because they are
all human languages and are all equally accessible to human learners
(Di Pietro 1971: 27).
In Di Pietro's deep structure all languages of the world share such
basic elements as SENTENCE, MODALITY, and PROPOSITION as
well as basic case relationships, i. e., agent, dative, object, instrument, etc.
Language-specific rules operating on various levels of derivation account
for surface-structure similarities and differences between particular pairs
of languages.
Di Pietro's model differs from both Chomsky's standard theory and
Fillmore's case grammar. In Di Pietro's model, in contrast with Chom-
sky's, those deep structure categories which are later realized as NP's are
not arranged in any linear or hierarchical order analogous to the order
in which NP's in Chomsky's deep structures appeared. Di Pietro's model
differs from Fillmore's in that the former contains language-neutral,
"universal" categories or, as Di Pietro calls them, "syntactic primes",
i.e. Names and Verboids, in addition to S(entence), M(odality),
P(roposition), and C(ase) CAT(egory). Di Pietro defines "Name" as a
"unit of arrangement which refers to that property of human languages
not only as verbs but also as adjectives and possibly as adverbs. The
other universal categories in Di Pietro's deep structures are arranged
hierarchically in the following fashion:
144 Chapter VI

[time] V CCATlObjJ CCATtagentJ

CM Ν CM Ν

[past] [drink, [wine, [John,

boirei vin] John]

where CM is a case marker and where the linear ordering of constituents

is irrelevant (Di Pietro 1971: 61).
Unlike in Fillmore's version of case grammar, in Di Pietro's model
the specifics of case, i. e. agent, dative, instrumental, etc. are provided
by the semantic projection rules and are not generated by the syntactic
rules, which is consistent with Di Pietro's claim that his base rules are
syntactic rather than semantic. We shall comment on this point later. At
this moment let us observe that Di Pietro's hierarchical arrangement of
"syntactic primes" as well as his recognition of S as a syntactic prime in
this particular framework raise certain doubts.
If Noun Phrase and Verb are dismissed from the universal base on the
grounds that they are too "suggestive of the surface restrictions of
particular languages" (Di Pietro 1971: 56), then S should be dismissed
on the same grounds since surface structures in particular languages
observe numerous constraints characterized by grammars of those lan-
guages. In the standard theory, S directly dominated NP and Predicate
Phrase, which reflected the assumption that in English every sentence in
its surface structure consists of a noun phrase followed by a predicate
phrase, and that they appear in the specified order unless transformed
otherwise. This sort of deep structure was a direct result of the analysis
of every sentence into two immediate constituents expressed by categories
related to the reality of the surface structure of English sentences. Ad-
 Linguistic models and contrastive studies 145

mittedly, not all sentences could be analysed in this way, but at least
every sentence could be "untransformed" to restore the basic underlying
shape. It was, furthermore, known that every noun phrase contained a
noun as one of its constituents and every predicate phrase contained a
verb or be. Thus, the categories appearing in the deep structure were not
different from those appearing in the surface structure; consequently, the
initial element S was justified in that it dominated all the categories which
appeared in the surface structure of sentences in addition to whatever
symbols were necessary to trigger off appropriate transformations map-
ping deep structures into surface structures.
In Di Pietro's deep structure, S no longer dominates those categories
which in the surface structure are constituents of sentences. Instead it
dominates a number of "syntactic" units which in various languages are
realized as various surface-structure categories in a wide variety of ways.
For example, Modality cannot be associated with any single type of
expansion. As a matter of fact, in purely syntactic terms, it cannot be
expanded at all, at least in terms of PS-rules, as it incorporates various
semantic elements which have highly heterogeneous syntactic realizations,
such as time, aspect, question, negation, etc. Moreover, it is possible to
envisage a situation in which a particular deep structure will be realized
as one surface-structure sentence in some language and as two or possibly
more surface-structure sentences in another language. Non-restrictive
relative clauses constitute a good example:

John is a coward. He ran away.

John, who is a coward, ran away.

Whether or not the two phrase markers underlying the two sentences
come under the domination of one S is a matter which cannot be decided
at the level of deep structure suggested by Di Pietro. A single Sentoid(?)
may be realized as two surface-structure sentences, or two Sentoids may
be realized as one surface-structure sentence without affecting the basic
meaning. The common deep structure correctly reflects this situation, but
it contains no indication of the number of sentences which are its surface-
structure realizations.
Sentences with concessive or resultative clauses, which Di Pietro as-
sociates with Modality, furnish another example illustrating the point in
question. Di Pietro's deep structure, which purports to be syntactic in
nature, does not distinguish between the following sentences:

If it rains, I'll take my umbrella.

It will rain, and I'll take my umbrella.
(understood conditionally rather than sequentially)
146 Chapter VI

Di Pietro's syntactic deep structure does not distinguish between the

two sentences on semantic grounds as the two sentences express the same
semantic content. Yet, a purely syntactic base, such as Di Pietro's base
purports to be, ought to be capable of distinguishing between complex
and compound sentences.
The difficulty results from the fact that although Di Pietro assumes
that his deep structure contains syntactic primes, in reality it is a construct
representing semantic relations underlying surface structures of sentences.
Such descriptions of terms appearing in Di Pietro's deep structures as
"[things] which represent man's universal need to label objects, ideas and
sentiments which are subject to cognitive powers" (Di Pietro 1971: 56)
clearly point to the semantic nature of the objects thus described.
The difficulties alluded to above seem to suggest that if syntactic
contrastive studies are to be conducted whithin the framework of a
generative transformational grammar, it is necessary to adopt a version
of the theory (possibly a notational variant) in which distinction is made
between the semantic structure as input to transformations and the
syntactic structure as output of transformations. Consequently, a consis-
tent difference must be made between semantic primes of semantic rep-
resentations and surface structure categories. In his attempt to construct
the universal base, Di Pietro is only half successful inasmuch as he
introduces semantically grounded terms into his syntactic model of the
base. For the purposes of syntactic contrastive studies, in order to intro-
duce a semantic tertium comparationis, it is necessary to go further and
to postulate an universal semantic base, free of all syntactic relations,
which are inevitably language-specific. The existence of such an universal
semantic base would be the initial postulate for the construction of
contrastive grammars of various languages (see Chapter VIII).
Chapter VII

Towards a theory of semanto-syntactic equivalence1

Syntactic contrastive studies, even if based on transformational generative

grammars, remain taxonomic in nature as long as they aim at producing
inventories of differences and possibly similarities either between parallel
places of grammatical systems or between parallel constructions or even
between parallel rules operating at various levels of derivation. Admit-
tedly, applying the generative-transformational framework allows the
investigator to grasp differences and similarities at those levels of repre-
sentation which have been made available thanks to the developments in
generative-transformational grammar, i. e., the levels which have been
sometimes referred to as "deep structure" However, numerous contras-
tive studies conducted within the generative-transformational framework,
discussed in the previous chapter, do not aim at providing an explicit
format for contrastive studies where explicitness would not refer merely
to the way in which the materials compared are presented individually
but to the contrasting technique itself. Although "translational equiva-
lence" was often adopted as the criterion of comparability, no precise
definition of the concept emerged from those studies. Equivalent con-
structions (and sentences) were said to be at least sometimes mutually
translatable, and the relation which holds between such objects was called
"textual equivalence" (Catford 1965: 27 — 34). In Chapter II, we pointed
to certain difficulties connected with adopting translation equivalence as
a basis for syntactic contrastive studies. We postulated that it is necessary
to distinguish between translation as a concept in the domain of perform-
ance and semanto-syntactic equivalence as a concept in the domain of
competence. In this chapter, we propose to elaborate on the concept of
semanto-syntactic equivalence in order to provide grounds for a more
rigorous definition. We shall first put forward a hypothesis concerning
semanto-syntactically equivalent sentences, and we shall provide a pre-
liminary justification for this hypothesis. We shall then discuss and clarify
certain difficulties that arise from the hypothesis. Finally, we shall re-
phrase the hypothesis as an initial postulate (axiom), using it as a basis
148 Chapter VII

for a theory of semanto-syntactic equivalence, which we shall outline in

Chapter VIII.
In order to discover equivalents across languages, one has to rely on
the authority of a competent bilingual informant, usually the investigator
himself. The informant's judgments are based on his intuition, which is
connected with his competence in the two languages. Although the notion
of semanto-syntactic equivalence is intrinsically connected with the mean-
ing of the linguistic objects compared, a complete constrastive study must
also give an account of their form. If we characterize various constructions
in any two languages in terms of equivalence and formal correspondence,
we shall observe that the compared sentences may be: (a) equivalent,
formally different, (b) equivalent, formally similar, (c) non-equivalent,
formally similar, (d) non-equivalent, formally different. In this chapter
we are mainly concerned with situations (a) and (b).
We shall work on the assumption that the first term in each pair refers
to the deep structures of the compared sentences while the second term
refers to their surface structure. Moreover, we shall try to justify the
following hypothesis: Equivalent sentences have identical deep structures
even if on the surface they are markedly different.
Before we provide some justification for this hypothesis, let us specify
what is meant by "deep structure" in the standard theory. The argument
which follows will show that "deep structure" as conceived in the standard
theory is too "shallow" to accomodate the above hypothesis.
Lakoff (1968) derives a definition of deep structure from Katz —Postal
(1964) and from Chomsky (1965). Deep structure in the standard theory
is viewed by Lakoff as a level of linguistic analysis where (a) basic
grammatical relations between fundamental grammatical categories are
defined; (b) selectional restrictions and co-occurrence relations are stated;
(c) appropriate grammatical categories receive lexical representations; (d)
inputs to transformational rules are provided.
Condition (b) implies that the selectional restrictions and other co-
occurrence relations stated at the level of deep structure make it possible
to formulate correct generalizations about selectional restrictions and co-
occurrence relations among the elements of the sentence, and that they
remain constant in the surface structure.
Condition (c) emerges from the assumption that the level of deep
structure directly determines the semantic interpretation of the sentence.
Since semantic interpretation involves semantic rules defined both in
terms of semantic content of lexical items and by grammatical relations,
conditions (b) and (c) are interdependent. Nevertheless, the semantic
 Towards a theory of semanto-syntactic equivalence 149

component as viewed by Katz—Pastal (1964) involves rules of a different

order ("projection rules") than the rules of the Base.
This definition of deep structure allows one to make decisions as to
the identity of deep structures of synonymous constructions which are
not identical on the surface. By demonstrating that the same selectional
restrictions and other co-occurrence relations are in effect for synony-
mous, though superficially different, constructions, it is possible to infer
that such constructions are essentially identical at the level of deep
structure, in accordance with the restrictions as stated in (b). As we shall
see later, the identity of selectional restrictions and co-occurrence relations
is a necessary, but not a sufficient, condition for two synonymous con-
structions to have a common deep structure. At this stage, however, let
us look at how Lakoff demonstrates the identity of deep structures of a
pair of constructions represented by

(1) Seymour sliced the salami with a knife.

(2) Seymour used a knife to slice the salami.

Before we continue with our task, let us introduce the Polish equiva-
lents of (1) and (2) in order to be able to subject them to the same
procedure to which Lakoff subjects (1) and (2). The Polish equivalents
of (1) and (2) are, respectively

(3) Seymour pokrajal salami nozem.

(4) Seymour uzyl noza aby pokrajac salami.2

In (3), the instrumental inflection in the word nozem is an equivalent

of the preposition with in (1) since the former like the latter has the
purposive, instrumental sense (Lakoff 1968). Also in (4) the verb uzyc is
an equivalent of use in (2) as both appear in the instrumental sense. The
surface structures of the two pairs of sentences can be represented as

(5) NP, V NP 2 - with - NP 3

(6) NP, use - NP 3 to - V NP 2
(7) NP, V NP 2 - Inst - NP 3
(8) NP, - uzyc NP 3 - aby - V NP 2

where Inst in (7) stands for the instrumental inflection.

Lakoff demonstrated that the same set of selectional and co-occurrence
restrictions holds in (5) and (6), and he thus proved the identity of the
deep structures underlying the two constructions. It appears that a parallel
set of restrictions applies with reference to (7) and (8):
150 Chapter VII

I. In Polish as in English the verb represented by V must be characterized

by the feature [ + Activity] for the sentence to be grammatical:
(9 a) Albert computed the answer with a sliderule.
b) Albert used a sliderule to compute the answer.
(10 a) Albert obliczyl (uzyskal) odpowiedz suwakiem.
b) Albert uzyl suwaka aby obliczyc odpowiedz.
(11 a) * Albert znal odpowiedz suwakiem.
b) * Albert uzyl suwaka aby znac odpowiedz.
II. Both in English and in Polish NP, must be [ + animate]:
(12 a) Peter hit Oliver with a hammer.
b) Peter used a hammer to hit Oliver.
(13 a) Piotr uderzyl Olivera mlotkiem.
b) Piotr uzyl mlotka aby uderzyc Olivera.
(14 a) *Cios uderzyl Olivera mlotkiem.
b) *Cios uzyl mlotka aby uderzyc Olivera.
III. In none of the constructions can NP 2 have the same referent as NP 3 :
(15 a) He unlocked the door with a master-key.
b) He used a master-key to unlock the door.
(16 a) Otworzyl drzwi wytrychem.
b) Uzyl wytrycha aby otworzyc drzwi.
(17 a) * Otworzyl drzwi nimi samymi.
b) * Uzyl drzwi aby je otworzyc.
IV. In none of the constructions can NP] have the same referent as NP 3 :
(18 a) Anatol broke the pane with the scared soldier.
b) Anatol used the scared soldier to break the pane.
(19 a) Anatol wybil szyb§ przerazonym zolnierzem.
b) Anatol uzyl przerazonego zotnierza aby wybic nim szybe.
(20 a) ?Anatol wybil szybg sobg.
b) * Anatol uzyl siebie aby wybic szybg.
V. Questions which are derived from (7) and (8) are both ambiguous in
the same way as are the equivalent questions in English:
(21 a) Did Seymour slice the salami with a knife?
b) Did Seymour use a knife to slice the salami?
(22 a) Czy Seymour pokrajal salami nozem?
b) Czy Seymour uzyl noza aby pokrajac salami?
 Towards a theory of semanto-syntactic equivalence 151

All the four questions are ambiguous in at least two senses. First, they
may be questions asking whether the actual action of slicing took place.
Second, if the action is presupposed, they may ask for confirmation that
the instrument used for the slicing of the salami was a knife.
VI. Negative constructions based on (7) and (8) are also ambiguous in
the same way as are the equivalent constructions in English:
(23 a) Seymour did not slice the salami with a knife.
b) Seymour did not use a knife to slice the salami.
(24 a) Seymour nie pokrajal salami nozem.
b) Seymour nie uzyl noza aby pokrajac salami.
All four sentences may mean that the action of slicing did not take
place or, if the action is presupposed, that the instrument used was not
a knife.
VII. In Polish as in English the verb phrase functioning as the complement
of uzyc cannot be negated:
(25 a) *I used the knife not to slice the salami.
b) */ used the knife to not slice the salami.
(26 a) *Uzylem noza aby nie pokrajac salami.
The fact that Polish equivalents of (5) and (6) are subject to a parallel
set of selectional and co-occurrence restrictions provides preliminary
justification for the hypothesis that equivalent sentences have identical
deep structure.
In his paper Lakoff did not attempt to formulate any explicit rules
representing relations between various grammatical categories in the deep
structure. He contented himself with some inferences about deep struc-
ture, the most significant one being that
Deep structures must be somewhat more abstract (further removed
from the surface) than previous research in transformational grammar
has indicated (Lakoff 1968: 24).
Chomsky questioned the validity of Lakoff s argumentation support-
ing the conclusion quoted above and claimed that different deep structures
can "express the required selectional and grammatical relations in a
unified way" (Chomsky 1972: 84). Therefore, according to Chomsky,
there is no need to postulate a more abstract deep structure where these
generalizations could be stated. Nevertheless, for independent reasons,
many linguists also expressed the conviction that deep structure must be
152 Chapter VII

more abstract, and that it should incorporate semantics. It was claimed

that semantics is not only interpretive but also generative, and that,
consequently, semantic representation and syntactic representation are of
essentially the same formal nature (McCawley 1968: 167 — 169).
The formal apparatus used in this uniform representation is the general
framework of symbolic logic or, more specifically, predicate calculus,
where propositions correspond to deep structure sentences, functions
correspond to verbs, and arguments to noun phrases or nouns. Fillmore's
base, expressed in terms of case relations of noun phrases to verbs, can
be viewed as a possible implementation of the logical notation in the
description of a natural language. Before we explore the possibilities of
applying such a semantic base as tertium comparationis for syntactic
contrastive studies (see also the account of Di Pietro's attempt in the
previous chapter), let us consider various problems which arise in con-
nection with the hypothesis that equivalent sentences have an identical
deep structure and the possible implications that this hypothesis, if true,
could have for the nature of deep structure. For the sake of brevity we
shall refer to the above formulated hypothesis as the "identical-deep-
structure hypothesis" However, it should be remembered that when this
hypothesis was first formulated (Krzeszowski 1971: 38), the term "deep
structure" was used to mean a more abstract level of representation than
the level of deep structure as formulated in the standard theory. This
more abstract level could be more appropriately called "semantic struc-
ture" or "input structure", where "input" would mean "input to syntactic
derivations"
The most elaborate criticism of the identical-deep-structure hypothesis
comes from Bouton (1976). He points to two difficulties involved in its
original formulation. The first difficulty is that "tense and aspect con-
tained within them [two identical semantic structures] must be identical
also" (Bouton 1976: 154). In a great number of Polish and English
sentences, including those discussed above, i.e., (1) and (3), aspect is
apparently different since in Polish but not in English, completed (perfect)
vs. incompleted (imperfect) aspect of the verbs involved in the sentence
is obligatorily expressed. Thus, not only (1) and (3) but also (2) and (4)
are not equivalent since the Polish sentences, in contrast with their English
counterparts, contain information that "Seymour had finished his work,
having cut all the salami that he had intended to cut" (Bouton 1976:
154). The incompletion of the action of cutting would have been expressed
by means of a different aspectual form, viz. uzywac, as in
(27) Seymour uzywal noza aby pokrajac salami.
 Towards a theory of semanto-syntactic equivalence 153

which would presumably mean "Seymour was cutting salami and did not
finish"
Despite certain inaccuracies in Bouton's interpretation of the Polish
aspectual forms, to which we shall return shortly, the point that he is
making is clear: since in contrast with English, Polish must specify the
feature [ ± completion] on the verb, Polish and English sentences must
have different semantic structures in spite of the fact that they are attested
to be equivalent. Therefore, the hypothesis that equivalent sentences have
an identical semantic structure is wrong.
The other difficulty that Bouton saw in the identical-deep-structure
hypothesis involved active and passive constructions in English and their
Finnish equivalents:

(28) Tom broke the window. (28F) Tom rikkoi ikkunan.

(29) The window was broken by Tom.
(30) Someone broke the window. (30F) Ikkuna rokotiin.
(31) The window was broken by someone.

According to Bouton, since (28) and (29) are mutually translatable

with (28F) and, and (30) as well as (31) with (30F), the two respective
sets constitute textual equivalents in the two languages. Since, however,
(28) and (29) cannot be derived from the same deep structure and neither
can (30) and (31), one or the other of the two English sentences in both
of the triplets must be derived from a different structure, thus contra-
dicting the hypothesis that translation equivalents have identical deep
structures.
Bouton's first example functions as evidence against the identical-
deep-structure hypothesis only on the assumption that aspect, like tense,
must be represented in the semantic structure of sentences. As Bouton
does not support this assumption with any arguments, it must be taken
that the assumption has the character of an axiom and as such requires
no proof. Under this particular axiom, it may be surmised that any
semantic distinction must be reflected in the semantic structure of sen-
tences as is indeed the case in the so-called generative semantics. However,
an alternative way of viewing semantic phenomena, namely that not all
semantic distinctions are to be included in the semantic representation of
sentences, should also be allowed. Such an alternative might exclude
aspect (though not necessarily tense) from the universal stock of semantic
distinctions to be represented in the semantic structure of sentences. It is
quite possible to adopt the initial view that semantics falls into sentence
154 Chapter VII

semantics and word semantics (see, for example, Bartsch — Vennemann

1972). This distinction is drawn in the following manner:
The sentence semantics of a sentence represents the meaning of the
surface syntax of a sentence in logical syntax without reference to the
meaning of any particular word of the sentence (Bartsch — Vennemann
1972: 39).
The following aspects of sentences contribute to their semantics: the
categorial and subcategorial status of constituents, morphological mark-
ers (inflection), word order, and intonation. These markers serve as
exponents of such semantic information as declarative, interrogative,
tense, sentence connection, sentence embedding, topic/comment relations,
focus formation, and presuppositions. What exactly constitutes the do-
main of sentence semantics and what does not is a moot point, but at
least it is possible that the aspect of Polish verbs should not be included
there but rather in the domain of word semantics, which is treated in the
lexicon. On the other hand, the English durative and perfective "aspects"
should rather be represented at the level of sentence semantics. Some of
the reasons motivating different treatment of aspect in Polish (and other
Slavonic languages), in comparison with the treatment of "aspect" in
English, are the following:

1. Aspect in Slavonic languages is an intraword phenomenon while in

English both durative and perfective require the presence of auxiliary
verbs.
2. In Slavonic languages changes in the aspect do not affect the syntactic
structure of the sentence while in English sentences expressing aspect
have a different structure in questions, negations, etc., from sentences
without aspect.
3. Every Slavonic verb is inherently completed or incompleted while no
English verbs are inherently "durative" or "perfective", just as in the
majority or cases, nouns are not inherently singular or plural and need
not be listed separately in the lexicon as plural.
4. There is a high degree of unpredictability concerning the completed
and incompleted forms of verbs in Slavonic languages, and these forms
have to be listed in the lexicon. Rules, if at all possible to formulate,
abound in exceptions. In English all durative and perfective forms
obey a simple syntactic rule:

Aspect —> (be 4- ing) (have + en).

 Towards a theory of semanto-syntactic equivalence 155

If the distinction between sentence and word semantics is drawn, and

if the Polish aspect is treated in the lexicon, the Polish and English
"Seymour sentences", objected to by Bouton, and other equivalent Polish
and English sentences would still have identical semantic structures at
the level of sentence semantics. In addition to sharing identical sentence-
semantics structure, equivalent sentences would contain equivalent lexical
items drawn from a bilingual dictionary where particular lexical entries
in LI would correspond to lexical entries in L2 according to complex
networks of grammatically and semantically conditioned relations. Such
a dictionary would have to contain as its entries on both sides the same
lexical entries which appear in respective monolingual dictionaries, al-
though the organization of the definitions and the information included
in the bilingual dictionary would not be a simple juxtaposition of the
respective monolingual dictionaries, but rather a principled network of
matchings of lexical items across languages (cf. Nowakowski 1977: 32 —
33). Completed and incompleted verbs in Slavonic languages would
constitute separate lexical entries in monolingual dictionaries of Polish,
as part of the overall account of the Polish language, as well as in the
bilingual Polish-English dictionaries, as part of the overall account of
equivalence relations between Polish and English sentences (and texts).
Thus such Polish verbs as uzyc (perfective) and uzywac (imperfective) as
well as pokrajac jpokroic (perfective) and krajacjkroic (imperfective) would
be entered separately in the Polish part of the dictionary.
Marginally, let us note that the difference between Seymour uzyl noza
aby pokrajac salami and Seymour uzywal noza aby pokrajac salami does
not consist in one action being completed in contrast with an incompleted
one but rather in the iterative (repetitive) character of the action expressed
by the verb. Thus, Seymour uzywal noza aby pokrajac salami would
suggest "using a knife more than once on separate occasions" and could
be rendered in English as "Seymour used to use a knife to slice the
salami" On the other hand, uzywac may express the concept of duration
as in Seymour uzywal noza przez trzy godziny 'Seymour was using knife
for three hours'
However, both kroic and pokroic will correspond to the English verb
slice. (Of course, other English equivalents, such as cut, carve, etc., must
also be included, but this fact is irrelevant to the problem of aspect
correspondences.) The fact that both the imperfect verb kroic and the
perfect verb pokroic correspond to only one form in English, which is
aspect neutral, does not ruin our theory of equivalence, if the notion of
156 Chapter VII

markedness is employed to constrain equivalence along the lines ex-

pounded below.
The notion of markedness is involved in the Polish aspect in the
following way: the so-called perfect verbs in Polish enter into oppositions
with the imperfect ones, with perfect verbs being marked and imperfect
ones being unmarked. This means that an imperfect verb can be used in
all those situations in which a corresponding perfect verb can be used
but not vice versa, as in
(32) Seymour krajal salami przez caly dzien. (unmarked)
(32E) Seymour sliced the salami all day. (unmarked)
(33) Seymour krajal salami do skutku. (unmarked)
(33E) Seymour sliced the salami to the end. (unmarked)
(34) Seymour pokrajal salami do skutku. (marked)
(34E) Seymour sliced up the salami to the end. (marked)
(35) * Seymour pokrajal salami caly dzien. (marked)
(35E) ?Seymour sliced up the salami all day. (marked)
Thus, the imperfect verb, being unmarked, can express both completed
and incompleted actions, while the perfect verb, the marked one, can
only refer to completed actions.
In English, the completion of the action can be optionally marked by
means of the particle up added after the verb as in (34E), which is then
equivalent to (34) in that both are marked for completion. (32) and (32E)
as well as (33) and (33E), which are respectively equivalent, are not
marked for the feature [ ± completion]. Note, however, that
(36) Seymour pokrajal salami, (marked [ + completion])
and
(36E) Seymour sliced the salami, (unmarked)
are also equivalent, while
(37) Seymour bezskutecznie krajal salami, (marked [— completion])
and
(37E) Seymour sliced up the salami, (marked [ + completion])
are not equivalent.
Under no circumstances can (37) and (37E) be considered as equivalent
since they have conflicting marking polarity.
 Towards a theory of semanto-syntactic equivalence 157

The situation is somewhat more complex with respect to the verbs

uzyc, zuzyc, and zuzywac. In this case, too, we deal with the opposition
marked vs. unmarked, but in addition to the feature [ + completion] the
feature [ + iterative] is also involved. In every case it is possible to mark
polarity by three types of specification. A lexical item W may be positively
specified for a feature F as [ + F], negatively specified for F as [ — F] or
remain unspecified [ ± F], We will say that if W is either positively or
negatively specified for F, it is thereby marked. Otherwise, it is unmarked.
Adopting this marking convention, we can discover the following system
of marking for uzyc, zuzyc, and uzywac, and their English equivalents:

(38) marked unmarked

[ + completion] zuzyc 'use up' uzyc
[—completion] uzywac 'use' uzyc
[ + iterative] miec zwyczaj uzywac uzyc
'use to use' (chodzic) 3
(chadzac 'use to go')
[— iterative] zuzyc 'use up' uzyc
(38) can be rearranged to spell out the features accounting for markedness
in the following way:
zuzyc + completion use up + completion
— iterative — iterative
miec zwyczaj — completion use to use — completion
uzywac + iterative + iterative
uzyc ± completion use + completion
+ iterative ± iterative
uzywac — completion
+ iterative
Note: the form miec zwyczaj uzywac, literally 'have the habit to use'
is synonymous with *zwykngc uzywac. The latter, however, is defective
as it never appears in the infinitival forms, In finite forms, such as zwykl
uzywac 'used to use', it will appear in our examples as marked for
[— completion, + iterative].
This system of marking allows us to formulate the following constraint
on equivalence of lexical items:
(C,) N o two lexical items will be assigned the status of equivalence if
they have conflicting marking polarity with respect to a specific
feature F.
158 Chapter VII

In order to determine whether or not a lexical item Wj in a language

L p (W L ) is equivalent to a lexical item Wj in a language L q (W L ) the
constraint (Ci) has to be applied in the following way:
WPL wqL Equivalence
polarity
Yes
+ + Yes
+ ± Yes
± + Yes
± Yes
± Yes
+ No
+ No

Constraint (C t ) will permit us to assign the status of equivalence to

the following pairs of sentences:

marking polarity:
+ c = ± completion
+ i = + iterative

(40) Seymour used the knife ± c ± i

Seymour uzyl noza ± c ± i
(41) Seymour used the knife ± c ± i
Seymour uzywal noza - c ± i
(42)? Seymour used the knife ± c ± i
Seymour zwykl uzywac noza - c + i
(43)? Seymour used the knife + c ± i
Seymour zuzyl ηόζ + c - i
(44) Seymour used to use the knife - c + i
Seymour zwykl uzywac noza - c + i
(45) Seymour used to use the knife - c ± i
Seymour uzywal noza - c ± i
(46)? Seymour used to use the knife - c + i
Seymour uzyl noza ± c ±i
(47) Seymour used up the knife + c - i
Seymour zuzyl ηόζ + c - i
(48)? Seymour used up the knife + c - i
Seymour uzyl noza ± c + i
 Towards a theory of semanto-syntactic equivalence 159

The following pairs, because of their conflicting polarity, will not be

assigned the status of equivalents:
(49) Seymour used to use the knife - c + i
Seymour zuzyl ηόζ + c - i
(50) Seymour used up the knife + c - i
Seymour zwykl uzywac noza - c+ i
(51) Seymour used up the knife + c - i
Seymour uzywal noza - c ± i

As is to be expected, those sentences that share both polarities seem

to be better equivalents than those that share only one polarity. The least
acceptable as equivalents are those pairs which do not share any polarity,
even if no conflict of polarity occurs. Such is the case with (42), (43),
(46), and (48). It is possible that further constraints are needed in those
cases when more than one polarity is involved. It may be required, for
instance, that in such cases equivalent lexical items should share at least
one polarity. Such a more severe constraint would rule out (43), (46),
and (48) as equivalents. However, this problem has no bearing on the
present discussion.
In view of C t , the fact that some of the examples discussed by Bouton
carry the implication that the action expressed by the verb was completed,
while the English equivalents carry no such implication, does not lead to
the conclusion that the relevant sentences have different semantic struc-
tures. At the level of sentence semantics, they can be said to have identical
semantic representations. As long as at the level of word semantics these
sentences do not violate Q , they can be recognized as equivalents despite
noticeable differences in the meaning at the level of word semantics.
Let us now turn to the second of Bouton's objections to the identical-
deep-structure hypothesis. According to Bouton, it can be saved only if
one of the two mutually exclusive situations can be shown to be true: (a)
English passive and active sentences can be derived from the same deep
structure. In that event (28) and (29) would be considered as paraphrases,
with (28 F) as their Finnish equivalent; all these sentences could then be
derived from the same semantic structure, (b) The Finnish sentences such
as (28F) and (30F) could be considered as ambiguous, each being derived
from two different semantic structures, such as those underlying the
English sentences (28) and (29), respectively.
The first solution is impossible in view of the fact that ever since the
publication of Katz —Postal (1964), English active and passive sentences
160 Chapter VII

have been repeatedly demonstrated to be derived from different under-

lying structures. The second solution, according to Bouton, is also im-
possible as the Finnish verbs in (28) and (30), as well as in other similar
sentences, observe selectional restrictions which correspond exactly to the
selectional restrictions of English active sentences. They do not observe
any selectional restrictions that characterize passive sentences. Thus, in
the first place, Finnish verbs are not subject to any stativity restrictions,
which characterize English passive sentences. In the second place, intran-
sitive verbs can occur in both English active sentences and in Finnish
sentences, but not in English passives, so that we can have:

(52) Sunnuntaina mennaan kirkoon.

'On Sunday people go to church'
(53) Suomessa hudetaan paljon.
'In Finland one skis a lot'
But not
(54) * Was laughed loudly by Bill, (from Bill laughed loudly.)
(55) *It sat here all day. (from Someone sits here all day.)
Therefore, Bouton concludes that the identical-deep-structure hypoth-
esis fails since the translation equivalents (28), (29) —(28F), and also (30),
(31) —(30F) cannot be shown to have identical deep structures: either
(28) and (29) have two different deep structures, in which case (28F) has
the identical deep structure with only one of them, or, quite implausibly,
(28F) and (30F) must be recognized as ambiguous. The latter solution is
impossible to adopt as the Finnish sentences observe selectional restric-
tions which characterize active sentences and do not observe any of those
restrictions which characterize passive sentences. Therefore, (28F) and
(30F) cannot be interpreted as ambiguous and can each have only one
deep structure. Since, however, they each correspond to at least two
English sentences, each having a different source, the identical-deep-
structure hypothesis fails as at least one of the translation equivalents of
both (28F) and (30F) is derived from a different deep structure.
It must be admitted that Bouton's reasoning is impeccable and that
the sentences in question cannot be claimed to have an identical deep
structure as established through the examination of selectional restrictions
since these restrictions are quite obviously different for active and passive
sentences.
However, the problem must be seen somewhat differently and must
involve the decision as to what in Finnish corresponds to the English
 Towards a theory of semanto-syntactic equivalence 161

sentences (28), (29), (30), and (31), respectively. In order to answer this
question, it will be necessary to examine the semantic content of English
passive sentences and the ways in which the same content, if at all, is
rendered in Finnish. Such problems can be formulated and answered
within linguistic models in which deep structure is not an independent,
well-defined, and well-delimited level of analysis, as in the standard
theory, but only within such models in which sentences are derived from
semantic representations of a more abstract nature. Bouton's reasoning,
as well as arguments presented by other authors, including the present
one (cf. Van Buren 1974; Krzeszowski 1974, 1979, see also the previous
chapter) show that the deep structure as formulated in the standard
theory cannot serve as a possible tertium comparationis for conducting
syntactic contrastive studies.
It should be noted in this connection that if two sentences across
languages do not observe the same set of co-occurrence and selectional
restrictions, it does not necessarily follow that they do not share an
underlying structure at some level of linguistic representation. It only
means that they do not share that level of analysis at which these
restrictions are stated, i.e. the level of the standard theoretical deep
structure. Therefore, if two sentences do share the same set of co-
occurrence and selectional restrictions, they share deep structure and, in
all probability, also more abstract levels of representation, ipso facto
being equivalent. However, the fact that two sentences do not share the
same deep structure does not rule out the possibility of their sharing a
more abstract semantic structure and thereby being equivalent, as well.
Nevertheless, considering the constraints on equivalence formulated be-
low, we shall try to show that the technique of establishing equivalence
through testing co-occurrence and selectional restrictions obtaining in
pairs of sentences across languages is valid, even if more abstract levels
of representation are accepted as tertium comparationis for syntactic
contrastive studies.
At this point, it is necessary to make a digression in order to recall
the essential distinction drawn in Chapter II between translations and
semanto-syntactic equivalents. The term "equivalent sentences" refers to
those sentences in two or more languages which have identical input
structures; this is not synonymous with the term "translations" in the
actual translation practice. The ability to recognize equivalents is a part
of a bilingual person's competence, while ability to translate is a part of
translation performance. In actual translation practice, sentences which
are translations often differ in their meaning, not necessarily due to errors
162 Chapter VII

in translation, but owing to a number of grammatical, stylistic, and

pragmatic reasons, which result in frequent instances of translations being
semantically non-equivalent. For instance, a line from Pygmalion:
(56) Nobody's going to touch you
is translated by Florian Sobieniowski as
(56P) Nie böj si§ nie damy ci krzywdy zrobic.
'Don't fear, we won't let anybody hurt you'
(56) and (56P) are obviously semanto-syntactically non-equivalent, but
(56P) does appear as the actual translation of (56). At this point we do
not intend to discuss all the implications of examples such as (56) and
(56P), especially for text-bound contrastive studies (cf. Chapter III). In
this chapter we are not dealing with such obvious deviations from se-
manto-syntactic equivalence. They usually result from the translator's
conscious attempt to preserve invariables other than semanto-syntactic
in his translation and only marginally from faulty competences. In any
event, such conscious deviations from equivalence do not constitute
counterexamples to the identical-deep-structure hypothesis. On the con-
trary, they confirm the empirical reality of equivalence as manifested in
the translator's awareness that his translation does not convey all the
invariables of the original text.
In Chapter III, we said that semanto-syntactically equivalent sentences
are those which are the closest approximations to acceptable word-for-
word translations and their synonymous paraphrases (if indeed such
paraphrases exist). Let us rephrase this description of equivalent sentences
in the form of the following Constraint 2 (C2):

(C2) No two sentences will be assigned the status of semanto-syntactic

equivalents unless they are the closest approximations to accept-
able word-for-word translations and their synonymous para-
phrases, if such exist.
The two constraints seriously restrict the scope of data to be accounted
for by a possible theory of equivalence. Furthermore, since it will be
required that equivalent sentences have an identical deep structure, the
semanto-syntactic equivalence will be constrained in the following two
ways:
(1) On the surface it will be constrained lexically (by C t ) and structurally
(by C2).
 Towards a theory of semanto-syntactic equivalence 163

(2) In the sentence-semantic structure it will be constrained by the re-

quirement that equivalent sentences (in the sense of (1)) have identical
sentence-semantic structure.
In view of these constraints on semanto-syntactic equivalence, Bou-
ton's examples involving English active and passive sentences and their
Finnish counterparts no longer constitute counterexamples to the iden-
tical-deep-structure hypothesis. More precisely, these examples do not
jeopardize its empirical validity. Only (28) and (28F) and (30) and (30F),
respectively, are equivalent in the constrained sense, (29) and (31) are not
equivalent to (28F) and (30F), respectively, since they are not the closest
approximations to grammatical word-for-word translations and, as was
said earlier on, they manifest different co-occurrence and selectional
restrictions. Therefore, (28) and (28F) will be derived from some identical
semanto-syntactic structure and so will (30) and (30F), while (29) and
(31) must be derived from some different structures.
In English (28) and (30) are derived from a similar semanto-syntactic
structure, and so are (29) and (31). The only difference in both pairs is
that in (28) and (29), but not in (30) and (31), the agent is specified.
Thus, on the one hand we have:
(28) Tom broke the window.
(29) The window was broken by Tom.
with the specified agent, and
(30) Someone broke the window.
(31) The window was broken (by someone).
with an unspecified agent.
On the other hand, the four sentences can be grouped pairwise,
according to the underlying semanto-syntactic structure, thus:
(28) Tom broke the window.
(30) Someone broke the window.
realizing the semantic structure.
SI: A G E N T VERB PATIENT
and in contrast with
(29) The window was broken by Tom.
(31) The window was broken by someone.
realizing the semantic structure
164 Chapter VII

S2: PATIENT STATE PREPOSITION AGENT.

51 and S2 are, of course, drastically oversimplified, but they adequately
represent the relevant difference in the semantic content of the discussed
sentences. Only SI can be realized in Finnish, but unlike in English, in
Finnish SI can be realized in the surface structure exemplified by (28F)
if Agent is specified and only as (30F) if Agent is not specified. Thus in
Finnish, if Agent is unspecified, it is obligatory deleted and the obligatory
"impersonal" transformation yields (30F). If Agent is specified, sentences
like (28F) result (cf. Kudzinowski 1978: 90: "The Finnish passive is not
in fact an impersonal class; one always associates it with some personal
agent" [translation is my own]).

52 is never realized in Finnish while the translations such as (31) —

(30F) constitute further examples of non-equivalent sentences functioning
as translations.
It might be added that in Polish the "impersonal" transformation is
optional, so that both
(57) Ktos wybil okno.
'Someone broke the window'
and
(58) Okno wybito.
'Window broken (by someone)'
are possible. However, unlike in Finnish, in Polish realizations of S2 are
also possible:
(59) Okno zostalo wybite przez Tomka.
'Window was broken by Tom'
(60 Okno zostalo wybite (przez kogos).
'Window was broken (by someone)'
It may also be worth noting that the English passives, having no
equivalents in Finnish, are not the only instance of semanto-syntactic
gaps. The so-called perfective tenses, which also exhibit properties of
stative constructions, are, for the most part, rendered in other languages
by means of constructions in which similar forms either do not appear
at all (as in Finnish and Polish), or they do not observe the restrictions
characterizing stative verbs (as in German and French). Therefore, Ger-
man sentences containing "Perfekt" tenses and French sentences contain-
ing "passe compose" must be derived from semantic structures which
 Towards a theory of semanto-syntactic equivalence 165

underlie agentive constructions rather than from semantic structures

underlying Stative constructions, as in the case with the English present-
perfect tense and possibly other perfective forms.
The problem with the English aspect (as well as with the passive
sentences in English and their Finnish translations) suggests another
important difference between translations and semanto-syntactic equiv-
alents, which consists in the fact that whereas some kind of translation
of any given sentence into another language is always available (even if
in the process of translation a great number of various linguistic properties
at various levels are not preserved), equivalent sentences across languages
are not always available due to the existence of semanto-syntactic gaps
in various languages. Of all possible sentence-semantic structures each
language realizes only a subset proper to that language. For instance,
English passive sentences do not have equivalents in Finnish; likewise
sentences containing stative (perfective) tenses can hardly be considered
as equivalent to any sentences in other languages. Such a constrained
view of equivalence lends further rigour to contrastive studies but also
severely limits their usefulness, at least in their present form.
The constraints on equivalence (C, and C 2 ) and the distinction between
semanto-syntactic equivalence and translation allow us to dismiss all the
examples discussed above, including Bouton's active and passive sentences
in English and their Finnish translations, as irrelevant to the identical-
deep-structure hypothesis.
All the remaining examples discussed by Bouton can be dismissed on
similar grounds. For example, the English translation
(61) Where have you been?
of the Persian question
(61 Pr) Az koja miyaid?
is certainly non-equivalent. The closest approximation to the grammatical
word-for-word translation would have been
(61') Where do you come from?
and (6Γ) rather than (61) should be regarded as the semanto-syntactic,
though not necessarily translational or statistical (see Chapter X), equiv-
alent of (61 Pr).
Finally, to invalidate the identical-deep-structure hypothesis, Bouton
quotes an extremely interesting example of answers to negative questions
in English in contrast with other languages, including Korean. 4 English
166 Chapter VII

negative questions are answered yes if the action mentioned in the ques-
tion actually took place and no if it did not. The answers to equivalent
questions in Korean correspond, respectively, to no and yes. In Korean
(and in Polish) the answerer affirms or denies the existence of the situation
to which the question refers. On the basis of these facts Bouton concludes
answers of opposite polarity, which seem to be responding to different
aspects of interrogative situation and in that sense have a different
meaning from each other, should not be assigned to a common un-
derlying structure merely because they are translation equivalents
(Bouton 1976: 158-159).
They should not be assigned a common structure indeed! And there
is nothing in the present proposal that would induce anyone to take such
an unmotivated step. Unfortunately, my knowledge of Korean is too
limited to enable me to analyse self-made examples, and I have no access
to native informants. Therefore, I shall confine the subsequent discussion
to the Polish material, trusting that the situation in Korean is similar in
the relevant respect and as such does not require a separate treatment
from the point of view of the theory of semanto-syntactic equivalence.
Bouton's examples concerning answers to negated questions in English
(and perhaps in Korean too) does not seem to be complete. The dialogues
(62) A: Didn't you go to school today?
B: Yes. (if Β did go to school)
and
(63) A: Didn't you go to school today?
B: No. (if Β did not go to school).
do not sound like genuine dialogues, and especially in the answer in (62)
appears to be vague and is likely to be followed by A's Yes, what?,
meaning Did you or didn 't you go to school?
In Polish, likewise, the plain Tak 'Yes' and Nie 'No' are insufficient
as answers:
(62P) A: Nie poszedles dzis do szkoly?
B: Nie, poszedlem. (if Β did go to school)
'No, I went'
(63P) A: Nie poszedles dzis do szkoly?
B: Tak, nie poszedlem. (if Β did not go to school)
'Yes, I did not go'.
 Towards a theory of semanto-syntactic equivalence 167

There are two ways of handling these examples, depending on what

one expects as correct answers in (62P) and (63P):
1. In my own idiolect the answers Tak 'Yes' to the question in (62P)
and the answer Nie 'No' in (63P) are also possible. Therefore, in my
idiolect answers in (62P) and (63P) could be equivalent to those in (62)
and (63), respectively. Otherwise, in those cases when Nie 'No' corre-
sponds to Yes in English and Tak 'Yes' to No in English, we would have
to recognize again the presence of semanto-syntactically non-equivalent
translations. In view of the distinction between semanto-syntactic equiv-
alence and translation, these instances would not constitute counterex-
amples to the identical-deep-structure hypothesis, especially in consider-
ation of the fact that that word-for-word lexical equivalents would also
be available.
2. In the case of those speakers in whose idiolects only Nie in (62P)
and only Tak in (63P) are acceptable as answers, the following solution
may be offered. It is true, as Bouton suggests, that

The difference in the response in the two languages seems to be that

the English speaker affirms or denies the existence of the situation
about which the question asks; the Korean [and the Pole] affirms and
denies the accuracy of the statement underlying the question (Bouton
1976: 158).

These affirmations and denials are in fact present in more complete

English dialogues, in which complete answers could be, respectively, Yes,
I did (meaning: Yes, I did go to school today) and No, I didn't (meaning:
No, I didn't go to school today) with the ellipses of the repeated material.
In Polish, indeed, the affirmation and the denial may refer to the form
of the question itself and to the statement implied in the question rather
than to the actual situation. Therefore, the Polish answer actually means
Nie to nieprawda, ze nie poszedlem do szkoly, bo poszedlem 'No, it is not
true that I did not go to school because I did go' and Tak to prawda, ze
nie poszedlem do szkoly, bo nie poszedlem 'Yes, it is true that I did not
go to school because I didn't' In real conversations, nobody ever gives
such full answers unless there is a reason motivating a complete answer.
However, such full answers, which presumably underlie the short ones,
do constitute equivalents across the languages. The shorter answers may
not be equivalent, but often function as translations. Competent bilingual
informants are well aware of this lack of semantic equivalence. However,
as was repeatedly emphasized before, the identical-deep-structure hy-
168 Chapter VII

pothesis is not intended to account for translation but for semanto-

syntactic equivalence.
The discussion of Bouton's counterexamples shows that the hypothesis
can be defended if certain constraints are imposed on equivalence, and
if translation performance is distinguished from bilingual informant's
competence in recognizing semanto-syntactic equivalents across lan-
guages. Therefore, it is natural to conclude that any theory of semanto-
syntactic equivalence which would assign different semantic structures to
equivalent sentences would be false since it would fail to explain the
bilingual informant's competence in distinguishing semanto-syntactically
equivalent sentences from non-equivalent ones.
In the next chapter, we shall outline a theory based on the asumption
that all semanto-syntactically equivalent sentences have identical semantic
structures. Thus, the identical-deep-structure hypothesis will serve as the
initial postulate (axiom) of the theory. To avoid terminological confu-
sions, we are going to abandon the term "deep structure" (which ought
to be used only in the context of Chomsky's standard theory), and we
shall use the self-explanatory term "semantic structure" (referring to
sentence semantics) or the term input structure. It ought to have become
clear by now that Chomsky's deep structure is a level of representation
considerably "shallower" or less abstract than the level of representation
which has emerged from the aforegoing discussion.
Chapter VIII

Contrastive Generative Grammar

At this stage we are ready to introduce the concept of "Contrastive

Generative Grammar" as a device explicating semanto-syntactic equiva-
lence as described in the previous chapter. 1
Since equivalent sentences are derived from identical semantic repre-
sentations, a claim which we take to be axiomatic, it necessarily follows
that whatever structural differences between them occur must be due to
the operation of various language-specific rules at various levels of der-
ivation in the respective grammars of particular natural languages. A
generative grammar of a particular natural language may be conceived
as a diversifying device, whose input are semantic representations of
sentences (or, briefly speaking, input structures), generated by the uni-
versal semantic component, and whose output are well-formed sentences
characterizing a particular language. The level at which two particular
equivalent sentences are diversified for the first time marks the extent to
which such sentences are similar: the lower (i.e., the less abstract) the
level of the first diversification, the more similar the equivalent sentences
are since they share more rules in their respective derivations (see also
Di Pietro 1971: 27).
Let us look at some examples. First, consider the following two
sentences in English and Polish, respectively:

(1) They painted the fence white.

(2) Pomalowali plot na biato.

These two equivalent sentences are first diversified at the level of the
categorial component, i.e. the level at which the rules assigning gram-
matical categories to various portions of semantic representations operate.
Both (1) and (2) have identical semantic structure consisting of the roles
Agent and Patient and of a Predicate. 2 The categorical component assigns
the categories Noun Phrase to Agent and Patient in both languages. In
contrast with Polish, however, a part of the English Predicate is assigned
the category Adjective, eventually realized lexically as white whereas in
170 Chapter VIII

Polish the corresponding part of the Predicate is assigned the category

(Prepositional) Noun Phrase. Thus, the first diversification of (1) and (2)
occurs at the level of the categorical component.
The sentences

(3) He was asked a lot of questions.

(4) Zadano mu mnostwo pytah.

are not diversified at the level of the categorical component since in both
of them there occurs an equal number of major grammatical categories
as realizations of various semantic roles and predicates. Thus, he of (3)
corresponds to mu of (4), a lot corresponds to mnostwo, of questions
corresponds to pytan, and was asked to zadano. Assuming that both
sentences are derived from an underlying structure which, when spelled
out in English would read as

(5) Somebody asked him a lot of questions

where no diversification occurs yet, we observe that (3) and (4) are
diversified for the first time at the level of syntactic transformations. In
order to generate (3) it is necessary to apply the English passive trans-
formation, and in order to generate (4) the Polish 'impersonal' transfor-
mation must be applied.
Finally, a pair of sentences like

(6) One of the signatures was illegible.

(7) Jeden ζ podpisow byl nieczytelny.

are diversified for the first time at the level of lexical insertions.
In languages which are genetically closely related (Slovak and Polish,
Spanish and Italian, etc.) diversifications at even later stages are possible.
Di Pietro (1971: 27) quotes a poem entitled La gloria del momento by
Juseph Tusiani, which is written in such a way that it can be either
Spanish or Italian. The only diversifications occur at the level of pho-
nological rules: the poem sounds different when read aloud. In writing,
only certain accentual marks identify it as Spanish.
Contrastive Generative Grammar is to explicate contrasts at various
levels of derivation and provide an explicit account of semanto-syntactic
equivalence. In order to fulfil these tasks, Contrastive Generative Gram-
mar must be capable of achieving some more specific aims. We state
these aims in the form of five postulates characterizing it:
 Contrastive Generative Grammar 171

(1) If L; L n is a set of natural languages, Contrastive Generative

Grammar must recursively enumerate sentences in any L; and Lj; this
means that for every sentence, the grammar must decide whether or not
the sentence has been generated either by G; or by Gj, where Gj and Gj
are generative grammars of Lj and Lj, respectively.

(2) For each sentence in L, and Lj, Contrastive Generative Grammar

must assign one or more structural descriptions (each ambiguous sentence
must receive as many structural descriptions as there are ways in which
it can be disambiguated).

These first two postulates characterizing Contrastive Generative Gram-

mar entail a definition of generative grammar for a particular language
Lj (cf. Chomsky 1965: 3ff.). The remaining three postulates define Con-
trastive Generative Grammar as a device different from something which
is merely a collection of generative grammars of η languages:

(3) For each pair of sentences in L, and Lj, Contrastive Generative

Grammar must determine whether these sentences are equivalent. At this
point, we recall again the distinction between semanto-syntactic equiva-
lence and translation drawn and described earlier (see Chapters II and
VII). It is to be expected that semanto-syntactically equivalent sentences
will have identical parts in their structural descriptions, which follows
from the assumption about the universality of input structure and from
the empirical observations that there exist similarities between specific
grammars. Therefore, Contrastive Generative Grammar must also do (4):

(4) For each pair of equivalent sentences in Lj and Lj? it must specify
those parts of the equivalent structural descriptions which are identical
and those which are not. In other words, it must note the level of
derivation at which the first diversification occurs. It will do so by
scanning the derivation, beginning with the semantic input and moving
by the successive strings, generated by successive rules, where any output
of a rule (except the terminal ones) constitutes input to a successive rule.
Let L 0 be original input, common for both Lj and Lj, and let R n and
R m be sets of rules in L; and Lj? respectively. Then for Lj, each I p _, will
be an input to a rule R p whose output is O p and each O p will serve as
input Ip to a rule R p + 1 , unless O p is terminal. In this way I 0 will be the
input to R,, whose output will be O,, which will be the input i! to R 2 ,
whose output will be 0 2 , etc. Similarly, for Lj? each output O q of R q will
serve as input I q to a rule R q + i , unless O q is terminal.
172 Chapter VIII

In order to fulfill postulate (4) Contrastive Generative Grammar will

have to inspect the derivational history of each equivalent sentence,
beginning with the semantic input I c . If a particular Ioi for a sentence S p
in Li also serves as input for a sentence S p in Lj, then the two sentences
are equivalent in accordance with postulate (3). The grammar then scans
further rules operating on Ioi to stop at the place where the output of a
rule R k in Lj is not identical with the output of a rule R k in Lj. It will
thus note that place in the derivation where the first diversification occurs.

(5) For each pair of pairs of sentences in L; and Lj, Contrastive Generative
Grammar may also determine the degree of similarity according to the
relation "more similar", where this relation is equivalent to the relation
"diversified at a lower level" Thus, for each pair of pairs of sentences S p
in L; <=> Sq in Lj and St in L; <=> Sk in Lj, where the double arrow represents
equivalence in terms of the identity of input structures, Contrastive
Generative Grammar must determine which of the two pairs is more
similar, i.e., diversified at a lower level of derivation. Postulate (5) cannot
be fulfilled without having all the rules of the grammar extrinsically or
intrinsically ordered. Since extrinsic ordering of the rules has been shown
to be superfluous in explaining facts about natural languages (see, for
example, Koutsoudas 1972: 88 — 96), the full realization of this postulate
is dependent upon the construction of a grammar in which all rules would
be intrinsically ordered. It is not clear whether such a grammar can be
constructed at all.

Although the existence of postulate (5) has no bearing either on the

theory of semanto-syntactic equivalence or on the theory of translation,
it may prompt research strategies for constructing hierarchies of diffi-
culties in second-language learning, e. g., it may clarify some problems
involved in negative and positive transfer (see Chapter IX). It may also
make it possible to calculate a "similarity index", i. e., the ratio of identical
and different rules employed in various languages (cf. Di Pietro 1971:
27).
In order to fulfill these postulates, Contrastive Generative Grammar
must be based on suitable theories of the languages to be contrasted.
Thus, in order to fulfill postulate (3), it must be based on a model of
linguistic description in which the semantic input is universal. In order
to fulfill postulate (4) it is, furthermore, required that all generative
grammars of particular natural languages should be based on category-
neutral semantic input represented in terms of a metalanguage free from
 Contrastive Generative Grammar 173

such notions as noun phrase, verb phrase, tense, modal, etc., since these
grammatical categories are neither universal nor semantic. Postulate (4)
can be fulfilled only if all generative grammars introduce grammatical
categories at the same level of derivation, i.e., the level intermediate
between the semantic representation and syntactic transformations. It is,
therefore, to be anticipated that the organization of each generative
grammar will require at least five levels of representation:
(1) semantic, where the fundamental semantic relations, i. e., the mean-
ing of sentences is represented in the form of a universal, category-neutral
semantic input to sentence derivation; (2) categorial, where language-
specific rules assign various categories, such as noun phrase, verb, adjec-
tive, tense, modal, etc., to various portions of the semantic representation;
(3) syntactic, where syntactic transformations arrange the major syntactic
categories (nouns, verbs, adjectives, and perhaps adverbs) in the linear
order in which they appear in actual sentences and introduce some of
the minor categories (prepositions, auxiliaries, and perhaps some ad-
verbs); (4) lexical, where lexical items are inserted from the dictionary, in
which lexical entries are defined in terms of syntactic frames of gram-
matical categories; (5) post-lexical, where "cosmetic" transformations
arrange minor syntactic categories in the linear order in which they appear
on the surface, provide words with inflectional endings, and introduce
word boundaries.
Anticipating our discussion in Chapter IX, let us note here that not
all languages seem to require an equal number of rules at these five levels
in order to derive sentences from the semantic representations. For
instance, there are languages remarkably poor in inflections (Chinese,
English), in which the number of "cosmetic" transformations seems to
be negligible in comparison with highly inflected languages (Russian,
Polish, Latin). Various "reduced" registers (telegraphese, baby-talk) and
pidgins seem to function without many syntactic transformations which
account for complex structures and multiple embeddings. Such languages
are derivationally "shallower" than fully elaborated codes. It is possible
that in such languages, lexical entries are inserted directly into semantic
structures rather than into outputs of transformational rules, which
account for most structural complexities of fully elaborated codes.
Another feature of Contrastive Generative Grammar is that it must
be based on text-grammars in order to remove ambiguity. No sentence-
grammar (i.e. a grammar generating sentences rather than their se-
quences and longer stretches of texts) is capable of serving as a foundation
for Contrastive Generative Grammar even in principle, because it does
174 Chapter VIII

not provide grounds for determining whether or not particular sentences

across languages are equivalent in those cases when they are ambiguous.
For example, given an ambiguous sentence in a language Li? Contrastive
Generative Grammar will be unable to match it with an equivalent
sentence in Lj without analysing the ambiguity of the sentence in L;. The
analysis may reveal that a sentence Sk in Lj is in η ways ambiguous, and
that a sentence Si in L, is in m ways ambiguous. The analysis may also
reveal that some of the η readings of Sk are equivalent to some of the m
readings of S; in some contexts but not in others. The determination of
which specific reading is involved cannot be effected without inspecting
stretches of texts longer than a sentence. Suppose, for example, that
Contrastive Generative Grammar confronts the English sentence

(8) Drinking water can be dangerous.

which is ambiguous in at least two ways. One reading roughly corresponds

to

(9) The drinking of water can be dangerous.

while the other meaning corresponds to

(10) Water for drinking can be dangerous.

In Polish (8) cannot be rendered by means of one, equally ambiguous,

sentence. One of the following two must be chosen:

(11) Woda do picia moze bye niebezpieczna.

(12) Picie wody moze bye niebezpieczne.

Without inspecting some disambiguating contexts in which (8) appears,

and which stretch beyond sentence boundaries, Contrastive Generative
Grammar will be unable to choose either (11) or (12) as the equivalent
of (8).3
In order to fulfill postulate (3), Contrastive Generative Grammar must,
furthermore, be based on a generative grammar which accomodates the
notions of reference and co-reference. The notion of reference must be
understood as an association obtaining between noun phrases of a natural
language and mental entities present in the language-user's mind. In
 Contrastive Generative Grammar 175

languages without noun phrases (if such indeed exist) 4 , this association
obtains between some other appropriate syntactic categories and respec-
tive entities in the language-user's mind. This concept of reference follows
from the realization of the fact that the relation between syntactic cate-
gories and what they refer to in the world at large is a composite one. 5
Thus, it is necessary to distinguish noun phrases themselves, entities in
the world at large, and an intermediate set of abstract entities inhabiting
the language-user's mind. The composite character of reference consists
in the fact that noun phrases are not directly associated with entities in
the world, but they refer to an intermediate set of conceptual mental
entities, which in turn can be related to entities in the world at large.
This concept of reference provides a uniform way of treating fictitious
and non-fictitious, living and dead, existing and extinct characters, ob-
jects, and events as there is no syntactic evidence in support of the view
that these various kinds of phenomena require a fundamentally different
semantic treatment within the realm of linguistics (cf. Sampson 1970b:
13). The differences can be formulated only in non-linguistic terms and
consist in the fact that some concepts can be associated with objectively
and currently existing phenomena in the world at large while some cannot.
This view of reference enables us to associate mental entities we
shall call them referents, with a set of integers from 1 to η in such a way
that each referent is labelled with a unique integer, the cardinality of the
set of referents formed in this way being definite at any given moment
for an individual language user and determined by his former linguistic
career. It becomes possible to associate each referent with one or more
linguistic expression by attaching respective indices to those portions of
derivations which underlie these expressions. 6 Information of this sort is
indispensable for discovering equivalent sentences across languages; it is,
therefore, relevant and necessary in constructing Contrastive Generative
Grammar (for more details see Krzeszowski 1979: 28 ff.).
Let us look at some English and Polish sentences in order to see how
the equivalence relation between them can be established only if larger
stretches of texts over which referential indices are established are taken
into account. For clarity, we shall use only examples in which the relevant
portion of the text appears in the form of a sentence immediately pre-
ceding the examined sentence, although this is by no means the only
possible situation. Let us consider sentences containing referentials, i.e.,
words which refer to other words in the text. On numerous occasions,
Contrastive Generative Grammar will be unable to decide whether two
sentences containing referentials are equivalent without finding which
176 Chapter VIII

words in the neighbouring sentences the referentials refer to, i. e. without

establishing co-reference. Confronting sentences like

(13) They do not need mending yet (Krzeszowski 1979: 21).

Contrastive Generative Grammar is incapable of deciding on the correct

choice of the Polish equivalent without examining the antecedent of they
which may occur in the previous sentence. If the antecedent of they is a
word such as clothes, then depending on which of the possible four lexical
equivalents of clothes in Polish is selected in the preceding sentence, one
of four personal pronouns will have to be selected:

(14) Ten ubior (masc.) jest calkiem nowy. On nie wymaga jeszcze
naprawy.
(15) Ta garderoba (fem.) jest jeszcze calkiem nowa. Ona nie wymaga
jeszcze naprawy.
(16) To ubranie (neut.) jest jeszcze calkiem nowe. Ono nie wymaga
jeszcze naprawy.
(17) Te ciuchy (plur.) sg. jeszcze nowe. One nie wymagajg. jeszcze
naprawy.

In order to select the correct equivalents of sentences containing

referentials, Contrastive Generative Grammar must match those sen-
tences with other sentences appearing in the same text and containing
words having the same reference as the referentials for which it seeks to
find equivalents. This procedure can only be effected if the information
concerning the referential identity of particular fragments of texts is
contained in the semantic inputs to those texts. This kind of information
cannot be presented at the level of syntactic representations since the
notion of reference is semantic in nature (Hiz 1968: 5 ff.). It follows that
Contrastive Generative Grammar cannot enumerate equivalent sentences
in two or more languages and thus fulfill postulate (3) if the semantic
structure of these sentences does not contain information concerning co-
referentiality of nominal concepts not only within the boundaries of
sentence derivation but also within the boundaries of a particular text.
In brief, a linguistic model which is to serve as the foundation for
Contrastive Generative Grammar must fulfill the following requirements:

1. It must contain a semantic base, generating universal semantic rep-

resentations as inputs to language-specific grammatical rules. The
 Contrastive Generative Grammar 177

semantic base serves as tertium comparationis for establishing semanto-

syntactic equivalence. Being universal, it must be free of language-
specific syntactic categories appearing at less abstract levels of repre-
sentation.
2. The semantic input must contain information about reference since
reference is a relevant concept in establishing semanto-syntactic equiv-
alence of sentences across languages.
The model cannot be restricted to generating sentences, but must be
capable of generating longer stretches of texts (sequences of sentences),
relevant in establishing reference and co-reference.

In Krzeszowski (1974, 1979), I suggested a model of linguistic descrip-

tion, which, in my view, met the requirements of Contrastive Generative
Grammar and, in contrast to other currently available linguistic frame-
works, could serve as the foundation for it. Specifically, the model was
based on a conversion grammar, that is, a device which enumerates well-
formed sentences and their sequences by mapping an infinite set of
semantic inputs onto an infinite set of well-formed sentences and their
sequences by imposing conditions of well-formedness upon semantic
inputs in terms of well-formedness of sentences, which constitute the
output of the grammar. Such a concept of output-monitored generative
grammar presupposes an infinite number of semantic inputs representable
in terms of, for example, graphs arranged in a finite number of patterns
but endowed with the property of self-embedding to ensure the infinity
of their combinations. Such graphs serve as inputs to a finite number of
conversion rules, whose purpose is to filter out ill-formed configurations
of graphs and convert well-formed configurations of graphs into surface
structure sentences and their sequences. In terms of a framework of this
kind, well-formedness of graphs is determined by their suitability to serve
as inputs to conversion rules, which have to be constructed in such a
way as to guarantee the enumeration of all and only well-formed sentences
and their sequences in a given language. Conversion rules select certain
configurations as inputs to derivations and assign grammatical categories
to particular portions of these configurations. In the present framework
the expression "conversion rules" embraces all types of rules required in
a grammar to convert semantic inputs (configurations of graphs) into
sentences of a particular language. These rules will include universal
semantic categorical rules, language-specific syntactic categorical rules,
syntactic transformational rules, lexical insertion rules, post-lexical trans-
formational rules, and morphographemic/morphophonemic rules.
178 Chapter VIII

All inputs to all conversion rules (derivations) are constructed accord-

ing to the following general patterns, which we shall call the original
configuration:

(18) ·Α

•ρ

In the original configuration, the dots are uniquely associated with

roles, which are primitive, axiomatic notions, solely characterized by their
positions on the original configuration. These roles are: the uppermost
A(gent), the one below A P(atient), the one below Ρ R(esident),
the three roles below R fL (from Locus), atL (at Locus) and tL (to
Locus) from left to right, respectively. Both fL and tL are mutually
exclusive with atL, when dominated by the same R. In other words,
neither fL nor tL can be dominated by the same R as atL.
This general pattern is characterized by the property of self-embedding,
which means that some configurations may function as referents in other
configurations. A configuration which functions as a role in another
configuration is called a subconfiguration. Subconfigurations are assigned
role labels according to the pattern of the original configuration. In other
words, the position of each dot can be occupied by a referentially indexed
subconfiguration as one of its elements, etc., in principle ad infinitum. All
subconfigurations receive not only role labels, but also referential indices.
Thus, it is possible to have the following configuration:
 Contrastive Generative Grammar 179

(19)

In (19) there is a subconfiguration in the position of Resident, and it

contains another subconfiguration in the position of Patient.
All configurations which constitute inputs to derivations are structured
according to the original configuration. Every configuration which is
accepted by the conversion rules as an input to derivations consists of at
least one nuclear subconfiguration and of a certain number of non-
nuclear subconfigurations and elementary roles. A nuclear subconfigur-
ation is a subconfiguration which contains no subconfigurations. An
elementary role is a role which is not a subconfiguration. A nuclear role
is a role within a nuclear subconfiguration. In (19) 3 is a non-nuclear
subconfiguration and Resident; 1 (Agent), 2 (Patient), 4 (Agent), 11
(Resident), 12 (at Locus), 13 (from Locus), and 14 (to Locus) are nuclear
roles; and 5 (Patient) is a nuclear subconfiguration.
Universal semantic categorical rules operate upon initial configurations
and assign role labels to elementary roles and to subconfigurations,
erasing those elementary roles which are not referentially indexed. The
universal categorical rules assign roles to dots and circles of every con-
180 Chapter VIII

figuration in accordance with the patterning of the original configuration.

They also assign some universal semantic categories, such as Manner (to
the Locus directly dominated by each nuclear subconfiguration), Time
(to the Locus dominated by the outermost subconfiguration), and either
Time or Place to every Locus of every subconfiguration, except the
nuclear and outermost subconfigurations. These rules, moreover, assign
an index to every referentially non-empty role:

α — which refers to the speaker T , is always assigned to the A within

the outermost subconfiguration and to all coreferential nodes;
β— which refers to the content of the speech act uttered by a, is
always assigned to Ρ and R (which are later conflated by another
rule) within the outermost subconfiguration;
γ — which refers to the addressee of the content of the speech act
('you'), is always assigned to the tL within the outermost sub-
configuration and to all coreferential nodes;
δ — which refers to the speech act, is always assigned to the R
labelling the outermost subconfiguration;
ε — which refers to the location of the speech act in time, is always
assigned to the Time Locus (TL) dominated by the outermost
subconfiguration (within every configuration) as well as to all
tTL's, fTL's, and atTL's;
φ — which refers to an unidentified but specific person (not T , not
'you'), that is, to an unidentified third person, may be assigned
to any role which has not been mentioned above.

Moreover, indices from 1 to η (and a few others which we shall ignore

here to simplify the description of the framework) may be assigned to all
the remaining, referentially non-empty roles in all configurations under-
lying a particular text.
All roles with no referential indices are erased, while coreferential roles
occurring in the same subconfiguration are conflated (like PRß above).
The resulting objects can be represented as branching trees, such as those
used in the standard theory. Thus, the first universal rules filter out a
large number of all possible configurations and in doing so they specify
a subset of all configurations, which can serve as inputs to subsequent
rules. They also construct diverging trees (which can also be represented
as labelled bracketed strings), serving as inputs to language-specific cat-
egorical rules. For example, the text in (20):

(20) I stroke a dog. The dog was small.

 Contrastive Generative Grammar 181

is derived from an initial configuration, which after the application of all

the relevant universal semantic rules can be represented in the form of
the following branching (diverging) tree:

(21) Σ

Αα Δ Ra Δ Aa Δ Ρ3Δ tTL6R3 AatL5AtTLEA tLyA atTLc

where Σ stands for 'sequence of sentences within one performative com-

plex' — in this case declarative, and where A's correspond to traditional
predicates. (21) can also be represented as a labelled bracketed string
(22):

(22) Z(R 8 (A a Δ PRß(R a Δ t U R J A . Δ P3] Δ tTL c

R 4 [R 3 Δ atL 5 ] Δ tTL c ) Δ tLY)) Δ atTL e )

which should be interpreted as referring to the following situation: The

sequence of sentences (Σ) places the speech act (R s ) at the moment of
speaking (atTL c ), that is, 'now', and it embraces the content of the speech
act (PRß) directed by the speaker (A a ) to the addressee (tL 7 ). Thus, (22)
can be spelled out as:

I(A a ) declare that I(R a ) assent to the proposition (tL,) that there
occurred the event (R 2 ) of I(A a ) stroking a dog (P3) at a certain time
before 'now' (tTL,.) and the event (R 4 ) of the dog (R 3 ) being in the
state of smallness (atL 5 ) at the time before 'now' (tTL c ).

It must be noted that the content of the two nuclear subconfigurations,

indicated by the square brackets, when spelled out in ordinary English
words, is restricted to I stroke a dog (R 2 ) and The dog be small (R 4 ),
182 Chapter VIII

respectively. This means that neither sentence modality nor tense and
aspect are included since they are, whenever necessary, present elsewhere
in the appropriate strings but outside nuclear subconfigurations. We
emphasize this point here since it will be crucial in understanding the line
of reasoning in the next chapter.
Subsequent language-specific, categorical, and transformational rules
assign grammatical categories to labelled objects such as (21). The process
is too complicated to be described briefly here, and in fact the details of
the derivation need not concern us. Suffice it to say that these rules
produce labelled bracketed strings (or trees) with all grammatical cate-
gories and the relevant syntactic structure in a given language. It is into
those strings that lexical items from the dictionary are inserted. At this
stage of the derivation I stroked a dog would look as follows (the places
where lexical items are inserted have been indicated:
(22) ZiRjiA. Pres-V PRß(Ra V tLNP,(S(RNP2((Def)-NP-(N3)) Past-V-ed stroke
the boy
(Indef) — NP — (N4))) # ))tLT)
a dog
Semantic representations in terms of nuclear subconfigurations, cor-
responding to simple logical propositions and various indexed roles and
their combinations, provide a universal framework whereby sentence
semantics of all types of languages can be represented. It is essential to
notice that all types of sentences, complex and compound, in various
moods, aspects, and tenses, involve nuclear subconfigurations. Thus, the
nuclear subconfiguration (Αα Δ P3), appearing in I stroked the dog is also
involved in Did I stroke the dog? I might stroke the dog. I have been
stroking the dog. I didn't stroke the dog. I said I would stroke the dog.,
etc. while the elements underlying these various forms appear elsewhere
in the relevant strings, but outside the nuclear subconfiguration. For
example, past, present and future tense are derived from tTL c , atTLB,
and fTLE, respectively.7 Thus

(23) John stroked a dog.

(24) John strokes a dog.
(25) John will stroke a dog.
would, respectively, have the following tense specification:
(26) E(R 5 (A a Δ PRp(R a Δ tL 1 (R 2 (A 3 Δ P4) Δ tTL£) Δ tLy)) Δ atTL £ )
(27) tL,(R 2 (A 3 Δ P4) Δ atTLE
(28) tL,(R 2 (A 3 Δ P4) Δ fTL e
 Contrastive Generative Grammar 183

English sentences containing verbs in the so-called continuous aspect

are derived from configurations containing a nuclear subconfiguration
functioning as at L; for either AR or PR co-referential with A or Ρ within
the nuclear subconfiguration. (It will be remembered that complex roles
like AR or PR are derived from co-referential roles within the same
(sub)configuration by virtue of a universal semantic rule conflating co-
referential roles within the same configuration). The entire subconfigur-
ation, in turn, functions as R in one of the three TL E 's to account for
the distinction between present, past, and future continuous forms. Thus
the sentences

(29) John is stroking a dog.

(30) John was stroking a dog.
(31) John will be stroking a dog.

are derived from

(32) tl,(R 2 (AR 3 Δ atL 4 (A 3 Δ P 5 )) Δ atTL c ...)

(33) tl,(R 2 (AR 3 Δ atL 4 (A 3 Δ P 5 )) Δ tTL E ...)
(34) tl,(R 2 (AR 3 Δ atL 4 (A 3 Δ P 5 )) Δ fTL E ...)

where 1 — John/is/was/will be stroking a dog.

2 — John be stroking a dog.
3 — John
4 — John stroke a dog.
5 — a dog.

When spelled out in ordinary English words (32) would look as follows:

(35) John be in the state of [John stroke the dog] at now.

(33) and (35) can be represented analogously.

Sentences containing the so-called perfect aspect are derived from

configurations containing a nuclear subconfiguration that functions as R
in tTL e , and the subconfiguration formed in this way functions as R in
atL co-referential with either A or Ρ within the nuclear subconfiguration.
Therefore, such sentences as

(36) John has stroked a dog.

(37) John had stroked a dog.
(38) John will have stroked a dog.
184 Chapter VIII

are derived, respectively, from:

(39) tl 1 (R 2 (R 3 (R 4 (A 5 Δ P 6 ) Δ tTL £ ) Δ atL 5 ) Δ atTL e )
(40) tTL e )
(41) fTL £ )
where 1— John hasjhadjwill have stroked a dog.
2— John have stroked a dog.
3— John stroked a dog.
4— John stroke a dog.
5— John.
6— a dog.
Generally speaking, the meaning of sentences with verbs in the Perfect
Aspect involves the indication of the current relevance of some past
action, which is formally represented in the fact that the subconfiguration
referring to such actions is placed not only in the tTL c but also in atTL E .
Consequently, the meaning of such sentences as (36) can be explicated
as:
(42) The fact that John stroked a dog is at John ( = relevant for John)
at now.
(42) is the spelling out of the semantic representation (39) underlying the
sentence (36).
Declarative sentences and their sequences are derived from configu-
rations beginning in:
(43) A a Δ PR p (R a Δ {fL/tL} (Χ)) Δ tLY
where the first Δ corresponds to the performative verb DECLARE, and
the second Δ corresponds either to ASSENT TO or DISSENT FROM,
depending on whether tL or fL is selected (for details and justification
of these terms see Krzeszowski 1979: 48 ff.)· The version with ASSENT
TO underlies all declarative positive sentences. The version with DIS-
SENT FROM underlies sentences with "modal" negation as distinct from
ordinary negation relating to one of the elements of the sentence. "Modal"
negation can be signalled by means of the surface performative verb
doubt, as in
(44) I doubt that John stroked a dog.
The superordinate sentence I doubt that can be paraphrased with
the negative morpheme not or η 't overtly present, as in
 Contrastive Generative Grammar 185

(45) I don't think that John stroked a dog.

On the other hand, the performative complex (involving the two A's)
can be deleted altogether, and the sentence can be realized as
(46) John didn't stroke a dog.
In all three cases, the resulting sentences are ambiguous in the same
variety of ways as all other negative sentences. From the point of view
of the performative analysis, however, it is more important to note that
sentences such as (46), that is, all negative sentences without performative
verbs present in the surface structure, are also ambiguous between the
"modal" reading of the negation and the "pure" negation. The "modal"
negation involves the act of denial on the part of the utterer of the
sentence and is represented in the semantic structure by the second
performative DISSENT FROM, with R a as its subject. This reading of
negative sentences can be paraphrased with the use of the surface-negative
performatives, such as doubt and don't think. The "pure" negation consists
in negating an element within the nuclear subconfiguration (proposition).
In this case, the second performative is ASSENT TO, with R 3 as its
subject. Such readings of negative sentences cannot be paraphrased with
the use of overt negative performatives.
The distinction made here reflects the difference in the attitudes of the
speaker to the contents of every sentence that he utters. The speaker
either believes that what he says is true, or that it is false. Therefore, any
negative sentence can either be an expression of doubt concerning the
truth of the proposition contained in that sentence or a plain report
concerning the non-occurrence of events, actions, or facts expressed by
that sentence. Observe that the two readings of every negative sentence
such as (46) differ with respect to presuppositions involved in each
reading. Thus (46) can be paraphrased either as

(47) I think it is false that John stroked a dog.

(Someone says he did.)
or as
(48) I think it is true that John didn 7 stroke a dog.
(Someone says he did not, or, in fact, John did not stroke a dog.)
Sentences containing "pure" negation are derived from subconfigur-
ations containing R(X) Δ fL(Y), where the role to be negated is specified
as Ri(X) which is co-referential with an element within Y, e.g.,
186 Chapter VIII

(49) R, Δ fl2(R3(A, Δ P4) Δ tTL £ )

where X is null since R] is elementary, 1 — John, 2 — John stroked a dog,
3 — John stroke a dog, 4 — a dog. This configuration is involved in the
derivation of
(50) John did not stroke a dog.
with a special emphasis on the word John.
It is also possible to have
(51) R 4 Δ fL 2 (R 3 (A, Δ P4) Δ tTLJ
which underlies
(52) John did not stroke a dog.
Another input underlying 'pure' negation is represented by
(53) R,(R 2 (A 3 Δ P4) Δ tTL £ ) Δ fL,(R 2 (A 3 Δ P4) Δ tTL e )
where 1 — John stroked a dog, 2 — John stroke a dog, 3 — John, 4 — a
dog. (53) underlies
(54) John did not stroke a dog.
uttered with "normal" stress and intonation.
When spelled out in ordinary English words, (49), (51), and (53) can
be represented as
(55) R1 John not involved in ^(ai John stroked P4 a dog)
(56) R4 a dog not involved in ^(ai John stroked P4 a dog)
(57) R , (John stroked a dog) not involved in ^John stroked a dog).
Note that the subconfigurations accounting for negated elements in
sentences resemble the subconfigurations underlying sentences with /
doubt that associated with the performative DISSENT FROM insofar
as the contextual features characterizing the respective A's involve R's
with the related fL's. This is a formal reflection of the meaning of
negation, which always relates either to a disassociation of an element
from the truth involving that element, as in (49), (51), and (53), or the
disassociation of the speaker from a certain truth, as in the sentences of
the type / doubt that discussed earlier on and derived from a version
of (43).
Positive questions and commands, as well as their sequences, are
derived from configurations with COMMAND as the first performative.
 Contrastive Generative Grammar 187

In the case of positive questions, the second performative is either SPEC-

IFY or YES/NO, depending on whether a specific or a general question
is involved. The following configuration underlies all specific (wh-)ques-
tions, that is, questions about a particular role:

(58) A a Δ PRp(AY Δ PRj Δ atLj(X)) Δ tLY

which, when spelled out in less abstract terms, gives
(59) / IMP (to) you (you SPECIFY (SOMETHINGi/SOMEBODY;)
in (X)),
where X is a subconfiguration containing a role i, co-referential with
SOMETHINGj/SOMEBODYj.
General questions ask for the confirmation or the denial of a subcon-
figuration underlying a certain statement. The appropriate configuration
is as follows:

(60) Αα Δ PRp(AY Δ P ; (Rj Δ atL k (X))) Δ tLY

with the performative complex involving C O M M A N D and YES/NO.
When spelled out in less abstract terms, (60) can be represented as

(61) / IMP (to) you SOMETHING,, (you YES/NO (SOMETHING;

(SOMETHINGj/SOMEBODYj) be in (X))),
where SOMETHINGß is the direct object of the first performative and
introduces the subconfiguration containing the content of the command;
SOMETHING; is the direct object of the second performative and intro-
duces the subconfiguration underlying the proposition to be questioned;
SOMETHINGj/SOMEBODYj is an element of the questioned subcon-
figuration X (proposition), co-referential with an element within X and
constituting the focus of the question, signalled on the surface by a special
stress.
Although a complete description of Contrastive Generative Grammar
is available elsewhere (Krzeszowski 1979), we have taken the liberty of
presenting the appropriate semantic inputs in considerable detail since in
the chapter that follows we are going to explore a possible application
of this model in the prediction of certain errors that foreign learners
make in attempting to learn and use the target language. The explanation
of the possible linguistic causes of these errors crucially depends on the
proper understanding of how, according to Contrastive Generative Gram-
mar, sentences are derived from underlying representations.
Chapter IX

Error analysis, interlanguage, and Contrastive

Generative Grammar

Prediction of errors made by learners attempting to use a foreign language

was one of the motivating factors in contrastive studies, especially in the
United States. Lado hoped that contrastive studies would reveal similar-
ities and differences between the native (source) language and the foreign
(target) language, which would in turn make it possible to predict whether
positive or negative transfer is likely to take place:

If the expression, content, and association are functionally the same

in the native and the new languages, there is maximum facilitation.
Actually no learning takes place since the student already knows the
unit or pattern and merely transfers it (Lado 1964: 40).

In other situations, negative transfer or interference would presumably

take place.
Lado defined T R A N S F E R , either positive or negative, as the extension
of a source-language habit into the target-language, with or without the
awareness of the learner. He expected that similar places would lead to
cases of positive transfer, which would facilitate learning, while different
places would cause negative transfer or interference with the target-
language patterns, which would lead to errors having their source in the
source language (Lado 1964: 222). Thus, analysing foreign language
learners' errors would provide empirical verification of contrastive pre-
dictions.
Empirical research quickly invalidated this simplistic view: many sim-
ilar places across languages turned out to be troublesome and were
erroneously handled by foreign learners, and, conversely, different items
across languages would be assimilated in a more expeditious way. Briere
(1968) demonstrated the relative independence of hierarchies of difficulties
and of linguistic predictions. Politzer (1968: 35 — 43) produced examples
of parallel structures in French and English which cause enormous learn-
ing difficulties. The inadequacy of contrastive methods was partly to
190 Chapter IX

blame, but in many instances these experiments revealed the basic non-
identity of linguistic predictions based on the study of competence and
errors accessible through the study of performance, which, as is well
known, is affected by various other factors, escaping purely linguistic
approaches. Let us quote as another example the instance of the Russian
verb form ponjal, notoriously mispronounced by Poles as ponjäl, in spite
of the fact that the equivalent Polish word 'pojgl has the stress on the
first syllable, i.e., exactly where it falls in Russian.
It was, moreover, realized that if an error is a deviation from some
norm, it depends on the norm whether a particular linguistic form is
recognized as an error. If a Pole says I goes, he makes an error. The same
form in Black English may be considered perfectly acceptable (cf. Arabski
1979: 12).
Since the relation between contrastive studies and traditional error
analysis turned out to be more complex than Lado originally expected,
the latter continued to flourish as a more direct and more reliable source
of information about the learner's problems than contrastive studies,
which only indirectly and often inaccurately predicted certain types of
errors among all those that foreign-language learners make, namely only
those which have their source in negative transfer from the source lan-
guage. Traditional error analysis was thus an ad hoc attempt to deal with
the practical needs of the classroom teacher. It was confined to impres-
sionistic collections of "common" errors and their classification into
various categories, such as phonetic errors, grammatical errors, stylistic
errors, etc. In addition, traditional error analysis also attempted to dis-
cover frequencies of occurrence of various errors in texts. Conceived in
this way, it was more immediately useful in determining the sequence of
presentation of target-language items, in deciding on the relative degree
of emphasis and practice, and in devising remedial exercises than was
contrastive studies. More sophisticated traditional error analysis would
attempt to analyse the source of errors (e. g., source-language interference,
overgeneralization, etc. (see Arabski 1979: 31) and to evaluate the "seri-
ousness" of the error in terms of communication disturbances or the
distance from the norm (Olsson 1972; Enkvist 1973, 1977; James 1974,
1977).
Three major reasons determined the practical usefulness of traditional
error analysis vis-ä-vis contrastive studies:
1. Traditional error analysis did not suffer from the limitations of con-
trastive studies, which restricted the latter to errors caused by interlin-
 Error analysis 191

gual transfer; it discovered many other types of errors, like those

resulting from overgeneralizations, faulty teaching and learning strat-
egies.
2. Traditional error analysis pointed out actual, attested difficulties, most
directly connected with language usage while contrastive studies was
concerned with the study of competence.
3. Traditional error analysis was not beset by formidable theoretical
problems, which plagued contrastive studies, for example the problem
of equivalence.
This is why traditional error analysis, originally intended as a method
of testing constrastive predictions and enhancing the power of the ped-
agogical applications of contrastive studies (cf. Banathy — Madarasz 1969:
77 — 92), became an autonomous field of study, more useful than con-
trastive studies in syllabus planning and in designing materials for foreign-
language instruction.
A new approach to error analysis emerged when the focus was shifted
from teaching to learning, as a consequence of the realization that learning
strategies do not necessarily correspond to teaching strategies. This ob-
servation was based on the fact that learners persist in making certain
types of errors regardless of the amount of time and effort invested in
the teaching of relevant linguistic forms, and that the order in which the
learners master various linguistic forms does not always correspond to
the order in which these forms are taught.
Corder (1967 [1974]: 24) drew the important distinction between errors
of performance (slips, lapses) and systematic errors of what he called
"transitional competence", concentrating his attention on the latter. He
noticed that systematic errors are merely a realization of the learner's
transitional competence in the target language, and as such they are
evidence of the learner's success rather than of his failure. The process
of acquiring transitional competence is not fundamentally different from
the process whereby the child acquires his native tongue in that both
these processes involve the formation of hypotheses about language rules
and the subsequent testing of these hypotheses. As an illustration Corder
quotes the following dialogue:

Mother: Did Billy have his egg cut up for him at breakfast?
Child: Yes, I showeds him.
Mother: You what?
Child: I showed him.
Mother: You showed him?
192 Chapter IX

Child: I seed him.

Mother: Ah, you saw him.
Child: Yes, I saw him. (Corder 1974: 2 5 - 2 6 )
In this dialogue the child has tested three hypotheses:
1. Concord: subject verb in the past tense;
2. The meaning of the verb to show different from the meaning of to see;
3. Irregular past of to see.
Having received negative reinforcement on all the three hypotheses, the
child has formulated new hypotheses about the respective rules, which
eventually yielded linguistic forms conforming to the norms of standard
English. However, the original forms used by the child in the quoted
dialogue were manifestations of the child's transitional competence in his
native language.
According to Corder, in learning a foreign language the learner has to
test the following hypotheses:
Are the systems of the new language the same or different from those
of the language I know? And if different, what is their nature?
(Corder 1974: 27).
In testing such hypotheses, the learner goes through successive tran-
sitional competences, which Nemser (1971 [1974]: 55) called "approxi-
mative systems" According to Nemser "an approximative system is a
deviant linguistic system actually employed by the learner attempting to
utilize the target language" (Nemser 1974: 55).
Nemser's description of approximative systems was based on three
assumptions:
1. Learner speech at a given time is the patterned product of a linguistic
system, L a (approximative system), distinct from Ls and Lt and inter-
nally structured.
2. L a 's at successive stages of learning form an evolving series, L a j n,
the earliest occurring when a learner first attempts to use Lt (merger,
the achievement of perfect proficiency, is rare in adult learners).
3. In a given contact situation, the L a 's of learners at the same stage
of proficiency roughly coincide, with major variations ascribable to
differences in learning experience (Nemser 1974: 56).
Implicit in Nemser's presentation of L a 's is an additional assumption
that at some stage of development, L a 's of a particular learner cease to
 Error analysis 193

evolve and the system becomes stable at the most advanced stage available
to a given learner. If this is true, such a stabilized approximative system
can be studied and described by the same methods in which other natural
languages are described. We shall elaborate on the significance of this
conclusion later in this chapter.
Selinker (1972 [1974]) also hypothesized "the existence of a separate
linguistic system based on observable o u t p u t which results f r o m the
learner's attempted production of a T L n o r m " (Selinker 1974: 35). Se-
linker called such a system "interlanguage" He observed that there are
certain linguistic items, rules and subsystems, which foreign learners tend
to keep in their interlanguage regardless of their age and the a m o u n t of
time devoted to explanation and instruction in the target language.
Selinker designed the term "fossilization" to refer to this phenomenon.
It seems that Selinker's fossilized interlanguage corresponds to Nemser's
stabilized approximative system.
In attempting to describe interlanguage, Selinker identified five m a j o r
(and a few minor) processes determining its formation. In our account
of these m a j o r processes, we shall augment Selinker's original illustrations
with our own examples.
The following processes f o r m an interlanguage:

1. Overgeneralization of target language rules, which consists in extending

the use of certain linguistic forms to contexts in which they are not
appropriate, or in which they result in ungrammaticalities due to false
analogy with other forms. F o r example:

(1) What did he intended to say? (Selinker 1974: 38)

(2) I didn't took a big trip.

in which the past tense form is used where it could be logically expected
by analogy with the forms in declarative sentences. Also

(3) After thinking a little I decided to start on the bicycle as

slowly as I could as it was not possible to drive fast.

in which the speaker falsely extends the use of the verb to drive to all
types of vehicles, including bicycles, which one normally rides.

2. Transfer of training, which has its source in the way in which drills
and exercises are constructed and ordered. F o r example, Serbo-Croats
(and also Poles) have considerable problems with the he-she distinction,
and they tend to use he on almost every occasion, even when she is called
194 Chapter IX

for. The source of this persistent error cannot be attributed to the transfer
from the source language since in both Serbo-Croatian and in Polish,
unlike in Finnish and Hungarian, the analogous sex-based distinction is
codified in the grammar. Selinker attributes this error to the fact that
"textbooks and teachers in this interlingual situation almost always pres-
ent drills with he and never with she" (Selinker 1974: 39). My own
examples involve
(4) John is a sitting
in which the article a erroneously appears in front of the present participle
as a result of the prior overpractice of the pattern John is a boy, John is
a pupil, etc. and
(5) Housework is the work which is doing at home
in which the overpractised present continuous effectively interferes with
the less intensively practised passive forms.

3. Strategy of target-language learning. The general strategy is that of

simplification (cf. Widdowson 1975: l l f f . ) . Foreign-language learners
tend to reduce the target language to a simpler system and make errors
typically involving omission of various grammatical formatives, articles,
and sometimes function words. For example,
(6) Swimming is 0 very interesting and nice kind of sport.
(7) It is not to 0 done.
(8) This man kill-0 the most dangerous lion.
Another manifestation of this type of strategy is the coalescence of
aspect, as in
(9) Shakespeare have been born in 1564.
(10) I have read many plays by Shakespeare.
in which the simple and perfect aspects coalesce to produce a simpler
system of tenses.

4. Strategies of communication involve yet another form of simplification.

The errors in this domain usually have their source in the learner's
conscious or unconscious conviction that one cannot waste too much
time reflecting on how to say things even if one knows how to say them
properly. This leads to ignoring many grammatical phenomena which do
not affect communication. These include the plural number of nouns,
 Error analysis 195

certain function words, and certain conjunctions. A spectacular example

was uttered by a fluent speaker of English, under considerable strain,
when he was trying to get a ride when hitch-hiking in Sweden:
(11) Do you going nach Stockholm?
The strategy of communication influenced by the lack of time for reflec-
tion caused by the immediate necessity to get a lift, prevented the speaker
from making use of the grammatical resources that he had already
acquired. The example involves other processes, such as false analogy,
transfer from German, etc.

5. Transfer from the source language, which is the familiar process of

interference of the source language patterns. Examples are numerous and
varied. Among them one finds
(12) I bought yesterday five electric bulbs.
in which the Polish word order has influenced the speaker's production
in English. Also
(13) This things sometimes haven't in the shop.
where the Polish word order is accompanied by the faulty use of haven't,
which must be attributed to the analogous use of the verb miec in Polish
negative sentences.
Using Selinker's ideas as a starting point for his own interesting
speculations, Widdowson claimed that
all of the processes which Selinker refers to are tactical variations of
the same underlying simplification strategy and in general error
analysis is a partial account of basic simplification procedures which
lie at the heart of communicative competence (Widdowson 1975: 12).
According to Widdowson, simplification is a result of an attempt to
adjust the language behaviour in the interests of communicative effect-
iveness. Simplification may, therefore, affect any stage of interlanguage,
and it may involve
a movement away from the reference norm of the standard language
(or even a particular interlanguage) so as to arrive at forms of speaking
judged to be dialectally appropriate in certain contexts of use (Wid-
dowson 1975: 15).
Simplification presupposes the existence of non-simplified or more
complex forms of language. "Simplified" is a property of language by no
196 Chapter IX

means restricted to the foreign learner's interlanguages. It also charac-

terizes such types as pidgins and Creoles (De Camp 1968; Valdman —
Phillips 1975), "baby-talk", "lover's talk", and "foreigner talk" (Ferguson
1964, 1975) and various other types of "reduced registers", such as
"telegraphese", technical descriptions, and the language of the language
of instruction (Corder 1975: 5 - 7 ) .
It is interesting to acknowledge some results of research into the
structure of all these "simplified" codes. It appears that all of them share
some structural similarities, such as an extremely simple, or non-existent,
morphological system, a comparatively rigid word order, a reduced system
of pronouns, a small number of grammatical function words, a reduced
use of the copula, and the absence of articles (and to a smaller extent of
deictic words). According to Valdman (1975: 23): "pidgins are deriva-
tionally shallower [italics supplied] than other natural languages and
reflect more closely universal cognitively-based deep structures" In view
of the overt similarities between all types of "simplified" codes, Vald-
mann —Phillips' contention can be extended to cover them all. We shall
presently return to this observation in the context of Contrastive Gen-
erative Grammar.
Corder (1975) expresses an unorthodox view on the role of simplifi-
cation in language learning, taking Valdman — Phillips' reasoning to its
logical conclusions. He suggests that it might be possible to regard
"standard" codes as "elaborated" forms of "basic" simple codes such as
pidgins, Creoles, interlanguages, and all types of "reduced" registers. His
proposal assumes the existence of some universal process of elaboration
or complication involved in all types of language learning. Though
universal, the process would, of course, involve language-specific "com-
plication rules" The development of a pidgin into a Creole would be a
case of progressive complication. Another case would be the development
of approximative systems into an interlanguage closely approaching the
target language. All types of "reduced" registers used in appropriate
situations would be instances of fossilized intermediate approximative
systems or institutionalized, stereotyped stages in the process of compli-
cation towards the standard version of the language.
Contrastive Generative Grammar as a device generating equivalent
constructions seems to be well suited for providing a fairly explicit account
of "reduced" codes, including interlanguages, vis-ä-vis fully elaborated
codes. It can provide a linguistic account of how a simplified code can
be related to a more complicated code, by spelling out formal compli-
cation routes by which the latter can be derived from the former. 1 It will
 Error analysis 197

be remembered that Contrastive Generative Grammar involves two kinds

of lexicalizations (lexical insertions). Minor lexicalizations insert function
words and grammatical formatives, and they constitute part of syntactic
transformations. Major lexicalizations take place after syntactic trans-
formations and insert content words. In fully elaborated codes, all major
lexicalizations take place after all the relevant syntactic transformations.
In various types of "simplified" codes, notably also in interlanguages,
major lexicalizations may occur at earlier stages of derivation. In extreme
cases major lexicalizations may occur at the level of semantic represen-
tations. In such cases, "cosmetic" transformations, accounting for agree-
ment, concord, government, and generally for inflections, are either
completely left out, or they apply to a limited extent. Major lexicaliza-
tions, as it were, "petrify" the syntactic structure of constructions since,
at least in Contrastive Generative Grammar, once major lexical items are
inserted, no syntactic transformation can apply. Therefore, the degree of
syntactic complication (elaboration) of a construction in an interlanguage
directly depends on the stage of derivation at which major lexicalizations
take place. Early lexicalizations result in the syntactically simplest con-
structions, characterized by the absence of function words (inserted by
syntactic transformations which in such cases cannot apply) and a rigid
word order, often different from the way in which words are arranged in
fully elaborated codes. Within the framework of Contrastive Generative
Grammar, complication can be defined as the gradual shift of the place
at which major lexicalizations occur, from the deepest level of represen-
tation, i.e., the semantic level, to the level which constitutes input to
"cosmetic" transformations, and where major lexicalizations occur in
fully elaborated codes.
Among the first sentences in child speech, Brown (1970: 220) quotes
Adam put, Eva read, Mummy lunch, dog bite, hit ball, etc., which represent
various relations such as agent —action, action —object, agent —object,
etc. Such examples illustrate extreme cases of early lexicalizations occur-
ring in the simplest codes, whereby lexical items are inserted directly into
semantic representations, or more specifically into nuclear subconfigur-
ations, thus causing "petrification" of their structure. In these cases,
syntactic transformations do not apply, so that all other distinctions
represented by the material outside the nuclear subconfiguration is left
out of the derivation (cf. Chapter VIII).
In more elaborated codes, more non-nuclear material is involved in
the derivation of sentences. As an example, let us consider w/z-questions
in English. It will be recalled that in the framework of Contrastive
198 Chapter IX

Generative Grammar all w/j-questions are derived from configurations

marked in Chapter VIII as (58) and (59), which we now repeat for
convenience as
(14) A a Δ PRß(AY Δ PR^ Δ atLj(X)) Δ tLY
and
(15) I IMP (to) you {you SPECIFY (SOMETHING./SOMEBODYj)
in (X)).
A full specification of the semantic representation for a sentence like
(16) What did John touch?
can be formulated as
(17) Αα Δ PRß(AY Δ PR; Δ atL 2 (R 3 (A 4 Δ Ρ,) Δ tTL e ) Δ tLY) Δ atTL e
where 1 — the object touched by John about which the question is
asked (SOMETHING^SOMEBODYO
2 - JOHN touched SOMETHING/SOMEBODY
3 - John touch SOMETHING/SOMEBODY (nuclear subcon-
figuration)
4 — John
In ordinary English words (17) can be spelled out as
(18) AaI command tLyyou atTu that A7you specify
PRisomething/somebody in atL2 the proposition involving
Ri[John touch something/somebody] at tTLi:some time before now.

In order to convert (17) into the fully elaborated question, several

operations must be performed by the rules at various levels.

1. At the categorical level, appropriate categorical rules will assign gram-

matical categories to various portions of (17). By virtue of these rules,
the categories Σ and S (in the case, the Σ-sequence consists of just one
S-sentence) will be assigned to R's placed in TL's, various roles will be
assigned the categorical status of NP's; some Δ'β will be categorized as
verbs while other Δ'β will be erased; tTLE's and atTL c 's will be, respec-
tively, assigned the status of past and present grammatical formatives;
and the interrogative morpheme Q will be adjoined to the left of PR,
(for details see Krzeszowski 1979: 62 f f ) . As a result of these operations,
(17) becomes (19):
 Error analysis 199

(19) I(R 5 (A a V PRß(AY Δ V Q-PR, V atLNP 2 (S(RNP 3 (ANP 4 V

PNP,)past)#)) tL Y )present)

2. At the level of syntactic transformations, the following operations will

be involved:
(a) the deletion of the PRj co-referential with a role in the nuclear
subconfiguration. In our example, PR, and P) are involved, so the former
is deleted together with its verb. Simultaneously, Q is shifted to be
adjoined to that role in the nuclear subconfiguration which is co-refer-
ential with the deleted PR. In our example Q is shifted to P, in the
nuclear subconfiguration. Thus (19) becomes (20):

(20) E(R5(Aa V PR P(AY V atLNP2(S(RNP3(ANP4 V Q-

PNP,)past))#)) tLj)present)
(b) adjoining the tense morphemes of each Σ- and S- subconfiguration
to the verb within that (sub)configuration. Thus (20) becomes (21):

(21) E(R 8 (A a present-V PRß(AY V atLNP 2 (S(RNP 3 (ANP 4 past-V Q-

P N P , ) ) # ) ) tL,))
(c) preposing the element with Q to the initial position within the
nuclear subconfiguration. Thus (21) becomes (22):

(22) E(R 5 (A 2 present-V PR ß (A T V atLNP 2 (S(RNP 3 (Q-PNP, past-V

ANP 4 ))#)) tLY))

3. At the level of "cosmetic" transformations (after the insertion of lexical

items) introducing do to support the tense in those cases when the V is
not an auxiliary verb be or have. In this way (22) becomes (23):
(23) I(R 8 (A a present-V PR ß (A a V atLNP 2 (S(RNP 3 (Q-PNP, past-do
V ANP 4 ))#)> tLY)).2

The successive application of all these operations (categorical, syntac-

tic, and "cosmetic") mentioned in (1), (2), and (3) yields an increasingly
complex or "thick" structure as more and more rules are involved in the
derivation of the question. However, at each of these steps the derivation
can be terminated by the insertion of lexical items. In such cases, which
might be metaphorically called "premature lexicalizations", 3 we obtain
structures which are derivationally shallower as they involve fewer rules
than the derivationally fuller structures, and these shallower structures
resemble underlying semantic representations more closely. Thus, we can
200 Chapter IX

hypothesize that due to such "premature" lexicalizations the following

questions will be generated:
(24) (a) John touch? (lexical insertions at the semantic level
or at the categorical level before the
rule adjoining Q to the node co-ref-
erential with the deleted PR)
(b) John touch what? (lexical insertions before the rule ad-
joining tense to the verb and the rule
preposing the questioned element)
(c) John touched what? (lexical insertions before the rule pre-
posing the questioned element)
(d) What John touched? (lexical insertions before the rule pre-
posing tense)
(e) What touched John? (lexical insertions before the rule in-
troducing do)
Since the rules operating on each level are intrinsically ordered, the
rule adjoining tense may also operate after the rule preposing the ques-
tioned element. Therefore, one may also expect sentences such as
(f) What John touch? (lexical insertions after the rule pre-
posing the questioned element but be-
fore the rule adjoining tense to the
verb)
Thus, Contrastive Generative Grammar generates questions of various
degrees of derivational complexity. Some of these questions correspond
to questions actually attested in various "reduced" codes, including child's
speech and interlanguages. In the materials studied by Brown (1970) and
Ravem (1974), one finds types (a), (d), (e), and (f). Ozga (personal
communication) finds types (d) and (e) to be frequent among Polish
learners of English. It is perhaps slightly puzzling to realize that none of
the researchers found intermediate hypothetical questions representing
patterns (b) and (c), that is, those in which lexicalizations take place
immediately before the rule preposing the questioned element, except in
the case of that Brown calls "occasional" questions. 4 Although Contras-
tive Generative Grammar generates such structures as (b) and (c) as
potential sentences, it cannot explain why they do not actually occur in
the examined corpus. However, since the semantic material triggering off
the rule which introduces Q into derivations actually precedes in the
linear order from left to right the relevant nuclear subconfiguration in
 Error analysis 201

the semantic representation, and since after being adjoined to the ques-
tioned element in the nuclear subconfiguration, it is subsequently pre-
posed again, the latter two steps could be ignored altogether, while Q
would simply attract the questioned element from the nuclear subconfi-
guration. In view of the evidence presented by Brown, Ravem, and Ozga,
the appropriate rules could be reformulated in the following way: instead
of two rules — one that adjoins Q to the nuclear role, coreferential with
the deleted PR; and one that preposes that role to the initial position
within the nuclear subconfiguration one rule could be postulated. It
would perform three operations at once: it would delete the PR; to which
Q is originally adjoined, it would move Q to the initial position within
the nuclear subconfiguration, and it would prepose that role within the
nuclear subconfiguration which is coreferential with the deleted PR ; to
adjoin it to Q. In this way, the stage at which Q is placed in the nuclear
subconfiguration in a position other than the initial one would be elim-
inated, and the grammar would not generate (b) and (c) as intermediate
constructions between (a) and (d). Although this solution lacks internal
motivation, it is at least plausible in the framework of Contrastive
Generative Grammar, and it is motivated externally by the empirical
evidence furnished by Brown, Ravem, and Ozga.
This example reveals one of the weaknesses of Contrastive Generative
Grammar. It can accomodate many facts in the domain of performance
in an interlanguage, but it cannot predict those facts in an internally
motivated way. In other words, Contrastive Generative Grammar in its
present form, is too powerful. It is capable of generating a number of
constructions which are not actually found in performance. However, it
seems that many generative grammars suffer from the same shortcoming.
In terms of error analysis, the above discussion concerns various errors
which result from "premature" lexicalizations, and which consist in in-
serting lexical items into syntactic structures which have not been fully
elaborated (in comparison with fully elaborated codes). Such errors can
be defined as those forms in an interlanguage which are different from
the norm formulated in terms of generative rules characterizing the fully
elaborated version of target language, and which result from the omission
of various syntactic rules which "complicate" the structure. Contrastive
Generative Grammar accounts for the fact that these "errors" are similar
across various "reduced" codes and across interlanguages, regardless of
the source languages: lexical items are inserted at early stages of sentence
derivation into underlying structures which are similar in proportion to
the distance from the surface at which such lexicalizations take place.
202 Chapter IX

Thus, Contrastive Generative Grammar can at least predict that errors

resulting from such "premature" lexicalizations, involving the absence of
articles, function words, and other formatives derived from the material
outside the nuclear subconfiguration, will be similar in various interlan-
guages.
Another type of deviation from target-language norms that may be
handled by Contrastive Generative Grammar consists in the instances of
inserting lexical items of the target language into structures generated
according to the rules of the source language. For example, the grammar
of the Polish interlanguage of English may generate such sentences as
(25) I wanted that Peter opened the door
in which English lexical items have been inserted into the structure
generated by the Polish syntactic rules which generate such sentences as
(26) Chcialem, zeby Piotr otworzyl drzwi.
Errors of this sort are traditionally described in terms of transfer of
patterns from the source language and are to a large extent language-
specific. They result from substituting complication routes of the source
language for complication routes of the target language and can be
predicted by Contrastive Generative Grammar on the basis of a com-
parison of particular equivalents it enumerates.
As a model of bilingual competence, Contrastive Generative Grammar
cannot make any claims about actual occurrence of these various forms
in the performance of particular interlanguage users, nor can it pretend
to be a model of psycholinguistic processes which attend on the actual
production of various deviant utterances in particular interlanguages (cf.
Chesterman 1980: 2 1 - 2 3 ; James 1980: 54). It can, however, predict
potential errors and suggest falsifiable hypotheses about the occurrence
or non-occurrence of particular forms in particular interlanguages (cf.
Ravem 1974:128 ff.). The formulation and verification of such hypotheses
falls within the domain of error analysis proper.
Chapter Χ

Quantitative contrastive studies1

Quantitative contrastive studies belong to text-bound contrastive studies.

As was pointed out in Chapter II, quantitative contrastive studies may, but
do not have to, use translations as their primary data. It is possible to
conduct quantitative contrastive studies on texts which are not translations,
and which are not systematically equivalent nor even equivalent in any
sense of the word "equivalent" (cf. Chapter II). The constituent texts of 2-
texts used for quantitative contrastive studies may be chosen for comparison
only on the grounds that they represent the same register, or the same style,
or the same literary genre, or on any other grounds which provide the
common platform of reference motivating the comparison. Using such
texts, one can look for statistical equivalence, represented by certain fixed
patterns of frequencies of occurrence of various linguistic forms character-
izing a particular register, style, genre, etc. These patterns of frequencies
characterize specific "styles" in particular languages by following a certain
statistical norm. Deviations in plus or in minus from such a norm may be
connected either with non-native performance, as we shall see later, or they
may be intended, for example, in parodies. 2
Systematic contrastive studies are incomplete and inadequate unless
supported by quantitative data at all levels of linguistic analysis. 3 As an
example let us consider the case of personal pronouns across languages.
In Chapter IV, systematic equivalence was established between English
and Polish paradigms of personal pronouns: / — ja, you ty/wy, they
— onijone, etc. Correct, as far as it goes, this generalization is inadequate
for several reasons. For example, it does not say anything about the
appropriateness of specific pronouns in specific situations. Above all, it
does not say anything about the relatively low frequency of occurrence
of personal pronouns in Polish texts in contrast to a relatively high
frequency of occurrence of personal pronouns in English texts.
Systematic contrastive generalizations prove to be even less adequate in
all those numerous cases in which a particular construction in one language
has more than one nearly synonymous equivalent in another language.
204 Chapter Χ

Some of these equivalents may turn out to be nearly congruent (cf. Chapter
VI). In such cases, systematic contrastive studies would predict no learning
problems but rather something in the nature of positive transfer. In reality,
however, different frequencies of occurrence of such congruent forms in
texts could result in errors consisting in using congruent forms either too
frequently or too rarely. In the former case, we would be dealing with errors
of abundance, in the latter case, with errors of avoidance. In either case the
frequency of use by non-native users of the language would be different
from the frequency of use by native users.
Let us consider the following example. The French construction apres
etre revenu can be rendered in Czech (Becka 1978: 129) and in Polish in
a number of ways, none of which is a congruent equivalent of the French
original:

Prepositional Phrase: Czech po navratu

Polish po powrocie 'after the return'
French apres le retour
Active Past Participle: Czech vrativ se
Polish wrociwszy (si§) 'having returned'
French etant revenu
Subordinate Clause: Czech kdyz se vrätil
Polish kiedy wrocil 'when he returned'
French quand il etait revenu
Main Clause: Czech vrätil se a...
Polish wrocil i 'he returned and
French il etait revenu et
Becka discusses this example in the context of quantitative contrastive
stylistics and observes that "In the translation process a grammatical phe-
nomenon has become a stylistic one" (Becka 1978:129). As a stylistic issue,
it is primarily a matter of the frequency of use. All these constructions in
the three languages are nearly synonymous and can be arranged in pairs
which exhibit a certain degree of formal resemblance. Systematic contrastive
studies would assign to those pairs the status of equivalence and would
discover formal differences resulting from the application of non-identical
rules involved in the derivation of these constructions. Some valid gener-
alizations could thus be made about systematic differences and similarities
of specific constructions across languages. Such systematic analyses would
provide no clues to the ways in which native speakers employ these equiv-
alent constructions in actual texts. The fact that there is an equivalent or
even a congruent construction in a target language does not necessarily
 Quantitative contrastive studies 205

mean, and in fact rarely does, that it is employed with equal willingness by
native users of the language. In reality, frequencies of occurrence of equiv-
alent constructions are often quite different. For example, a native user of
French would seldom use apres le retour in the contexts in which Czech
would have po navratu or Polish would have popowrocie. Although in terms
of systematic contrastive studies such pairs would constitute semanto-syn-
tactic equivalents, in statistical terms they would probably have to be con-
sidered as non-equivalent.
In her contrastive analysis of simple sentences in Czech and English,
Duskovä notes that
One of the features in which Czech and English differ is the tendency
of English to what has been called complex condensation, viz. the use
of non-finite verb-forms where Czech has subordinate clauses (Dus-
kovä 1978: 83).
The differences, according to Duskovä, may be due to systematic
differences in those instances in which English possesses certain gram-
matical forms that Czech lacks (e. g., the gerund), or they may be due to
more restricted uses of such Czech forms as infinitives, participles, as
well as perceptive and causative verbs.
In quantitative terms, "more restricted uses" may suggest lower fre-
quencies of occurrence in texts although this is not always the case. For
instance, the fact that in Polish infinitives do not appear as object
complements in the construction called accusativus cum infinitivo nor in
their passive counterparts called nominativus cum infinitivo means that
infinitives in Polish have "more restricted uses" or, strictly speaking, that
they have a more limited distribution. But it does not mean that infinitives
are less frequent in Polish texts than in English texts. As a matter of fact,
according to my own investigations, about every 65th word in a contin-
uous Polish text is an infinitive, while only about every 90th word in a
continuous English text is a marked infinitive (with to). Therefore, Dus-
kova's suggestion that the statistical tendency in English to complex
condensation may be due to systematic differences must be regarded with
caution. Whether or not there is a direct correlation between the number
of various infinitival constructions and the frequency of occurrence of
infinitives in continuous texts is a matter that remains to be investigated. 4
Quantitative contrastive studies provide means of approaching the
poorly investigated phenomena connected with errors of avoidance and
errors of abundance, which cannot be handled by systematic contrastive
studies since the very concepts of avoidance and abundance are quantitative
206 Chapter Χ

rather than qualitative in nature. Interestingly enough, it has been observed

that differences in the frequency of use of certain constructions are con-
nected with native vs. non-native use of a language. For example, Graczyk
(n. d.) noted the tendency of Polish learners of English to overuse extraposed
subjects as in It is a fact that it is hard to convince people that there is much
sport in boxing. Although grammatical, this sentence suffers from clumsi-
ness, which could be remedied by writing Boxing may jean be a good sport
though few people know/admit it. According to Graczyk, one finds fewer
extraposed subjects in compositions written by native speakers of English.
This sort of overuse of a favourite construction could be labelled "error of
abundance" One can easily envisage the opposite tendency to avoid certain
types of constructions. If a foreign-language learner manifests this tendency
in his writing and/or speaking, he is committing an "error of avoidance"
One such error, involving a comparatively low frequency of a certain type
of nominals in texts written by Polish learners of English is discussed below.
In either case, the learner deviates in plus or in minus from a certain statistical
norm which characterizes native performance in a particular language. To
ascertain such an error, one has to perform a quantitative contrastive study
of texts written by native users of a particular language and by a non-native
user of the same language and compare the frequencies of use of the inves-
tigated forms. In other words, one has to conduct a quantitative contrastive
study of texts written in the target language (by native users) and texts
written in an interlanguage (by foreign-language learners). Whatever dif-
ferences in frequencies of occurrence of various linguistic forms in native
and non-native productions will be thus discovered, are of great interest in
themselves, as they can prompt teaching strategies that could lead to the
elimination of both types of frequency errors.
However, it is at this point that one begins to wonder what possible
impact native frequency patterns might have on foreign-language per-
formance. It would seem plausible to assume that the performance in a
foreign-language is determined by the frequency patterns of distribution
of systematically equivalent forms in source-language texts. One could
hypothesize that this is indeed the case and then proceed to verify this
hypothesis in the course of statistical research.
Let us take a look at one such study, whose results were at first surprising
in that they apparently contradicted the contrastive predictions but in the
end provided a reasonable explanation. In the course of the research, I
investigated frequencies of occurrence of Gerundive and Action Nominals
in M.A. theses on linguistics written by two Polish students (both non-
native, average students of the English Department at the University of
 Quantitative contrastive studies 207

Gdansk) and in a Diploma thesis written by a native user of English. For

convenience, I shall refer to the two theses written by the Polish students
as PEl and PE2, respectively, and to the thesis written by the British student
as BE.
For the purpose of this particular research, no distinction was made
between Gerundive Nominals and Action Nominals ending in -ing as de-
scribed by Lees (1963), so the two kinds of nominals were described jointly.
However, Action Nominals ending in suffixes other than -ing, as well as full
nouns ending in -ing (Regular Gerunds in Lees' misleading terminology),
were excluded. The frequency of occurrence of gerundive and action nom-
inals were calculated in relation to the number of sentences constituting the
three theses (citations and illustrative examples had been omitted from the
count). Interestingly enough, the average length of sentences was approx-
imately the same in each of the three theses and amounted to 23.9 words
per sentence in BE, 24.3 words per sentence in P E l , and 24.6 words per
sentence in PE2. Therefore, the relatively high number of gerundive and
action nominals in the thesis written by the native speaker (BE) can by no
means be attributed to the hypothetically greater length of sentences, which,
perhaps, might have been expected in a thesis written by a native user of
English. On the contrary, it turns out that the sentences in BE tended to be
slightly shorter though the difference had no statistical significance.
The frequency of occurrence of gerundive and action nominals in BE
was significantly higher than in PEl and PE2. Whereas in BE the number
of such nominals averaged 0.26 per sentence, in PEl it amounted to only
0.09 and in PE2 to even less, viz., 0.07. Characteristically, PE2 was graded
as the lowest passing ("satisfactory") while PEl was evaluated one grade
above the lowest passing ("good"). Table 2 contains the relevant figures.

Table 2. Comparison of English written by a native speaker (BE) and by two non-native
speakers ( P E l , PE2)

BE PEl PE2

Number of sentences 300 231 341

Number of gerundive and action nominals 78 21 25
Gerundive and action nominals per sentence 0.26 0.09 0.07
Words per sentence 23.9 24.3 24.6

In order to see to what extent the frequency of occurrence of equivalent

Polish nominals in Polish continuous texts might have negatively influ-
enced the performance of the two Polish students in English (both
208 Chapter Χ

committed evident errors of avoidance), a quantitative contrastive study

of translationally equivalent (cf. Chapter II) texts in English and Polish
was undertaken.
As in English, one encounters problems with the classification of
nominals in Polish. The so-called nomina actionis in Polish basically fall
into two groups: substantiva verbalia and substantiva deverbalia. Sub-
s t a n t i a verbalia differ from substantiva deverbalia in that the former
are morphologically regular as they take regular suffixes -anie, -enie, -cie
and can be formed from nearly all verbs (with very few exceptions, which
can be easily listed). The deverbalia are formed from a fairly limited
number of verbs and abound in alternative forms which do not obey any
obvious rules. In comparison with English, substantiva verbalia are closer
to Lees' Gerundive Nominals as they are more "verby" in nature while
the deverbalia are closer to Lees' Action Nominals and Regular Gerunds
by being more "nouny" 5
Since the Polish nomina actionis fall into two groups while correspond-
ing English nominals fall into three groups, from the point of view of
quantitative contrastive studies, there arises the problem of what to count
and what to compare. The situation can be visualized thus:
Polish nominals:
Substantiva verbalia (verbal substantives) — regular
pisac pisanie
mowic mowienie
Substantiva deverbalia (deverbal substantives) — irregular
pisac pisarstwo
mowic mowa
Lees' English nominals:
Gerundive nominals — regular
write writing
speak speaking
Action nominals — regular
write the writing of
irregular
discuss discussion
Regular gerunds — regular
fill — one filling, two fillings
 Quantitative contrastive studies 209

It appears that gerundives are the least "nouny" while full gerunds are
the most "nouny" (cf. Ross 1973). In Polish, verbal substantives are less
"nouny" than deverbal substantives. The comparative problem consists
in the fact that, on the one hand, Polish verbal substantives seem to
combine some characteristic features of both English gerundive nominals
(teaching English) and action nominals (the teaching of English) while on
the other hand, Polish deverbal substantives combine some features of
English action nominals and of regular gerunds. The situation is further
complicated by the fact that within the Polish verbal substantives (in
-anie, -enie, -cie) it is possible to distinguish between at least two subca-
tegories, of which one is more "nouny" than the other. Thus, alongside
such verbal nouns as czytanie 'reading' and mowienie 'speaking', we also
have such regular nouns as wprowadzenie 'introduction', uzycie 'currency;
use', which resemble English irregular action nominals, but unlike the
latter they retain regular suffixes. The most "nouny" Polish nominals,
i.e. full nouns called deverbal substantives, have irregular and unpre-
dictable suffixes of various sorts, which accompany morphophonemic
and semantic shifts. In that latter respect, they resemble English irregular
action nominals rather than full gerunds, which regularly take the suffix
-ing and form regular plurals with -s. If we assign numerical indices to
the three kinds of nominals in Polish and English, indicating the rank on
the scale of "nouniness", we shall obtain the following situation:

Polish:
1. Deverbal substantives (irregular)
2. Verbal substantives, (regular)
3. Verbal substantives 2 (regular)

English:
1. Regular gerunds
2. Action nominals (regular, irregular)
3. Gerundive nominals (regular)
Thus, at the level two of "nouniness" we have a regular morphological
situation in Polish and an irregular situation in English. At the level one
of "nouniness" we have an irregular morphological situation in Polish
and a regular situation in English. Unfortunately, one cannot rule out
the possibility that the two taxonomies are based on erroneous criteria.
For example, Lees' distinction between gerundive and action nominals
rests on a mixture of morphological, syntactic, and semantic properties.
210 Chapter Χ

As a result, action nominals comprise both regular -ing of formations

and irregular ones.
For the purposes of the present study, it was necessary to unify the
criteria by restricting them to morphological ones alone. This meant that
in both Polish and English only regular formatives were considered.
These included all Polish forms in -anie, -enie, -cie and all English
gerundive nominals as well as those action nominals which take the
regular suffix -ing. Lees' Regular Gerunds (full nouns) were excluded
from the count, following another morphological criterion, viz. their
ability to form plural by taking the plural inflection -s. This property
places Less' Regular Gerunds outside all verbal forms, on the very top
of the scale of "nouniness" On the other hand, no such elimination was
possible in Polish since only semantic criteria allow one to distinguish
the more "nouny" from the less "nouny" of the Polish verbal substantives.
In view of the results of the quantitative contrastive study of PE1,
PE2, and BE, one would have expected similar results from the quanti-
tative contrastive study of a Polish text written by a native user of Polish
and a text written by a native user of English. In this way the scanty use
of gerundive and action nominals in the Polish-English interlanguage
(PE1 and PE2) would be accounted for by the negative transfer of the
presumably lower frequencies of occurrence of nominals in Polish texts.
However, it turned out that the hypothesis about the negative transfer
from Polish to English, in the case of the frequencies of nominals, is not
easily verified.
Two texts were compared with regard to the frequency of occurrence
of English gerundive and action nominals ending in -ing and of Polish
substantiva verbalia. The first 300 sentences of John Lyons' Introduction
to theoretical linguistics were examined and compared with a very "Polish-
sounding" version of the same book translated from the original by
Krzysztof Bogacki under the title Wst§p do j§zykoznawstwa.
In the original version, 60 gerundive and action nominals were found,
which means that the average occurrence of such nominals was about
0.2 per sentence and corresponded to the results obtained from the
previous study. In the Polish text, as many as 208 regular nominals were
found, which gives the average occurrence as high as 0.7 per sentence,
which is significantly higher than in English. A closer quantitative look
at the data helps to understand this unexpected result. It turns out that
only about 80 of the total 208 words in -enie, -anie, -cie constituted the
less "nouny" words of rank 1 on the scale of nouniness. These averaged
to 0.26 per sentence, that is, almost as many as in the case of English
 Quantitative contrastive studies 211

nominals. Interestingly enough, among them there were many words

which corresponded to English full nouns (Lees' Regular Gerunds) rather
than to gerundive nominals or to action nominals. They were such words
as wyzwolenie sig 'emancipation', zepsucie 'corruption', zalozenie 'estab-
lishment', etc. As a matter of fact only as few as 24 of the Polish
substantiva verbalia corresponded to English nominals ending in -ing,
which constitutes only about 10 percent of all occurrences of substantiva
verbalia in Polish. This situation offers a hint suggesting factors which
influenced the Polish-English interlanguage in which PE1 and PE2 were
written. It appears that in the preponderant majority of cases, Polish
substantiva verbalia correspond to English full nouns (Regular Gerunds)
or to irregular action nominals as their lexical equivalents. Therefore, the
Polish learners of English associate Polish substantiva verbalia with
English full nouns rather than with -ing forms of gerundive and action
nominals as a matter of a general principle, following such equivalents
as dodanie 'addition', stosowanie 'use', odrzucanie 'refusal', porownanie
'comparison', spostrzezenie 'insight', uzycie 'currency', omawianie 'dis-
cussion', etc. Once such an association is formed, it is easily overgener-
alized to other instances. This overgeneralization is manifested in the
learner's attempts to find a lexical equivalent for every Polish substanti-
vum verbale rather than follow a possible transfer rule from a grammat-
ical rule deriving substantiva verbalia in Polish to a grammatical rule
deriving gerundive and action nominals in English. Presumably, no such
transfer rule has been formed in the case of the authors of PE1 and PE2.
The absence of such a transfer rule of grammar and the overgeneralization
of the lexical transfer account for the non-native-like performance in the
case of PE1 and PE, manifested as errors of avoidance affecting the
frequencies of occurrence of gerundive and action nominals.
The quantitative contrastive studies described above leads to the fol-
lowing general conclusion concerning quantitative contrastive studies:

1. Quantitative contrastive studies are strictly dependent on systematic

contrastive studies in that the latter provide criteria for identifying lin-
guistic forms to be counted in actual texts. For instance, the decision
about the system equivalence of the Polish forms in -enie, -anie, -cie and
the English forms in -ing is only possible to make when systematic analyses
of Polish and English noun-forming suffixes are taken into consideration.
The results of such qualitative analyses provide structural categories,
which are then counted in the course of quantitative contrastive studies.
212 Chapter Χ

2. The quantitative contrastive studies presented above suggests that the

hypothesis about the native vs. non-native performance, as manifested
by different frequencies of occurrence of certain forms is valid. In this
case, quantitative contrastive studies consists in the quantitative compar-
ison of texts produced by native users of a target language and non-
native users of that target language; the latter presumably produce texts
in an interlanguage. The deviations from the native norms are either in
minus (errors of avoidance), as in the case of English nominals used by
Polish students, or in plus (errors of abundance), as in the case of
extraposed subjects, so frequent in the Polish English interlanguage texts.
Such quantitative contrastive studies might be called intralinguistic in-
asmuch as they involve a target language and an interlanguage associated
with that target language. If texts produced by native users in two different
languages are compared, we deal with interlinguistic quantitative con-
trastive studies. It would seem that its results in this area cannot explain
certain transfer phenomena if these results are considered in isolation
form other results of both a qualitative and quantitative nature. Transfer
appears to be determined by a complexity of factors and cannot be
accounted for by one particular kind of contrastive studies.

3. There seems to exist a need for massive statistical research involving

both intralinguistic and interlinguistic quantitative contrastive studies,
which could eventually lead to a better understanding of the complex
phenomena of transfer, as well as those phenomena which characterize
native and non-native performance in a foreign language. The practical
validity of this kind of research is easy to appreciate. One result might
be a quantitative description of "individual style" at the plane of inter-
language. One can expect that the Polish interlanguage of English is
quantitatively different from the Hungarian interlanguage of English, etc.
Each national interlanguage of English is probably characterized by a
unique matrix of frequencies of occurrence of various linguistic forms.
The empirical verification of such hypotheses, among other concerns, will
keep contrastive studies alive for some time to come.
Chapter XI

Cognitive linguistics and contrastive studies

Cognitive linguistics is a field in modern linguistics which stands in

opposition to the transformational generative tradition. 1 Rather early in
its history, that is at the beginning of the 1960s, the originally uniform
theory created by Noam Chomsky began to branch off as a result of the
criticism expounded in the works of such linguists as Lakoff, Ross,
McCawley, Fillmore, and many others (cf. Newmeyer 1980: 133 ff.). While
Chomsky subjected his own standard theory to endless revisions and
extensions, the dissidents broadened the scope of their linguistic investi-
gations, and, having gone through successive stages of generative seman-
tics (Lakoff 1971), global rules (Lakoff 1970) and fuzzy grammar (Lakoff
1973), they eventually reduced all generative grammars to the status of
an interesting but marginal enterprise within linguistics.
Under the inspiration of various cognitive sciences, notably psychol-
ogy, anthropology, ethnography, and philosophy, cognitive linguistics has
come into being in the works of Lakoff (1982, 1986), Langacker (1986),
Lindner (1981, 1982), Brugman (1981), Fillmore (1984a, b) and many
others.
Cognitive linguistics is based on the assumption that human language
cannot be viewed as an abstract system isolated from socio-psychological
settings in which it is used, but, on the contrary, it reflects the way people
experience the world:

A wide variety of experimental factors perception, reasoning, the

nature of the body, the emotions, memory, social structure, sensori-
motor [sic!] and cognitive development, etc. determine in large
measure, if not totally, universal structural characteristics of language
(Lakoff 1977: 237).

Moreover, cognitive linguistics is based on radically different ontolog-

ical foundations in comparison with those on which structuralism and its
continuations are based. Cognitive linguistics rejects the classical, Aris-
totelian view of the world as consisting of discrete categories; it thus
214 Chapter XI

rejects the most essential axioms of the classical theory of knowledge on

which traditional linguistics was based. Specifically, cognitive linguistics
rejects the view that the world can be described in terms of categories
defined by means of necessary and sufficient conditions which must be
fulfilled by all the members of a given category. The classical theory does
not admit intermediate cases: everything either is or is not a member of
a given category, depending on whether it fulfills the required conditions.
Classical categorization admits complex categories as logical combina-
tions of simpler categories. Thus complex categories can be always re-
duced to simple ones. This sort of reductionism was applied in the
description of all aspects of language and manifested itself in reducing
complex signs to simple signs (morphemes) and even to a finite set of
submorphemic elements (phonemes). Reductionism, inherent in classical
categorization, was connected with set theory, in terms of which all levels
of language were analysed. For example, in phonology, matrices of
distinctive features are sets of conditions whereby category membership
is defined while various kinds of brackets represent various logical op-
erations on those sets. Likewise, any generative grammar defines sets of
sentences, and every sentence is an ordered set of phonological features,
etc.
The classical theory of categorization is firmly rooted in the tradition
of human thinking as it satisfies man's needs for orderliness. Questioning
it might be considered as an attempt to question the very existential
foundations of many scientific disciplines, mainly those concerned with
human knowledge, that is, cognitive sciences: philosophy, psychology,
anthropology, linguistics, and artificial intelligence. Yet, in all those
disciplines the classical theory of categorization has been challenged (cf.
Lakoff 1982: 14).
With regard to philosophy and linguistics, Wittgenstein (1953 [1972]:
50 ff.) pointed out that classical categorization cannot adequately cope
with various concepts, such as "game", inasmuch as there exists no set
of necessary and sufficient conditions characterizing this concept, which
must embrace various heterogeneous activities like chess, bridge, football,
solitaire, or hide-and-seek. Particular "games" mutually resemble one
another in varying degrees, like members of one family, but one cannot
identify a feature, leave alone a set of features, which would be common
to all games. It is only the "family resemblance" that can serve as the
basis on which one can classify all games as belonging to one natural
category.
 Cognitive linguistics and contrastive studies 215

Eleanor Rosch, a psychologist, investigated natural categories and

found that not all members of a given category are equally representative.
For example, in our culture robin is more representative or prototypical
in the category of birds than chicken, penguin or ostrich. Thus, according
to Rosch, the degree to which some creature fits the category bird is a
function of humanly-relevant properties — perceptual, functional, motor,
intentional, etc. which it shares with prototypical birds, such as
sparrows, robins, etc. Moreover, she discovered psychologically "basic"
categories, which from the set-theoretical standpoint are not fundamental.
The experiments have shown that such concepts as D O G and C H A I R
are basic while M A M M A L and F U R N I T U R E are superordinate, and
R E T R I E V E R and R O C K I N G - C H A I R are subordinate (Rosch 1973;
Rosch —Lloyd, quoted after Lakoff 1982: 17 — 19). Basic level categories
were found to "have maximal clusters of humanly-relevant properties"
(Lakoff 1977: 9), a fact which manifests itself in having visual prototypes
for D O G and C H A I R but not for M A M M A L and F U R N I T U R E .
Lakoff additionally insisted that "thought, perception, the emotions,
cognitive processing, motor activity, and language are all organized in
terms of the same kind of structures" (Lakoff 1977: 11), which he called
gestalts. He deliberately left the concept vague but suggested that gestalts
are "at once holistic and analyzeable (sic!)" (Lakoff 1977: 11) and though
they have parts, they are not reducible to parts since they have additional
properties, which cannot be seen in their parts examined in isolation from
the gestalt. 2 In the framework of cognitive linguistics, gestalts are idealized
cognitive models of reality as it is reflected in the minds of language
users. As idealizations, such models necessarily involve oversimplifica-
tions, metaphorical understandings, and both expert and folk theories of
reality. Idealized Cognitive Models define natural categories, which are
not objective, but are strictly connected with human experience of reality.
They contain the usual propositional content (the sort of information
that one is likely to find in a dictionary) as well as mental images of a
visual, or more generally, sensory and kinaesthetic type. Words in lan-
guage are not defined relative to relevant fragments of "objective" reality
but rather relative to such models (cf. Lakoff 1982: 48 ff.). The concept
of Idealized Cognitive Model corresponds to Langacker's concept of
"abstract domain" (Langacker 1983: 55 ff.). One way to represent them
is by means of "mental schemes", which we shall introduce later in this
chapter.
Further research has revealed that natural categories, characterized by
lack of clear-cut category boundaries (fuzziness, ability to stretch), inter-
216 Chapter XI

nal prototypical structure, family resemblance of category members, and

gradience (a member may be more or less representative of the
category), permeate natural languages. Consequently, language cannot
be described adequately in terms of strict rules which can only handle
prototypical phenomena but are less and less effective when non-proto-
typical phenomena are involved. For example, it turned out that the
fundamental phonological category, phoneme, is also a natural category
(Jaeger 1980, quoted after Lakoff 1982: 94).
Classical phonology was incapable of handling the well-known phe-
nomenon of phonemic overlap, which involves situations in which many
non-prototypical variants of some phonemes belong to a given category
because of family resemblance, and because of functional considerations.
These non-prototypical variants are often phonetically quite remote from
prototypical variants. For example, the English apico-alveolar strong
stop /1/ has several variants, of which only the most prototypical ones
sound like [t]. The prototypical English [t] appears in syllable initial
positions in front of stressed vowels. Other variants of ft/, in other
positions, are sometimes quite remote from the prototype. For example,
between a stressed and an unstressed vowel, as in the word water, jtj may
be realized in a very non-prototypical way as a voiced stop [d] or even a
'flap' [J], resembling the [J] in the word right, [J] is also a variant of the
English phoneme /r/. This example, in cognitive linguistics, is not treated
as an inconvenient case, which the phonological theory cannot cope with,
but as a natural situation characterizing all languages, in which categories
have fuzzy boundaries. Eventually, the [i] in water is classified as a variant
of jtj and the [α] in right as a variant of /r/, not because of their inherent
structural properties, but because they are perceived as such by native
users of English. They are thus assigned to appropriate categories not on
structural grounds, which in this case fail, but on cognitive grounds.
Ross (1973, 1975, 1981) has demonstrated that virtually all grammat-
ical categories exhibit prototypical effects. So we have more and less
prototypical nouns, verbs, adjectives, sentences, prepositions, etc. Ross
has also shown that all types of grammatical constructions, such as
passive, relative, coordinative, etc., also display prototypical effects.
In comparison with the generative tradition, cognitive linguistics em-
braces a much larger scope of linguistic data, including the traditional
domain of grammar, that is, morphology and syntax. In terms of cognitive
linguistics, one may say that transformational grammar dealt only with
the most prototypical phenomena, those that are best motivated in our
experience, while it completely ignored all kinds of non-representative
 Cognitive linguistics and con t ras live studies 217

examples, so vital in everyday use of language. This restriction in the

scope is particularly evident in the case of figurative meanings, about
which generative grammars have virtually nothing illuminating to say.
Transformational grammar generates only those sentences which are
prototypical for a given category, which means that it ignores all those
numerous interpretations in everyday uses of language, which are meta-
phorical extensions of prototypical senses. Metaphor has been placed in
the very focus of attention of cognitive linguistics since human experience
and human thinking are largely metaphorical (Lakoff—Johnson 1980:
229 ff.), which cannot fail to influence language. In Lakoff s estimation,
"literal" senses of sentences amount to only 2 to 5 percent of all senses
used in everyday communicative acts. This is the area embraced by the
so-called core grammar consisting of about 20 rules (Lakoff 1983: 38).
This situation results from the fact that, in Lakoff s words: "There appears
to be a continuum between productive constructions and completely
frozen constructions" (Lakoff 1983: 38).
It is perhaps needless to add that the limitations of core grammars,
pointed out above, can be attributed a fortiori to Contrastive Generative
Grammar as it adds its own difficulties connected with semantic equiv-
alence to those which are already inherent in the nature of generative
grammars themselves.
The emergence of cognitive linguistics has created the necessity of
revising the work in contrastive studies. The first attempt in that direction
was made by Schlyter (1980), who produced an excellent contrastive
study of various nouns and verbs in German, French, Spanish, English,
etc. in terms of the prototype theory, coming up with a number of
insightful and valuable observations, unavailable outside the cognitive
framework.
In a brief section of his thesis, Kalisz (1981) attempted to restate the
concepts of congruence and equivalence in terms of "partial pattern
matching" as discussed in Lakoff (1977). This latter concept is based on
the assumption that linguistic forms can be characterized by clusters of
pragmatic, semantic, and syntactic properties (parameters). Various con-
structions, it appears, exhibit various degrees of correspondence of such
parameters. This kind of correspondence may be referred to as matching,
which extends over a continuum from full matching, through partial
matching, to no matching at all. Kalisz tries to employ the concept of
partial matching in explicating the familiar relations of equivalence and
congruence, crucial in all contrastive studies based on structural and
transformational models. He maintains that
218 Chapter XI

Equivalence between two given structures is a matter of degree of

partial pattern matching of properties. One can talk about a degree
of syntactic equivalence even if lexical properties do not match, a
pragmatic equivalence when the two structures produce the same
perlocutionary effect in spite of their syntactic and lexical properties
etc. The higher degree of matching of syntactic, semantic and prag-
matic properties reflects the higher degree of overall equivalence be-
tween two or more constructions (Kalisz 1981: 45 — 46).
As originally conceived in structural and transformational terms, con-
gruence and equivalence referred to formal and semantic identity of the
compared constructions (Krzeszowski 1967: 37; 1971: 38; Marton 1968b:
56 — 57). The separation of semantic and syntactic representations, as
formalized in the standard theory, led to the "identical-deep-structure
hypothesis" (cf. Chapter VII), which, as we remember, claimed that
equivalent constructions and sentences have identical semantic represen-
tations, even if on the surface they are markedly different (Krzeszowski
1971: 38, 1979: 11). Formal diversifications occurred at various levels of
derivations in the subsequent derivational histories of equivalent construc-
tions. More similar constructions shared more rules and were diversified
later, that is, at some level closer to the level of surface representations.
The distance from the surface at which the first diversification occurred
provided grounds for measuring the degree of similarity and difference
of the compared constructions (Di Pietro 1971: 27; Krzeszowski 1974,
1979: 13). The formal device which was to accomplish this task is called
Contrastive Generative Grammar.
Contrastive Generative Grammar rendered the concept of congruence
somewhat superfluous since all equivalent constructions were congruent
by definition at some level of representation (if only at the semantic level,
at which congruence was guaranteed as a matter of the initial postulate).
Therefore, the idea of degrees of syntactic similarity is not new with
Kalisz's suggestion.
Kalisz's proposal requires at least two amendments. Firstly, the re-
quirement that pragmatically equivalent structures must produce the same
perlocutionary effect should be relaxed through substituting "maximally
similar cognitive effects" for "the same perlocutionary effects" Not all
speech acts have perlocutionary effects, so that under Kalisz's proposal
a number of speech acts such as statements, rhetorical questions, etc.
would be unaccounted for. Kalisz's proposal is too restrictive also because
it contains an untenable and, I am sure, unintended, implication that
 Cognitive linguistics and contrastive studies 219

only those utterances which happen to be produced simultaneously in

the same setting and with the same perlocutionary effect are equivalent.
Even if restricted to perlocutionary effects, the requirement concerning
their "sameness" rules out any prospects of generalization and renders
the proposal uninteresting. It is obvious that a given perlocutionary effect
can occur only once. Instead, it is reasonable to consider identity of
effects as a special case of similarity, where identity is to be viewed as
the highest theoretically conceivable degree of similarity.
Secondly, the concept of partial pattern matching as envisaged by
Kalisz in the context of contrastive studies seems to require some con-
straining. According to Kalisz, "Equivalence between two structures is a
matter of degree of matching properties. Thus, it reflects a degree of
partial pattern matching of properties" (Kalisz 1981: 45). This claim must
be interpreted as suggesting that given a set S of pairs of equivalent
constructions, C's and K's, one can arrange these pairs on the scale
provided by the varying numbers of certain matching properties. Let us
assume that S contains the following pairs of equivalent constructions,
characterized by some properties which match and some which do not
match. Let P's stand for those properties which match, and let p's stand
for those properties which do not match:

S = C,(PPPP) C 2 (pPPP) C 3 (pppP) C 4 (pppp)

S = K,(PPPP) K 2 (pPPP) K 3 (pppP) K 4 (pppp)

The pairs of constructions in S display various degrees of similarity,

with Q and K, being identical in all relevant respects and with C 4 and
K 4 being different in all relevant respects. Considering this latter case,
the question should be asked: on what grounds are they compared? Note
that we have artificially limited the number of properties (parameters) to
four, providing the upper bound on the scale of similarity with the value
of 4 and the lower bound having the value of 1. N o such limitations can
be arbitrarily imposed on actual constructions. However, if we do impose
some more or less arbitrary limits on the number of properties considered,
we still face the following problem: if two linguistic forms match with
regard to one property, can they be said to exhibit partial pattern
matching? From the set-theoretical point of view the answer seems to be
"yes" After all, one property out of four is a part of the total four, and
if it matches, which can be expressed in terms of set intersection, it is
involved in partial pattern matching.
220 Chapter XI

However, consider the following examples from English and Polish:

(1) All visitors are kindly requested to leave the boat immediately.
(2) Proszg siadac. 'Please sit down'

(1) and (2) evidently share at least one property request. But can
they be said to be equivalent, pragmatically or otherwise, merely because
they exhibit this extremely low degree of pattern matching? Consider,
moreover:
(3) Spieprzajcie stqd. 'Get the hell out of here'
and
(4) Statek zaraz odplywa. 'The boat departs in a moment'

In certain cirumstances (1) and (3) may have identical perlocutionary

effects (visitors leaving the boat); yet, it is doubtful whether anybody
would consider them as pragmatically equivalent. On the other hand, (4)
may in some situations be a better equivalent of (1), even if it exhibits
even less pattern matching with (1) than do either (2) or (3).

The discussion so far leads to the following conclusions:

(a) The mere number of shared properties is not decisive in establishing

pragmatic equivalence, because it does not seem to be correlated with
the degree of similarity as grasped by partial pattern matching.

(b) The identity of perlocutionary effects does not guarantee that the
corresponding utterances are pragmatically equivalent.

(c) Pragmatic equivalence appears to be dependent on elements of ex-

tralinguistic contexts attending the production of equivalent utterances
(see also Janicki 1985: 15ff.).

Thus, pragmatic equivalence depends on the status attributed to var-

ious properties rather than on their number, and the status may vary
according to the effects that the speaker seeks to achieve. Given a highly
onomatopoeic poem such as "The Bells" by Edgar Allan Poe, the trans-
lator will seek to retain as many acoustic effects, imitating various kinds
of bells, as possible. He will, therefore, attribute privileged status to the
phonetic properties of the poem. A pragmatically equivalent translation
of such a text will focus on the phonetic substance of the poem, perhaps
 Cognitive linguistics and contrastive studies 221

to the detriment of semantic and most probably syntactic properties. 3

On the other hand, in the translation of legal documents, pragmatic
equivalence will be determined by maximum semanto-syntactic accuracy.
Pragmatically equivalent legal texts will probably approach semanto-
syntactic equivalents as defined in Chapters III and VII.
Briefly speaking, pragmatic equivalence of texts can only be seen as a
realization of a number of conditions which are necessary to obtain
particular cognitive effects evoked by those texts. The conditions vary
depending on the kind of text, its internal structure and its cognitive
function. Each of these factors may determine the privileged status of a
certain property or a cluster of properties of the text, which then serve
as tertium comparationis for pragmatic equivalence in a specific instance.
In view of the above considerations, I would like to claim that if two
linguistic forms across languages belong to one category within the
domain of contrastive studies 4 , the properties crucial in determining
category membership are of unequal status as category determinants. In
other words, some properties are more privileged than others in assigning
two linguistic forms in two languages to one category within the domain
of contrastive studies. Moreover, the properties which are critically im-
portant in determining pragmatic equivalence constitute a gestalt in the
sense described by Lakoff (1977). In the case of (1), tlu· most important
parameters constituting the relevant gestalt are:

polite request to leave the boat.

Other parameters, such as syntactic congruity or lexical congruity, in

this case, play a less important role. The three relevant parameters are:
type of speech act (request), modality (politeness), and anticipated per-
locutionary effects (visitors leaving the boat). Focusing on any one of
these three elements of the gestalt in the rendering of (1) in Polish results
in the distortion of the original pragmatic function. Thus in (2) the
anticipated perlocutionary effect is different, while in (3) the modality is
different (rudeness rather than politeness), which in turn may influence
the perlocutionary effect. This explains why (4) ranks higher on the scale
of pragmatic similarity than does either (2) or (3): (4) is neutral with
respect to politeness, and at the same time it does not rule out the
possibility of being interpreted as a request with the perlocutionary effect
such as anticipated in (1). Naturally, (5) and perhaps (6) would rank even
higher on the same scale:
222 Chapter XI

(5) Uprasza si§ wszystkich gosci ο natychmiastowe opuszczenie statku.

(Impersonal construction)
(6) Wszyscy goscie sq proszeni ο natychmiastowe opuszczenie statku.
(Passive construction)
(5) and (6) are the most accurate equivalents of (1) in all respects:
syntactic, semantic, and pragmatic. 5 At this stage I would like to suggest
that (5) and (6) are the most prototypical equivalents of (1), while (3) is
less prototypical. At the same time (4) is a poor representative of the
category of equivalents of (1), and (2) is probably outside the category.
We thus note that linguistic forms may have more (or less) prototypical
equivalents in other languages.
Let us consider one more example in order to explore some further
consequences of the prototype theory of equivalence. In her study of
English prepositions and particles, Brugman (1981) distinguishes nearly
100 senses of the preposition-particle-prefix (ppp) over and among them,
the following 20 non-metaphorical senses:
1. The plane flew over the town.
2. The plane flew over the hill.
3. The helicopter is hovering over the town.
4. The helicopter is hovering over the hill.
5. He is walking over the hill.
6. Sam lives over the hill.
7. The glider flew over the wall.
8. Harry jumped over the wall.
9. Harry jumped over the cliff.
10. The wall fell over.
11. Sam turned the page over.
12. Sam rolled the log over.
13. The power line stretches over my yard.
14. She spead the tablecloth over the table.
15. The guards were posted all over the hill.
16. He walked all over the hill.
17. She held the veil over her face.
18. He drove over the bridge.
19. He's over.
20. The water overflowed.
Each of these senses is associated with a specific image schema indi-
cating trajectors (the objects situated somewhere or moving), landmarks
 Cognitive linguistics and contrastive studies 223

(the places at which trajectors are situated or relative to which trajectors

move), and paths, whenever they are involved. (For details see Langacker
1987) These examples do not include metaphorical extensions of the ppp
over in such sentences as He has the authority over a staff of hundreds,
as an extension of 3 or He turned the question over in his mind as an
extension of 12. All these senses are related by virtue of family resem-
blance and can be chained in such a way that the most prototypical ones,
that is, 1 and 2, are situated in the centre while the progressively less
prototypical ones grow out in various directions. Towards the edges of
the chaining, one finds those senses which bear the least resemblance to
the prototypical ones, but which are locally more similar. The relevant
part of the chain is presented in (7):

(7)

© Θ Θ ©

In addition, senses 2, 3, 5, 6, 8, 9, 12, 14, 16, 18, and 20 have metaphorical

extensions, which are not indicated on the chain.
In attempting to compare this material with the relevant material in
another language, say Polish, the investigator faces the usual question:
what is the equivalent of over in Polish? Naturally, given at least twenty
senses of over in English one cannot expect a single word equivalent in
Polish.
In search of the equivalents Brugman's examples were submitted to a
group of Polish students of English with the request to translate them
into Polish. There were 25 respondents, all well advanced in English. The
purpose of the exercise was to find out how the ppp over would be
rendered in Polish. The results are presented in Table 3.
224 Chapter XI

Table 3. The preposition-participle-prefix over translated into Polish by 25 respondents

English senses Polish equivalents Number of answers

1 nadjponad 25
2a nadjponad 24
3 nad/ponad 25
4 nadjponad 25
5 po 10
przez 9
na 2
instrumental case 2
nadjponad, 1
woköl 1
6 za 18
na 3
obok 1
przy 1
po drugiej stronie 1
niedaleko 1
7 nad/ponad 24
przez 1
8b przez 16
accusative case 11
9" przez 10
accusative case 6
ζ 6
po 1
poprzez 1
10b other means 24
(phrases, prefixes)
accusative case 23
na drugp strone 2
12 accusative case 22
na drugg strone 3
13c nadjponad 19
przez 4
14 na 25
15 po 13
na 10
wokol 1
na obszarze 1
16 po 15
przez 4
accusative case 4
wzdluz i wszerz 1
wokol 1
 Cognitive linguistics and contrastive studies 225

English senses Polish equivalents Number of answers

17d na 15
accusative case 8
nad 1
18" przez 21
accusative case 2
po 1
19a other means 24
20 wy- (prefix) 18
prze- (prefix) 7

a
One respondent provided no answer.
b
In two cases two answers were provided: accusative case and przez.
c
Two respondents provided no answer.
d
In one case a completely irrelevant answer was given.
10 and 19 were nearly always rendered as pseudotransitive verbs przewrocic sig i skonczyc
sif, respectively.

These results are interesting for several reasons.

Firstly, if we coalesce nad and ponad as free variants, at least in the
contexts in question, we see that the most prototypical senses of over (1,
2, 3, and 4) are invariably rendered as nad/ponad, which must be recog-
nized as the most prototypical equivalent of over. Over 7, rendered as
przez, comes second by a very narrow margin of one. (Po)nad as the
most prototypical equivalent of over deserves the first mention in an
English-Polish dictionary although Stanislawski has 'na', which is the
only equivalent of the less prototypical over 14.
Secondly, in less prototypical senses, the Polish equivalents of over
vary over a considerable range of prepositions, prefixes, and certain other
means such as inflections and special forms of verbs.
Thirdly, the Polish equivalents of non-prototypical senses of over evoke
varying degrees of agreement among the respondents. On the one hand,
in the case of 8, 9, or 15, there occur considerable divergencies, on the
other hand, we deal with complete unanimity in the case of 14. Although
the reasons for these divergencies remain to be investigated, one can
surmise that they are partly due to the gaps in the linguistic competence
of the respondents, but also, at least to some extent, to their different
cognitive processing of the situations are depicted by the examples (es-
pecially in the case of 15 and 17).
Fourthly, less prototypical senses of over may have very prototypical
equivalents in Polish alongside less prototypical ones. For instance, po
226 Chapter XI

calym wzgorzu appears to be the most prototypical equivalent of all over

the hill in 16 while przez wzgorze and wzdluz i wszerz wzgorza are less
prototypical equivalents of over 16.
Fifthly, nad, the Polish equivalent of the five most prototypical senses
of over, itself has a number of senses, many of which have metaphorical
extensions and many other senses which do not even correspond to over.
I have been able to distinguish at least 14 less basic senses of nad as a
preposition and as a prefix. The eight basic senses of nad are exemplified
as follows:
1. Balon unosi si§ nad miastem.
'The balloon is hovering over the town'
2. Samolot przelecial nad miastem.
'The plane flew over the town'
3. Wierzba pochyla sie nad wodg.
'The willow leans over the water'
4. Balon unibsl sie nad wysp§.
'The balloon rose over the island'
5. Balon przylecial nad wyspg.
'The balloon flew to (the area) over the island'
6. Usiedli nad wodg.
'They sat near the water (front)'
7. On mieszka nad morzem.
'He lives near the sea'
8. Wyjechali nad morze.
'They went to the seaside'
Senses 1 and 2 are the most prototypical and constitute the core of
the following chaining:
(8) ®—@—@—®—®—CS>—@—®
In addition there are the following senses:
9. On nadbiegl. 'He came running'
10. Nadbudowali strych. 'They built the attic above'
11. Nadwozie 'car body'
12. Nadkomisarz 'supercommissar'
13. Nad snieg bielszy 'whiter than snow'
14. Nadcz/ow/e/c 'superman'
15. Nadcisnienie 'hypertension'
16. Nadspodziwany 'unexpected'
 Cognitive linguistics and contrastive studies 227

17. Nad/ac 'pour out a little'

18. Nad ranem 'before dawn'
19. Sqdownictwo nad chlopami 'peasant jurisdiction'
20. Prace nad siownikiem 'work on a dictionary'
21. Zachwyt nad urodg. 'delight at someone's beauty'
22. Nadbutwiec 'start being affected by rot'
Allowing for possible modifications and further extensions, the basic
chaining can be augmented thus:

In addition to the extensions mentioned here, many of which become

metaphorical towards the end of the chain (e.g., 13 onwards), some of
the spatial senses have direct metaphorical extensions, for example, 3 —
wyrastal nad przecigtnosc 'he grew above the mediocrity', 4 — pochylal
si§ nad kazdym bliznim 'he leaned over his every neighbour' The chaining
of the senses of nad is quite complex, but all these senses exhibit family
resemblance. Through relating all those divergent senses by means of
family resemblance, we obtain a fairly coherent view of otherwise dis-
parate phenomena.
It will be noted that each of these senses of nad will have its more and
less prototypical equivalents in English. 1, 2, and 3 will prototypically
228 Chapter XI

correspond to the English 'over' while 4 corresponds to 'above', etc. On

the basis of these data one can state that over and nad are prototypical
equivalents in the two languages, and that the relevant spatial senses,
that is, 1, 2, 3, and 4 in English and 1, 2, and 3 in Polish exhibit complete
pattern matching with respect to image schemas associated with those
senses.
To provide a complete contrastive study of the area of senses covered
by over and by nad, a similar procedure would have to be adopted for
each sense individually. Consequently, each sense would have to be
explored with the prospect of finding its prototypical and less prototypical
equivalents in the other language, until the entire area could be thus
explored. The resulting chainings would be mutually interconnected by
a complex network of links representing the equivalence of specific senses.
Such findings could be additionally supported by the study of image
schemas associated with particular linguistic forms. For example, the
research could either confirm or refute the prediction which seems to
ensue from the foregoing discussion that the image schemas evoked by
over in its four most prototypical senses and by nad in its three most
prototypical senses are centred around the following image schema:

where the trajector (TR) is either stationary or moving, but at a certain

time either its fragment or its complete body finds itself in the position
indicated by the black dot over the landmark (LM) in (10). Dashed lines
represent these potential paths (P's). The image schema is a gestalt, against
which the most prototypical senses of over and nad are centred. Apart
from this gestalt, there are other gestalts which serve as centres for other
senses of over and nad. These other centres attract other equivalents, etc.
The cross-language landscape of senses can, therefore, be seen as a
multifocal space with various linguistic forms in both languages, inter-
connected by identical gestalts. Every instance of the use of over involves
focusing on the relevant sense associated with the relevant gestalt.
In the case described here we have been dealing with linguistic phe-
nomena which exhibit a great deal of resemblance. The very fact that the
 Cognitive linguistics and contrastive studies 229

most prototypical equivalents of the most prototypical senses of the

English preposition over correspond to a preposition in Polish suggests
a high degree of pattern matching. Yet, when less prototypical (and more
conventionalized) senses are considered, or when typologically more
distant languages are involved, one can expect a smaller degree of simi-
larity (less pattern matching). Consider over 9, which is connected with
the image schema (11):

(11)

LM

A small majority of ten respondents rendered over 9 as 'przez', mis-

construing the image schema and visualizing it as

(12) TR

Another group (6 respondents) apparently had a different image,

namely something like (13), when they rendered over 9 as the accusative
case 'przeskoczyl ska!?':

TR
(13)

LM
230 Chapter XI

Finally, another group (6 respondents), by rendering over 9 as 'zes-

koczyl ζ cliffu' imagined something in the nature of (14) and thus obtained
the highest degree of pattern matching:

(14)

LM

Disregarding other, less numerous possibilities, we face at least three

distinct gestalt interpretations of the situation involved in over 9 and
three grammatical, though hardly accurate, translations of over 9 into
Polish. These divergent interpretations are due to the fact that Polish has
no way of expressing the situation depicted in (11) by grammatical means
similar to those which are needed to translate the preceding eight senses
of over. Therefore, any attempt to use an equivalent preposition or at
least an inflectional ending or a prefix yields results which only partially
match the original with respect to the relevant image schema. The rela-
tively conventionalized English expression to jump over the c/i#"associated
with (11) can be more fully matched with the Polish periphrastic (hence
highly motivated!) expression skoczyc w dol przez kraw§dz urwiska. Full
matching of the image schemas in this case is bought at a price: the
degree of syntactic pattern matching is now lower, and the degree of
conventionality of the equivalents is unequal since the English conven-
tional expression is rendered in Polish by means of a less conventionalized
and more productive collocation of words.
To conclude, in contrastive studies conducted in terms of the prototype
theory, gestalts and partial pattern matching, equivalent forms exhibit
various degrees of pattern matching. Thus, similarity (and difference) can
be evaluated by means of a gradient scale. The upper bound of the scale
is delimited inherently by complete pattern matching of semantic and/or
syntactic properties. This situation is most likely to occur when proto-
typical equivalents are involved, although, by no means, need this always
be the case. The lower bound of the scale is not delimited by the matching
patterns themselves since there is no a priori way of deciding on the
necessary minimum of similarity required for the recognition of two
linguistic forms as matching. Therefore, the lower bound of the scale is
delimited cognitively through the bilingual informant's recognition of two
linguistic forms in two languages as being in one category within the
 Cognitive linguistics and contrastive studies 231

domain of contrastive studies. Thus, the decision is made not on the

basis of the inherent properties of the compared categories but on the
basis of what Lakoff calls "background framing" (Lakoff 1982: 25).
Notwithstanding the fact that category boundaries may be fuzzy, their
area is delimited by the extent to which their various non-prototypical
senses stretch. Such delimitations are also based on cognitive grounds.
Chapter XII

"Theoretical" and "applied" contrastive studies

In 1980, Carl James published his Contrastive analysis, the second book,
after Di Pietro (1971), devoted to the theory and practice of contrasting
languages. In contrast to Di Pietro's, James' book is much more peda-
gogically oriented. James touches on many problems which have a direct
bearing on teaching foreign languages, though he addresses many con-
troversial theoretical problems beclouding contrastive studies without
claiming to provide definite solutions. In addition to a special chapter
devoted to "Pedagogical Exploitations of Contrastive Analysis", the book
abounds in digressions elsewhere whenever the author thinks that he has
something relevant to say about language pedagogy. In addition to James'
book, excellent accounts of the relevance of contrastive studies to teaching
foreign languages can be found in numerous published works, such as
N i c k e l - W a g n e r (1968), Catford (1968), Lee (1968a, b), Marton (1976),
Fisiak (1981), Sanders (1981), and very many others. Since James' book
is the most extensive treatment, so far, of the numerous controversial
problems connected with the distinction between "theoretical" and "ap-
plied" contrastive studies, I hope its author will not object to my treating
it as a take-off ground for my own subsequent thoughts. To a large extent
this chapter will be parasitic on James' book, inasmuch as it will mostly
contain ideas which the book has inspired.
The most conspicuous feature of James' monograph is its eclecticism
or, to use Sharwood-Smith's less emotive term, unorthodoxy (Sharwood-
Smith 1976: 50). James does not commit himself to any specific linguistic
theory nor indeed to any mode of contrastive analysis but provides the
reader with a large array of models, opinions, and proposals, while only
occasionally extending moderate criticism. James' criticism is not, how-
ever, anchored in any system of criteria of empirical or theoretical ade-
quacy. This rather liberal attitude to various proposals is to be applauded,
but it has its dark sides too. Most importantly it shows in James' attitude
to the very nature of contrastive studies, which in turn has a tremendous
impact on the organization of the book. James seems to be unable to
234 Chapter XII

decide whether contrastive studies is a part of applied linguistics or a

branch of linguistics and what the mutual relations between contrastive
studies, psychology, and pedagogy are.
The problems which James faces are the same ones that lie at the heart
of many linguistic debates. As such these problems cannot be avoided in
any work attempting to provide a synthetic view of contrastive studies
since the field is so closely connected with, and so obviously dependent
on, theoretical linguistics.
The first chapter promises a neat division into "pure" and "applied"
linguistics and disposes of the views denying the existence of the latter
as a theoretically-based discipline (Corder 1973: 10). James' main argu-
ment against considering an applied linguist as a consumer rather than
producer of theories is that in order to consume something one must be
selective and the selections must be guided by a theory of relevance and
applicability. James' next argument is that unlike "pure" linguistics "ap-
plied" linguistics (including contrastive studies as its central component
[p. 8]) is a hybrid discipline, embracing in some relevant aspects psy-
chology and sociology, and, therefore, it is a discipline distinct from
"pure" linguistics. It remains unclear how James' intends to show that
by being a hybrid, "applied" linguistics is ipso facto a theoretical disci-
pline. Being a hybrid only shows its different status vis-ä-vis "pure"
linguistics. It seems that James' argument in favour of the theoretical
status of "applied" contrastive studies rests on the use of the notion "a
theory of relevance and applicability" One would, therefore, expect that
this notion would be thoroughly discussed as a potential provider of the
necessary criteria of adequacy of contrastive studies. Unfortunately,
James' book falls short of these expectations. In the remainder of the
book, James does not maintain his initial stand on the status of contrastive
studies: occasionally recalling Fisiak's distinction between "theoretical"
and "applied" contrastive studies, he announces at one place: "From now
on I shall intend 'applied CA' whenever I use the term CA" (p. 8). Yet,
the relevance of this distinction remains obscure as James continually
refers to both theoretical and applied aspects of contrastive analysis. The
relevance of both of these in language pedagogy is well demonstrated,
even in the chapters dealing with supposedly "purely" linguistic aspects
of contrastive analysis, i. e., chapters 3, 4, and 5, which deal with linguistic
components, of contrastive analysis, and its microlinguistic and macro-
linguistic aspects respectively.
It appears that James' problems begin the very moment he declares
that contrastive analysis is merely a study of interlanguage (p. 3), thus
 'Theoretical" and "applied" contrastive studies 235

bluntly equating the two concepts, which, even though to some extent
related, are in fact products of completely different approaches to lan-
guage and are in no way identical. If contrastive analysis has the form
of a generative grammar, then indeed it may serve as a model of inter-
language along the strictly confined lines suggested above in Chapter IX.
But James insists that contrastive analysis should be a study of an
interlanguage as a dynamic, changing structure. In his view, contrastive
analysis cannot be a synchronic endeavour but must have the capacity
to account for the changing nature of an interlanguage, that is, it must
embrace the diachronic dimension. Let us marginally observe that the
word "diachronic" is not particularly fortunate to use in this context. If
contrastive analysis is to account for the changing nature of an interlan-
guage within an individual, then the word "dynamic" is probably more
appropriate. "Diachronic" would be fitting if someone attempted to
describe the changes which affect various interlanguages, and which take
place in time for a whole given population of foreign-language learners.
But, by his own admission, this is not what James means.
Nonetheless, even if one accepts this narrower concept of "diachronic"
meaning "changing", "evolving", one cannot help wondering about the
nature of interlanguage and its presumably changing status. At this point,
it is necessary to distinguish between approximative systems in the sense
of Nemser (1971) and interlanguage in the sense of Selinker (1972). Such
a distinction seems all the more necessary since the differences between
the two concepts are implicit in the cited papers, though to my knowledge
they have never been explicitly discussed. The two terms are often used
as synonyms alongside Corder's transitional competence (Corder 1967,
1974: 25).
An approximative system in Nemser's sense indeed refers to something
elusive, continually changing in the direction of the target language and
extremely difficult to describe except as a dynamic structure having a
distinctly temporal dimension. In contrast to this, interlanguage in Se-
linker's sense appears to be much more stable, a kind of fossilized
approximative system, not subject to changes, except those that charac-
terize any language of a mature individual. Therefore, interlanguage is
no less amenable to a synchronic description than any other language.
More generally, any frozen competence, whether in one's native or in a
foreign language (or interlanguage), is equally easy or equally difficult to
describe diachronically, synchronically, or otherwise.
Now, if this distinction is valid, it is obvious that the research into the
emergence and evolution of approximative systems is a different matter
236 Chapter XII

from any study of a particular interlanguage. From the psycholinguistic

and applied linguistic point of view, the study of approximative systems
is certainly more interesting, more rewarding, and more tempting. But a
linguist is also entitled to describe any interlanguage within the framework
of an available linguistic theory and the two fields must be distinguished.
There is no doubt that one can conduct contrastive psycholinguistic
research and within such studies investigate the rise and development of
approximative systems. It is equally possible to investigate interlanguages
and compare them with both the source and the target language. When
taken to logical conclusions, Corder's stand on transitional competence
leads to a complete annihilation of the traditional field of error analysis
but certainly not of contrastive analysis! Error is no longer a valid
linguistic concept as it can only be conceived in microsociolinguistic
terms. The so-called error analysis turns out to be a contrastive analysis
of interlanguage and the target language.
The question arises whether such "errors" are predictable. The predic-
tion of errors has always been a major concern of contrastive analysis so
naturally James, too, expresses his concern in this respect. But in view of
what we said above, the simple question whether contrastive analysis can
predict errors has acquired an unexpected depth since now the question
concerns no more nor less than the form of interlanguage and can be
rephrased thus: "Is the form of interlanguage predictable?" "Are there
any constraints on the form of interlanguage, and if so what are they?"
Such questions touch on the very essence of modern linguistic theory.
They are questions about the boundaries of language universals. If con-
trastive analysis is able to cope with such questions, it no longer deserves
the humiliating title of the Cinderella of linguistics but should be allowed
in the royal chambers of human knowledge.
James does not draw the distinction between approximative systems
and interlanguage along the lines presented above. Instead, he identifies
interlanguage with the dynamic, changing structure, which we would
prefer to call an approximative system. In James' view, contrastive anal-
ysis ought to be a study of such dynamic objects. In this sense, Fisiak's
distinction between applied and theoretical contrastive studies becomes
vacuous since contrastive analysis by definition must be concerned with
processes of foreign-language learning and acquisition, that is, with the
domains traditionally associated with applied linguistics. This is what
James himself readily admits:
In a sense, the contrastivist continually transcends his own competence,
in that he is first and foremost a linguist, whose proper concern is
 'Theoretical" and "applied" contrastive studies 237

with structure, and yet he presumes to draw conclusions about a mode

of human behaviour, learning. He seems to act thus out of a conviction
that his CAs possess some sort of psychological reality (James 1980:
178).

Such a view of contrastive analysis seems to leave no room for

"theoretical contrastive analysis" unless one is ready to accept the rather
uncomfortable conclusion that such analyses are concerned with the study
of interlanguages while "applied" analyses deal with the study of ap-
proximative systems. There is no obvious reason why "theoretical con-
trastive analyses" should be thus restricted.
James' eclecticism is very well seen in his attempt to match the outdated
behaviourist psychological model of language based on the Stimulus-
Response relation with modern linguistic theory, which he does in the
chapter devoted to the psychological basis of contrastive analysis. The
impossibility of reconciling the two approaches is manifested in the
contradiction with which James concludes the chapter. On the one hand
he writes:

'Transfer' is the psychological cornerstone of CA. I have shown how

it is manifested in L2 learning. The concept of transfer originates in
behaviourist psychology, which has been superseded by cognitive psy-
chology (James 1980: 25).

But immediately, on the same page, he adds:

The contrastive analyst is not, and need not inspire (sic!) to become,
a psycholinguist. It is the contrastive analyst's duty to chart the
linguistic (structural) routes in L2 learning. His findings and those of
the psycholinguist will be complementary, but their instruments and
methods must be different.

Thus, one remains in the dark: are contrastive analysis and psychology
autonomous, or does psychology affect the conceptual framework of
contrastive analysis? Furthermore, are contrastive analysis and psycho-
linguistics complementary, or is the former based on psychology, as the
title of the chapter suggests. These are not merely questions of terminol-
ogy. The neglect of this problem leads to numerous theoretical and
practical difficulties, especially in connection with the often discussed
"psychological reality" of linguistic models.
One of the most controversial problems connected with the pedagogical
applications of contrastive analysis is how cross-language differences are
238 Chapter XII

related to learning difficulties. James devotes a fair amount of space to

this vital matter but does not seem to be able to arrive at any definite
conclusions. Due to some editorial mishap, the subchapter entitled "Scales
of Difficulty" appears twice, first as a subchapter of Chapter 6 (James
1980: 147 — 148), then lengthy fragments are repeated virtually verbatim
as a subchapter in Chapter 7 (pp. 187 ff.). When James first quotes
Stock well, Bowen and Martin's scale of difficulty, he seems to be rather
sceptical about its validity. Having evoked Tran-Thi-Chau's criticism of
the scale (Tran-Thi-Chau 1975), he concludes:
The scale is, of course, subject to empirical validation, though when
one attempts this a whole array of other complicating factors
motivation, aptitude, teaching or learning style, etc. enters the
picture (James 1980: 148).
These "complicating factors" indeed reduce the scale to a mere ab-
straction, weakly reflecting didactic reality. But in the course of the forty
pages separating the two references to the scale of difficulty, James seems
to have changed his mind. Having quoted even more criticism, notably
by Nickel (1971) and Nickel — Wagner (1968), he unexpectedly and with-
out any justification concludes:
There is undoubtedly substance in all these criticisms of the scale of
difficulty. But all of them can be answered to some extent without
totally abandoning the conceptual framework it is based on (p. 192).
Unfortunately, this is where the book ends, and James does not show
how these criticism can be answered, nor whether indeed the framework
can be defended.
In addition to the critical comments that James himself quotes, one
may add that the proposed scales are based on criteria which completely
ignore semantic and communicative values of the relevant constructions.
It seems that since the role that various grammatical forms play in
communication is a distinctly motivating factor, it is impossible to design
a sensible hierarchy of difficulty without considering it. There is no doubt
that those forms which have low communicative value are less interesting
to the learners and subjectively more difficult. This is but one aspect of
why the original scales do not promise to provide a particularly useful
framework on which the preparation of materials for teaching could be
based. Another aspect is the following.
There is no doubt that in a large number of instances places which
are identical across languages are easier to learn than those that are not.
 'Theoretical" and "applied" contrastive studies 239

As a matter of fact, the learner does not have to learn them at all but
only to realize that they are identical. This truth can be verified by anyone
who has tried to learn two languages, one genetically close to his native
language and one genetically remote. The language which is genetically
close has extensive familiar areas, which require little or no learning. For
example, a Pole attempting to learn Slovak will find this language con-
siderably easier to acquire than a Portuguese learning Slovak. This is
particularly well seen in the case of receptive skills, listening and reading.
All Poles and Slovaks living in the border areas daily experience the
benefits of this situation. To various degrees, this concerns such languages
as Bulgarian and Macedonian, Dutch and German, Swedish and Nor-
wegian, Finnish and Estonian, and numerous other languages, repre-
senting various families. The same languages, which seem so easy to
people using genetically closely related languages, may appear to be
enormously difficult to people speaking genetically remote languages.
The truth of these claims is so obvious that it requires no further
comments or illustrations, but the only way in which it is possible to
account for these facts is by saying that some languages indeed share
more common properties than others.
Therefore, for every language, it is possible to arrange all other lan-
guages in the ascending order of difference, and, at least in principle, of
difficulty. But this type of scale would only define one kind of difficulty,
i. e. the wholesale, overall degree of difficulty of particular languages in
comparison with a given language. Specific grammatical phenomena
could not be plotted on a similar hierarchy in which difference and
difficulty would be strictly correlated, due to the reasons mentioned earlier
on. James is certainly aware of these problems when he mentions them
twice in his book. Unfortunately, in this case he contradicts himself again.
Having referred to Lado's observation that what is the same in a foreign
language requires no learning on the part of the learner inasmuch as he
already knows it (Lado 1957: 7), James proceeds to draw the following
conclusion:

A bizarre conclusion from this claim is that since all languages have
something in common — the 'linguistic universale' — each of us knows
at least parts of languages we have never heard or read. This is however
a reductio ad absurdum, not to be taken seriously (James 1980: 17).
Yet, towards the end of his book, James seems to have changed his
mind on this issue as well since he thus writes about identical places
across languages:
240 Chapter XII

Our assumption is that the LI L2 identities will not have to be learned

by the L2 learner, since he knows them already by virtue of his LI
knowledge. Thus, though I have never attempted to learn Icelandic,
some aspects of this language are nevertheless known to me in advance:
on the one hand, I 'know' those features of Icelandic that are universal,
and those that are shared by it and my native language. This is not
an absurd claim (p. 151).
Numerous observations as well as some experiments support James'
altered view on the issue. One might add that the native language is not
solely responsible for the facilitation which is involved here. Polyglots
over centuries have unanimously agreed that each successive language is
easier for them to learn than the previous language. This phenomenon
can only be explained by a theory claiming the existence of identical
places across languages, which do not have to be learned: the more
languages one knows the less one has to learn in each successive language.
In other words, in each new language a polyglot finds less and less of
what he still does not know.
Yet, in specific instances "identical" does not necessarily mean easier
to learn, where by "easier to learn" one reads "not conductive to errors"
In many cases, learners expect things to be different in a foreign language.
In Chapter IX we quoted the example of Polish learners of Russian
habitually misplacing the stress in the word ponjal ('he understood') in
spite of the fact that the equivalent Polish word is stressed identically.
Identical areas are also occasionally causes of errors of avoidance: antic-
ipating things to be different in the foreign language, the learner will
avoid identical constructions fearing an error. For example, many Polish
learners of English shun the phrase from time to time under the influence
of the congruent Polish phrase od czasu do czasu. These and many other
examples show that cross-language identity and difference cannot be
uniquely associated with ease and difficulty, respectively. A workable
scale of difficulty must be based on the investigation of various other
parameters in addition to the scales of difference.
One of the burning problems in the theory of contrastive analysis
concerns the selection of the most suitable linguistic model. The following
alternative presents itself: either select the best theory, regardless of
whether or not it can be usefully employed in contrastive analysis or
select the model which is best suited for it. Thus, the contrastive analyst
faces a rather dramatic choice: unless fortune has it that one theory is
best by both these counts, he will be unfaithful either to his linguistics
 'Theoretical" and "applied" contrastive studies 241

or to his contrastive analysis. James suggests two solutions to this di-

lemma:
1. Provide the fullest descriptions of LI and L2 in terms of models which
can best achieve this task, and then translate both into a form which
is model-neutral, a sort of a contrastive-analysis variant of "etalon
language" used in translations (Mel'cuk 1963: 62).
2. In the description concentrate on one of the two L's involved, namely
the one which is to be learnt, following Filipovic's procedure whereby
his contrastive analysis of Serbo-Croatian and English is descriptively
biased towards English as the target language (Filipovic 1975).

There exists a third possibility, and this involves the concept of pro-
totypes and partial pattern matching as described in Chapter XI of the
present work. The comparative procedure would not require of any
particular category or construction in LI to be an ideal realization of
some prototype. The equivalents in L2 would not have to be ideal
realizations of the prototype either. But contrastive analysis would reveal
that such equivalent pairs share some properties but not others, with one
or the other item being closer to the prototype by virtue of sharing more
features with it. This procedure differs from James' solution 1 in that the
two descriptions would not be independent but would both be anchored
in the description of the prototype.
James faces this dilemma practically when he compares advantages
and disadvantages of generative phonology in comparison with taxo-
nomic phonology. James opts for taxonomic phonology, committing
himself in the following way:
Given the choice between taxonomic and generative phonology, while
accepting that the latter is probably more powerful for 'pure' linguistic
purposes, we should, as Burgschmidt and Götz (op. cit.: 199) do, opt
for the former and weaker, for the simple reason that it is more
practical and concrete (James 1980: 82).

But even if one accepts taxonomic phonology one still faces alternative
possible descriptions of the same data in one language. In Chapter V, we
pointed out that for Polish one encounters a number of different analyses
of phonological systems, leading to different inventories of phonemes.
The Polish palatalized obstruents [p', b', t', d \ Γ, v'] can be analysed as
separate phonemes or as positional variants (before /i/) of non-palatalized
homorganic obstruents. The analysis depends on whether [i] is recognized
as an allophone of /i/, mutually exlusive with [i], or whether [i] and [i]
242 Chapter XII

are considered to be in contrast. In this latter case, it becomes necessary

to postulate contrast between palatalized and non-palatalized obstruents.
The two analyses are mutually incompatible, but the second one is
preferable because it allows to reduce the number of Polish obstruents
to non-palatalized ones only and to consider the palatalized obstruents
as positional variants of the non-palatalized ones. This reduction is
accomplished at the expense of having only one more vowel phoneme,
viz. /[/ in contrast with /i/. The principle of economy, so much valued
in taxonomic phonology, is decisive in this case. However, such an analysis
conceals a number of important phonetic (allophonic) differences between
Polish and English. One of these differences consists in the fact that in
English, unlike in Polish, obstruents are not palatalized in front of high
front vowels. Such an analysis will fail to predict negative transfer from
Polish unless the contrastive analysis is continued at the subphonemic
level. This indeed seems necessary in view of the fact that the most
conspicuous deviations from the norms of the target language occur at
the level of phonetic realizations of equivalent phonemes.
The example evoked above again shows that even the selection of one
linguistic model does not guarantee a uniform description within that
model, and the contrastive analyst must be continually on the watch to
seek a balance between the demands imposed by the accuracy of descrip-
tion and the accuracy of comparison. Still, James' readiness to sacrifice
"purely" linguistic merits, such as accuracy and economy, in favour of
such qualities as "more practical and concrete" (p. 82) leaves one with a
certain amount of discomfort, even if this position is probably practically
justified. And yet Gussmann (1984: 29) writes: "we reject any a priori
assumption that some type of phonology may be better suited to con-
trastive ends than others and maintain instead that whatever phonology
is judged to be most adequate in general terms should also be used in
contrastive studies Gussmann's assumption may well point to one of
the reasons why "pure" linguists have little sympathy for the dilemmas
of contrastive analysts: confidently preoccupied with the descriptive ad-
equacy of their models "pure" linguists turn a blind eye to the practical
needs of applied linguists.
The greatest merit of James' book is that it boldly faces all these
difficult problems of which the author is profoundly aware, Instead of
offering pseudosolutions and formulating sweeping and unmotivated
claims, he leaves many of those controversial problems open. Thus, to a
large extent the relevance of contrastive analysis to language teaching
remains a mystery. But this is certainly not James' fault. We still know
 'Theoretical" and "applied" contrastive studies 243

far too little about the nature of foreign language learning in general and
about the nature of possible contrasts in particular. For example, we still
do not know whether there is such a thing as an impossible contrast. If
contrastive analysis could specify possible contrasts in possible human
languages, and if it could rule out impossible contrasts, it would be
related to general linguistic theory in a natural way. But general linguistic
theory is still far from being capable of performing this sort of task. It
is not even possible to formulate reasonable research strategies along
these lines. For the time being, contrastive analysis must be limited to
predicting potential errors, even if one builds it, as James does, on
psycholinguistic foundations. Actual performance must still remain out
of reach.
All these dilemmas and controversies show that contrastive analysis
remains a living discipline, and that interest in it, which has not faded in
the past few years, is well founded in the wealth of theoretical and
practical problems that still remain to be investigated.
Notes

Introduction
Actually, we arc still oversimplifying matters in as much as "similar" cannot be
interpreted as an absolute value but only as relative to a feature or a set of features,
or more generally as "similar in some respect or in some respects" (see Chapter IV).
Furthermore, we do not analyse the concept "language element", which we allow to
subsume such various structural concepts as "phoneme" "morpheme" "syntagm" or
"a syntactic structure" For the purpose of argumentation, we assume that all the
problems inherent in analysing language into such "elements" have been satisfactorily
solved.

Chapter I
In fact, what is compared are particular aspects of languages, such as various gram-
matical systems, constructions or rules, not languages in their entirety. "Comparison
of languages" is an abbreviation embracing such specific comparisons. For details see
Chapter IV.

Chapter II
1. This chapter is largely based on Krzeszowski (1984).
2. One of the few exceptions is provided by James (1980: 90), who mentions "substantive"
tertia comparationis for phonological and lexical contrastive studies, as well as formal
(surface structure), semantic (deep structure) and translation equivalence as the best
tertia comparationis for contrastive studies, provided it embraces both semantic and
pragmatic equivalence (James 1980: 178). No attempt is made, however, to relate
translation equivalence as tertium comparationis to other kinds of possible tertia com-
parationis within the overall landscape of contrastive studies. Substance as tertium
comparationis is also briefly mentioned in Rusiecki (1976: 37).
3. Strictly speaking, squares constitute a proper subset of rectangles, but this does not,
of course, rule out the possibility of comparison.

Chapter III

1. Largely based on Krzeszowski (1989).

2. The terms "pure" and (for example, pedagogically) "oriented" seem to be more fitting
as expressions of the teleological difference between the two kinds of contrastive
activities.
3. No extensive statistical data are available to support these impressionistic observations.
However, Becka (1978: 135) has found that in a sample of 100 sentences from a Czech
translation of Three Man in a Boat by J. K. Jerome, not a single occurrence of participle/
gerund construction is to be found as a condensation of a subordinate (complement)
clause. My own quick count of the corresponding fragment in the Polish version of
the text has yielded the same result. These findings contrast sharply with English, as
in the original text eleven instances of the participle/gerund construction can be found.
246 Notes

Chapter IV
1. In her exhaustive survey of passive constructions in various languages Siewierska (1984:
43—44) considers Basque sentences with the so-called intransitive subject and the
intransitive auxiliary d-a (a form of izati 'be') to be cases of passives. If one accepts
her analysis, Basque and English would constitute situation (b). However, Siewierska
considers the broadest possible sense of the concept "passive" Under narrower senses,
it might be required of passive sentences to have special nominal morphology as well
as syntactic markers as features minimally required of passive sentences in contrast to
active ones. This, as Siewierska's examples show, is not the case in Basque. If this
contrast is required of passives, English and Basque would indeed represent situation
(c).
2. A detailed sociolinguistic analysis of the forms of address in English and Polish can
be found in Ludkiewicz (1985).

Chapter V
1. Characteristically, the most extensive contrastive study of intonation known to me ends
with a similar plea for more research in the domain of emotive aspects of intonation:
A contrastive analysis of emotional-attitudinal intonation would also be worth
making. The difficulties of such an analysis would be considerably greater
The different available descriptions do not examine the same sets of attitudes;
Moreover, in the attitudinal-emotional functioning of language the role of
paralinguistic features (which are largely undescribed) is increased (Varga 1975:
133).
2. In fact many examples discussed in Fisiak et al. do not involve semanto-syntactically
equivalent sentences, i.e., closest approximations to acceptable word-for-word trans-
lations (cf. Chapter VII). We shall return to this notorious problem below.
3. This section is a considerably expanded version of Krzeszowski (1981c).
4. Outside the scope of our investigations remain, for the time being, some further
phenomena which could also be studied contrastively, for instance diachronic distinc-
tions (archaisms), geographical distinctions (regionalisms), metaphorizations (see, how-
ever, Chapter XI), and morphophonological nets in what Nowakowski (1977: 27) calls
the LEXICON.
5. In fact, Lado mentions one more "pattern of difficulty" connected with geographical
distribution of certain lexical items, but since it is discussed within the context of one
language, not as a contrastive phenomenon, we will not discuss it here.
6. Skorupka divides phraseological units into 'zwi^zki frazeologiczne stale, fyczliwe i
luzne' which roughly corresponds to the tripartite division introduced above (Skorupka
1967: 6 - 7 ) .
7. Cf. Preston (1979), where as many as 40 categories of language variation are distin-
guished.
8. I wish to thank Ms. Katarzyna Mostowska for kindly providing Polish pragmatic
equivalents of Fillmore's examples.

Chapter VI
1. For detailed surveys of the classical tradition in grammar see Dinneen (1967: 70 ff.)
and Lyons (1968: 44 ff.).
2. The so-called Saxon genitive is a form intermediate between an inflectional ending and
a function word (Krzeszowski 1980: 140).
 Notes 247

3. One of the earliest fully comprehensive comparative grammars of two modern lan-
guages, written for pedagogical purposes, is Nicolas Salmon's Grammaire Angloise
comparee avec La Grammaire FratiQoise published in London in 1797.
4. For historical accuracy it must be noted that most structuralists distinguished between
two kinds of meaning: structural and lexical. For example, Francis defined structural
meaning as "The meaning which a linguistic structure has over and above the lexical
meanings of the words it contains" (Francis 1954: 595) and lexical meaning as "Meaning
of a morpheme or word apart from the meaning it acquires by virtue of its position in
a larger structure; 'dictionary meaning'" (ibid). Ignoring the problem of circularity and
indeterminacy involved in these "definitions", whenever we talk about "meaning", we
shall focus on "lexical" rather than "structural" aspects of what to us seems an overall,
integrated concept of meaning.
5. Based on Jassem (1954).
6. Based on Jassem (1962, 1966).
7. For a detailed study of such matters in Georgian and East Armenian see Kramsky
(1978).
8. Halle (1964: 326) distinguishes four degrees of narrowing of the vocal tract:
The most extreme degree of narrowing, termed contact, is present when two
opposite parts of the vocal tract touch. Stop consonants such as [p] [d] or [k]
are articulated with contact at different points of the vocal tract.
A less extreme degree of narrowing, termed occlusion, is one capable of
producing turbulence. Occlusions are characteristically involved in the production
of fricatives such as [v] [s] or [s].
The next degree of narrowing, termed obstruction, is exemplified in the artic-
ulation of glides such as [w] or [j].
The fourth degree of narrowing, termed constriction, is that manifest in the
articulation of diffuse ("high") vowels such as [i] or [u],
9. Some phoneticians consider English final consonants in match/mxt$l and bridge/bridal
to be clusters stop + sibilant rather than single segments called affricates. Some
arguments in favour and against either view can be found in Gimson (1962: 166fT.).
See also Jassem (1983: 222 — 223) for the view maintaining that /tf/ and /d$/ are
segments in contrast to his earlier view that they are clusters (cf. Jassem 1954: 38 fT.).
Structural phonology abounds in such alternative analyses. In Chapter V, we presented
two possibilities in the analysis of the Polish vowels. Such alternative analyses have an
obvious and immediate bearing on the results of contrastive studies.
10. In the most recent versions of transformational generative grammar, the concept of
rule no longer occupies the central position. Instead, various universal principles and
parameters are in the focus of attention, cf. e.g., Chomsky (1982). Since similarities
rather than differences (contrasts) preoccupy linguists of this persuasion, we shall not,
at present, evaluate these approaches in terms of their relevance to contrastive studies.
11. Curiously enough, Marton's definition of congruence is silent on matters of stress and
other suprasegmental phenomena, which have turned out to be of critical importance
in establishing cross-language correspondences. It must also be noted that later models
promoted the so-called lexicalist hypothesis whereby the discussed items were listed in
the dictionary (or derived by word-formation rules) rather than derived transforma-
tionally (Chomsky 1972).
12. Van Buren (1974: 303 ff.) expertly discusses some further difficulties connected with
implementing the standard theory in contrastive studies.
13. For some ideas on how to incorporate various elements of Μ in the case grammar see
Dillon (1973: 271 - 2 7 9 ) and Stockwell et al. (1973: 27 fT.).
14. The revisions represent two major trends: increasing the number of cases (e.g., Nilsen
1972) and limiting the number of cases, usually to four (e.g., Anderson 1971; Krze-
szowski 1974, 1979).
248 Notes

Chapter VII
1. Based on Krzeszowski (1971, 1981b).
2. At this point we ignore the question of aspect in Polish. We shall presently return to
this problem, which slightly complicates the issue in question.
3. Since the verb uzyc lacks the synthetic form of the iterative, we provide another example
to Fill the existing systematic gap of the verb in question.
4. A detailed discussion of the analogous English-Polish contrast can be found in Cygan
(1974).

Chapter VIII
The account of contrastive generative grammer in this chapter is a summary of
Krzeszowski (1974, 1979: 1 9 - 2 0 ) .
2. This representation is oversimplified since the predicate can be further analysed. In the
subsequent part of this chapter, we shall outline a way in which sentences such as (1)
and (2) can be analysed semantically in a more detailed way.
3. In Krzeszowski (1979) I erroneously analysed
(a) The invitation of the doctor surprised John
and its Polish translations
(b) To, ze doktor zostal zaproszony, wywolalo zdumienie Johna
(c) To, ze doktor zaprosil kogos, wywolalo zdumienie Johna
as semanto-syntactically equivalent. In view of the restricted definition of semanto-
syntactic equivalence, I now see that neither (b) nor (c) can be recognized as semanto-
syntactic equivalents of (a). The only possible such equivalents of (a) in Polish is
(d) Zaproszenie doktora zdumialo Johna.
which is ambiguous in the same way as (a), so the problem of disambiguation does
not arise in this case since contrastive generative grammar will correctly match (a) and
(d) as a semanto-syntactic equivalent on the basis of the identity of their (at least)
double semantic inputs. Naturally, this example does not affect the gist of the argu-
mentation presented here.
4. Reportedly, Navaho (Apresjan 1966, 1971: 5 7 - 5 8 ) and Nootka (Hockett 1958: 2 2 4 -
225) are void of the distinction between verbs and nouns. On the conceptual level
"things" in Navaho do not exist in a three-dimensional space but rather only in time,
so that everything is in the state of continual "verbing", a dynamic process resembling
Heraclitean panta rhei. In a fictitious language called Tlön, described by J. L. Borges
in one of his Ficciones, entitled "Tlön, Uqbar, Orbis Tertius", one encounters the same
situation as regards the dynamic status of concepts which in most languages are confined
to three-dimensional space and expressed as "nouns" rather than as verbs.
5. The view of reference adopted here is compatible with the concept of ICM's (Lakoff
1982: 48 ff. 1986: 68 ff.), relative to which meanings of words are defined, rather than
relative to elements of the world at large. For details see Sampson (1970a, b) and, in
a somewhat different context Ludskanov (1972, particularly pp. 78 ff.).
6. Such an understanding of the notion of reference as a composite association makes it
possible to explain why a semantic interpretation of sentences like
(a) I spoke to a man who spent three years on Mars
is possible even if there exists no phenomenon in the outer reality with which a man
of (a) could be associated. Nevertheless, there exists a relation of reference between the
modified noun of (a) and some mental image of a man present in my mind. The
 Notes 249

potential multiple association of one referent with more than one linguistic expression
can be seen in the possibility of associating such forms as the man whom I met yesterday,
Mr. McClusky or my greatest enemy with a single referent.
It may be argued, following Kurylowicz (1971: 9), that "future" is different from "past"
and "present" in that it is a sphere of time embracing events which are not real in the
sense that they only exist as predictions, that is, as acts of the speaker's mind. The
modal character of the future tense is well grasped by Boyd and T h o m e ' s performative
analysis of modal verbs in English (Boyd —Thome 1969). In terms of contrastive
generative grammar, could formulate the appropriate representation alternative to the
one suggested here, which derives the future tense from fTL^

Chapter IX
Contrastive generative grammar does not attempt to account for actual processes of
sentence production by the foreign language learner inasmuch as it is an account of
competence, not of performance. Therefore, the descriptions of equivalent constructions
across fully elaborated codes, as well as across fully elaborated codes and "reduced"
codes cannot be interpreted as descriptions of what the learners actually do when they
produce or understand sentences (cf. Chesterman 1980: 18 ff.). However, the descriptive
account furnished by contrastive generative grammar may suggest some strategies in
psycholinguistic research whereby the accuracy of such descriptions may receive addi-
tional support. For some evidence to that effect see below.
2. Another "cosmetic" transformation will ensure agreement of the verb with the subject
N P if the present tense is involved {do becomes does in the singular).
3. "Premature" by comparison with the situation in the fully elaborated code.
4. "Occasional" questions contain the Wi-word in the position before the preposing and
are uttered with heavy stress and rising intonation. "If someone said 'John will read
the telephone book', one might respond, 'John will read what Τ this response is an
occasional form" (Brown 1970: 133). Therefore, such questions must be derived from
other sources than those that underlie normal w/i-questions, and, consequently, they
cannot be considered as representing an intermediate stage in the derivation of normal
H'A-questions.

Chapter X
1. Based on Krzeszowski (1981d).
2. A parody typically involves a high concentration of forms which are already outstand-
ingly frequent in the text(s) which serve(s) as the object of parody. One can envisage
studies aimed at a statistical account of parody.
3. For an excellent phonological quantitative contrastive study see Kramsky (1978).
4. Such a correlation is highly plausible as Polish employs infinitives in certain ways in
which English does not, for example as categorical imperatives in trzymac sig mocno,
chlopaki 'hold on, boys' or in such constructions as trzeba nam myslec po nowemu 'it
is necessary that we think in new ways' (lit. 'needs us to think anew'). Without more
extensive investigations in more than two languages, it is impossible to demonstrate a
direct correlation between more restricted contexts in which particular forms appear
at the level of language description and less frequent occurrences of these forms in
continuous texts even if such correlations can be hypothetically expected.
5. A very detailed systematic contrastive study of morphological, syntactic, and semantic
properties of Polish nominals in contrast with English ones can be found in Lewan-
250 Notes

dowska (1975) and need not be recapitulated here. A detailed description of Polish
nominals can be found in Puzynina (1969) and Grzegorczykowa (1979). The equivalent
English phenomena are described in Lees (1963) and Fraser (1970).

Chapter XI
1. The present chapter is a considerably expanded version of Krzeszowski (1986b).
2. According to Lakoff (1982), gestalts exhibit various other characteristics which need
not concern us at this point.
3. Sometimes a pragmatically valid translation will resemble the original only by the
rhythmic pattern, as in the case of nursery rhymes or various fixed expressions which
do not convey any meaning outside a strictly limited historical and cultural context of
a given nation or a group of people. Such in the case with the famous couplet from
one of Mickiewicz's poems:
Emilia Plater
Dziewica-bohater
which is a distorted, folk version of two lines from Smierc pulkownika (The colonel's
death), functioning very much like a nursery rhyme with distinctly comic overtones
parasitic on a historical allusion. An inspired translator once extemporized the following
English version:
Helen Shapiro
virgin-hero
achieving the comic effect which unexpectedly issues from the originally pompously
sounding poem and not bothering to find an English historical equivalent of the said
hero or trying to explain her place in Polish history. It is this comic effect, reinforced
by the rhyme in which the word 'hero' matches the name of a girl, which received the
privileged status and determined pragmatic equivalence in this particular case.
4. Two linguistic forms belong to one category within the domain of contrastive studies
if there is a tertium comparationis whereby they can be compared. The common tertium
comparationis delimits the category membership in the following way: saying that a
form A in LI is in some way equivalent to form Β in L2 entails that there is such a
tertium comparationis whereby they can be compared.
5. Kalisz (1986: 1251) suggests that (6) may be a better (pragmatic?) equivalent of (1)
than (5) inasmuch as (6), like (1) is a passive sentence while (5) is an impersonal
sentence. Thus, other things being equal, (6) shares one more property with (1) than
does (5). Kalisz seems to be confusing pragmatic equivalence with semanto-syntactic
equivalence (for which he has designed the unfortunate terms "overall equivalence");
(6) is certainly the semanto-syntactic equivalent of (1), being the closest approximation
to acceptable word-for-word translation of (1) (cf. Chapter VII). It does not necessarily
follow that it is the best pragmatic equivalent since passive constructions in Polish do
not enjoy the same stylistic status as they do in English. Instead, Polish makes ample
use of impersonal constructions such as (5). In this case stylistic considerations may
override syntactic considerations, so that (5) may be a better pragmatic equivalent of
(1). Again the privileged status of properties rather than their mere number seems to
be decisive.
References

Abbreviations
IRAL International Review of Applied Linguistics in Language Teaching
ISB Interlanguage Studies Bulletin (Utrecht)
JESL Journal of English as a Second Language (Philadelphia, Pennsylvania)
L.A.U.T. Linguistic Agency. University of Trier
MSLL Monograph Series on Languages and Linguistics (Georgetown University.
School of Languages and Linguistics)
PAKS Projekt für Angewandte Konstrastive Sprachwissenschaft (Institut für Li-
teratur- und Sprachwissenschaft. Lehrstuhl Anglistik: Linguistik, Uni-
versität Stuttgart)
PSiCL Papers and Studies in Constrastive Linguistics (Adam Mickiewicz Univer-
sity, Poznan, Poland)
PSML Prague Studies in Mathematical Linguistics (Academia, Publishing House
of the Czechoslovak Academy of Sciences)
PTJ Polskie Towarzystwo Jezykoznawcze [Polish Linguistic Association]
SAP Studia Anglica Posnaniensia. An International Review of English Studies.
Adam Mickiewicz University, Poznan
SSLA Studies in Second Language Acquisition. Indiana University
YSCECP The Yugoslav Serbo-Croatian English Contrastive Project. (Institute
of Linguistics. Faculty of Philosophy. University of Zagreb, Yugoslavia:
Center for Applied Linguistics, Washington D.C., U.S.A.)

Aarts, Flor —Herman Wekker

1982 "Writing a contrastive grammar of English and Dutch", PSiCL 14: 25 —
43.
Agard, Frederick B. —Robert J. Di Pietro
1965 The sounds of English and Italian (Chicago — London: University of Chicago
Press).
Allen, J. P. B . - S . Pit Corder (eds.)
1974 Papers in applied linguistics (Edinburgh Course in Applied Linguistics I)
(Oxford: Oxford University Press).
Anderson, John M.
1971 The grammar of case: towards a localistic theory (Cambridge Studies in
Linguistics) (Cambridge: Cambridge University Press).
Anderson, Stephen R.
1974 The organization of phonology (New York: Academic Press).
Apresjan, Jurij D.
1966 Idei i metody sovremennoj lingvistiki [Concepts and methods in contempo-
rary linguistics] (Moskva: Izdatelstvo Prosvescenie).
1971 Koncepcje i metody wspoiczesnej lingwistyki strukturalnej [Concepts and
methods in contemporary linguistics]. Translated by Zygmunt Saloni (War-
szawa: Panstwowy Instytut Wydawniczy).
Arabski, Janusz
1979 Errors as indications of the development of interlanguage (Katowice: Uni-
wersytet Sl^ski).
252 References

Arnol'd, I. W.
1973 The English word: leksikologija sovremennogo anglijskogo jazyka [Lexicology
of modern English] (Moskva: Vyssaja Skola).
Awedyk, Wieslaw
1974 "The syllabic structures of English and Polish", PSiCL 2: 8 9 - 9 5 .
1976 "Some remarks on generative contrastive phonology", PSiCL 4: 53 — 60.
1979 "Towards pedagogical contrastive phonology", PSiCL 10: 1 2 5 - 1 3 3 .
Bach, Emmon
1968 " N o u n s and noun phrases", in: Bach —Harms (eds.), 91 — 122.
Bach, Emmon —Robert T. Harms (eds.)
1968 Universals in linguistic theory (New York: Holt, Rinehart and Winston).
Banathy, Bela H. — Paul H. Madarasz
1969 "Contrastive analysis and error analysis", JESL 4.2: Π — 92.
Bartsch, Renate —Theo Vennemann
1972 Semantic structures (Frankfurt am Main: Athenäum).
Becka, J. V.
1978 "Application of quantitative methods in contrastive stylistics", PSML 6:
129-147.
Berault, Peter
1688 A new, plain, short, and compleat French and English grammar (London).
Biedrzycki, Leszek
1964 "Fonologiczna interpretacja polskich glosek nosowych" [A phonological
interpretation of Polish nasals], Biuletyn PTJ 23: 25 — 46.
Bloomfield, Leonard
1933 Language (New York: Holt).
Boas, Hans Ulrich
1977 "Some remarks on case grammars as bases for contrastive studies", PSiCL
7: 2 1 - 3 2 .
Bolinger, Dwight
1965 "The atomization of meaning", Language 41: 555 — 573.
1966 "Transformation: structural translation", Acta Linguistica Hafniensia 9.2:
130-144.
Borkowski, G —L. Mickelsen
1963 "A contrastive study of the impersonal sentences of Polish and Russian"
(Paper presented at the annual meeting of the Linguistic Society of America,
Chicago, December 28 — 30).
Bouton, Lawrence F.
1976 "The problem of equivalence in contrastive analysis", IRAL 14.2: 143 —
163.
Boyd, Julian —James P. Thorne
1969 "The semantics of modal verbs", Journal of Linguistics 5: 57 — 74.
Briere, Eugene J.
1968 A psycholinguistic study of phonological interference (The Hague: Mouton).
Bross, J. S.
1962 "Problems of equivalence in some German and English constructions",
Mechanical Translation 7: 8 — 16.
Brown, Roger
1970 Psycholinguistics: selected papers (New York: MacMillan).
Brugman, Claudia
1981 Story of over [M.A. thesis] (Bloomington, Indiana: Indiana University
Linguistics Club).
 References 253

Burgschmidt, Ε. —D. Götz

1974 Kontrastive Linguistik, Deutsch-Englisch: Theorie und Anwendung (Mün-
chen: Hueber).
Catford, John C.
1965 A linguistic theory of translation (Oxford: Oxford University Press).
1968 "Contrastive analysis and language teaching", MSLL 21: 159 — 173
(Georgetown University Press).
Chesterman, Andrew
1980 "Contrastive generative grammar and the psycholinguistic fallacy", PSiCL
11: 1 7 - 2 4 .
Chomsky, Noam
1957 Syntactic structures (The Hague: Mouton).
1965 Aspects of the theory of syntax (Cambridge, Mass.: M.I.T. Press).
1972 Studies on semantics in generative grammar (The Hague — Paris: Mouton).
1975 Reflections on language (New York: Pantheon).
1982 Some concepts and consequences of the theory of government and binding
(Cambridge, Mass. — London: M.I.T. Press).
1984 Modular approaches to the study of the mind (San Diego: San Diego Uni-
versity Press).
Chomsky, Noam —Morris Halle
1968 The sound pattern of English (New York: Harper and Row).
Coles, Elisha
1675 Syncrisis: a very important introduction about what is similar and what is
different and how it facilitates learning (London).
Comrie, Bernard
1976 "The syntax of causative constructions: cross-language similarities and
divergencies", in: M. Shibatani (ed.), Syntax and semantics (New York:
Academic Press), 261 —312.
Corder, S. Pit
1967 "The significance of learners' errors" IRAL 4: 161 — 170 [1974] [reprinted
in: J. C. Richards (ed.) Error analysis. London: Longman 19 — 27].
1973 Introducing applied linguistics (Harmondsworth: Penguin Education).
1975 "'Simple codes' and the source of the second language learners' initial
heuristic hypothesis", SSLA 1.1: 1 — 10.
Cygan,Jan
1974 "Negation in English and Polish", PSiCL 2: 2 9 5 - 3 2 9 .
Danilewicz, Tadeusz
1982 Ways of expressing cause in English and Polish [unpublished doctoral
dissertation, Uniwersytet Sl^ski],
De Camp, David
1968 "The field of creole language studies" SAP 1: 2 9 - 5 1 .
De Geest, W.
1979 The relevance of relational grammar to contrastive linguistics. L.A.U.T. Series
B. no. 39.
Delattre, Pierre
1965 "Comparing the consonantal features of English, German, Spanish and
French" IRAL 2: 1 5 5 - 2 0 3 .
Dillon, George L.
1973 "Perfect and other aspects in a case grammar of English" Journal
Linguistics 9: 271 - 2 7 9 .
Dingwall, William O.
1964a Diaglossic grammar [unpublished Ph.D. dissertation, Georgetown Univer-
sity Washington D.C.].
254 Reference.

1964b "Transformational generative grammar and contrastive analysis", Language

Learning 14: 147—160.
Dinneen, Francis P.
1967 An introduction to general linguistics (New York — Chicago — San Fran-
cisco—Toronto—London: Holt, Rinehart and Winston).
Di Pietro, Robert J.
1971 Language structures in contrast (Rowley, Mass.: Newbury House).
Dluska, Maria
1976 Prozodia jgzyka polskiego [Prosody of the Polish language] (Warszawa:
Panstwowe Wydawnictwo Naukowe).
Duczmal, Stanislaw
1979 "A contrastive semantic analysis of colour adjectives in Polish and English",
PSiCL 9: 1 8 1 - 1 9 1 .
Duskovä, Libusa
1978 "The simple sentence in Czech and English", PSML 6: 83 — 92.
Enkvist, Nils Erik
1973 "Should we count errors or measure success?", in: Jan Svartvik (ed.), Errata:
Papers in error analysis (Lund: Gleerup), 16 — 23.
1977 "Contextual acceptability and error evaluation", in: Papers from the con-
ference on contrastive linguistics and error analysis. Stockholm and Abo
Akademi Foundation), 1—26.
Ferguson, Charles A.
1964 "Baby-talk in six languages", in: The ethnography of communication ( =
American Anthropologist 66(6), pt. 2) (Washington D.C.: American An-
thropological Association), 103 — 104.
1975 "Towards a characterization of English foreign talk" Anthropological Lin-
guistics 17: 1 — 14.
Filipovic, Rudolf (ed.)
1971 The Yugoslav Serbo-Croatian English contrastive project (Zagreb).
1975 Contrastive analysis of English and Serbo-Croatian (Zagreb: Institute of
Linguistics).
Fillmore, Charles
1968 "The case for case", in: Bach —Harms (eds.), 1 —88.
1984a "Remarks on contrastive pragmatics", in: Jacek Fisiak (ed.), 119 — 141.
1984b Frames and the semantics of understanding [unpublished ms. Berkeley:
Department of Linguistics, University of California].
Fisiak, Jacek
1971 "The Poznan Polish-English contrastive project", in: Filipovic (ed.), 87 —
96.
1975a "The contrastive analysis of phonological systems", Kwartalnik Neofilolo-
giczny 22: 341—351.
1975b Wstgp do wspolczesnych teorii lingwistycznych [Introduction to contempo-
rary linguistic theories] (Warszawa: Wydawnictwa Szkolne i Pedagogiczne)
(revised edition 1978).
1976 "Generative phonological contrastive studies", Kwartalnik Neofilologiczny
23: 1 1 9 - 1 2 4 .
1983 "Present trends in contrastive linguistics", in: Kari Sajavaara (ed.), Cross-
Language Analysis and Second Language Acquisition 1 ( = Jyväskylä Cross
Language Studies 9), 9 - 3 8 .
1984 "On the roots of contrastive linguistics", Folia Linguistica 18: 139—153.
Fisiak, Jacek (ed.)
1980 Theoretical issues in contrastive linguistics (Amsterdam: Benjamins).
 References 255

1981 Contrastive linguistics and the language teacher (Oxford —New York —To-
ronto — Sydney — Paris — Frankfurt: Pergamon).
1984 Contrastive linguistics. Prospects and problems (Berlin —New York —Am-
sterdam: Mouton).
Fisiak, Jacek —Maria Lipinska-Grzegorek — Tadeusz Zabrocki
1978 An introductory English-Polish contrastive grammar (Warszawa: Panstwowe
Wydawnictwo Naukowe).
Francis, W. Nelson
1954 The structure of American English (New York: Ronald Press).
Fraser, Bruce
1970 "Some remarks on the action nominalization in English", in: R . J a c o b s —
P. Rosenbaum (eds.), Readings in English transformational grammar (Wal-
tham, Mass. —Toronto —London: Ginn), 83—98.
Fries, Charles C.
1945 Teaching and learning English as a foreign language (Ann Arbor: University
of Michigan Press) [ l4 1960],
1952 The structure of American English: an introduction to the construction of
English sentences (London: Longmans).
Gazdar, Gerald
1979 Pragmatics: implicature, presupposition, and logical form (New York —San
Francisco —London: Academic Press).
Ginzburg, R. S . - S . S. K h i d e k e l - G . Y. K n y a z e v a - A . A. Sankin
1966 A course in modern English lexicology (Moscow: Higher School Publishing
House).
Gimson, A. C.
1962 An introduction to the pronunciation of English (London: Edward Arnold).
Gleason, Henry Α., Jr.
1955 Introduction to descriptive linguistics (New York: Holt) [revised edition 1961 ].
1965 Linguistics and English grammar (New York — Chicago — San Francisco —
Toronto —London: Holt, Rinehart and Winston).
Gniadek, Stanislaw
1979 Grammaire contrastive franco-polonaise (Warszawa: Panstwowe Wydawn-
ictwo Naukowe).
Graczyk, Ireneusz
n.d. On writing English as a second language: verb phrase in advanced learners'
written English, I (Ms. Uniwersytet Lodzki).
Grzegorczykowa, Renata
1979 Zarys slowotworstwa polskiego: slowotworstwo opisowe 3 [An outline of
Polish word-formation: descriptive word-formation] (Warszawa: Panstwowe
Wydawnictwo Naukowe).
Gussmann, Edmund
1978 Contrastive Polish-English consonantal phonology (Warszawa: Panstwowe
Wydawnictwo Naukowe).
1980 Introduction to phonological analysis (Warszawa: Panstwowe Wydawnictwo
Naukowe).
1984 "Contrastive analysis substantive evidence and the abstractness issue" in:
Stig Eliasson (ed.), Theoretical issues in contrastive phonology (Heidelberg:
Groos), 2 7 - 3 6 .
Halle, Morris
1964 "On the bases of phonology", in: Jay A. Fodor — Jerrold Jay Katz (eds.).
The structure of language (Englewood Cliffs, N.J.: Prentice-Hall, 324 — 333.
1973 "Prolegomena to a theory of word formation", Linguistic Inquiry 4.1:3 — 16.
256 References

Halliday, Michael A. K.
1970 "Language structure and language function", in: John Lyons (ed.), New
horizons in linguistics (Harmondsworth: Penguin), 140—165.
Halliday, Michael A.K. —Angus Mcintosh — Peter Strevens
1964 The linguistic sciences and language teaching (London: Longmans).
Harris, Zellig S.
1946 " F r o m morpheme to utterance", Language 22: 161 — 183.
1951 Methods in structural linguistics (Chicago: University of Chicago Press).
Hewes, John
1624 A perfect survey of the English tongue taken according to the use and analogie
of the Latine (London).
Hill, L. A.
1968 Prepositions and adverbial particles (London: Oxford University Press).
Hiz, Henry
1968 Referentials [Ms. University of Pennsylvania, Philadelphia, Pa.].
Hjelmslev, Louis
1961 Prolegomena to a theory of language. Translated by J. Whitfield (Madison:
University of Wisconsin Press).
Hockett, Charles F.
1958 A course of modern linguistics (New York: MacMillan).
Howel(l), James
1662 A new English grammar (London).
Hudson, Richard A.
1976 Arguments for a non-transformational grammar (Chicago —London: Uni-
versity of Chicago Press).
Hymes, Dell
1972 "On communicative competence", in: J. B. Pride—J. Holmes (eds.), Socio-
linguistics (London: Penguin), 269 — 293.
Ivir, Vladimir
1969 "Contrasting via translation", YSCECP B. Studies 1: 1 3 - 2 5 .
1970 "Remarks on contrastive analysis and translation", YSCECP B. Studies 2:
14-25.
Jaeger, Jeri
1980 Categorization in phonology: an experimental approach [Unpublished
Ph.D. dissertation, Berkeley: University of California].
Jakobson, Roman —Morris Halle
1956 Fundamentals of language ('s-Gravenhage: Mouton).
James, Carl
1974 "Linguistic measures for error gravity", Audio-visual Language Journal
(Birmingham): 3 — 9.
1977 "Judgments of error gravities", English Language Teaching Journal 21: 116 —
124.
1980 Contrastive analysis (Harlow, Essex: Longman).
Janicki, Karol
1985 "Tertium comparationis in contrastive sociolinguistics", Nordlyd 10: 7 — 27.
Jassem, Wiktor
1954 Fonetyka jezyka angielskiego [Phonetics of the English language] (Warszawa:
Panstwowe Wydawnictwo Naukowe).
1962 "The distinctive features of Polish phonemes", Speech translation laboratory.
Quarterly progress and status report 1 (Stockholm: Royal Institute of Tech-
nology), 7 — 15.
1966 "The distinctive features and the entropy of Polish phoneme system",
Biuletyn PTJ 24: 8 7 - 1 0 8 .
 References 257

1983 The phonology of modern English (Warszawa: Panstwowe Wydawnictwo

Naukowe).
Jones, Daniel
1950 The phoneme (Cambridge: HefTer).
Kakietek, Piotr
1980 English modal auxiliaries and their equivalent constructions in Polish (Ka-
towice: Uniwersytet Sl^ski).
Kalisz, Roman
1976 "On the kinship terms in English and Polish", PSiCL 5: 2 5 7 - 2 7 0 .
1981 The pragmatics, semantics and syntax of English sentences with indicative
that complements and the Polish sentences with ze complements. A contrastive
study (Gdansk: Uniwersytet Gdanski).
1986 "More on pragmatic equivalence", in: D. Kastovsky — A. Szwedek (ed.)
(Berlin —New York —Amsterdam: Mouton de Gruyter), 1247 — 1255.
Kalogjera, Damir
1971 "The expression of future time in English and Serbo-Croatian", YSCECP
Reports 2: 1 2 0 - 1 3 4 .
Karttunen, Kimno
1977 "The equivalents of the Finnish passive voice in English", Jyväskylä Con-
trastive Studies 4: 1 0 9 - 1 2 5 .
Kastovsky, Dieter —Aleksander Szwedek (eds.)
1986 Linguistics across historical and geographical boundaries 1 — 2 (Berlin — New
York —Amsterdam: Mouton de Gruyter).
Katz, Jerrold J . - P a u l M. Postal
1964 An integrated theory of linguistic descriptions (Research Monograph No. 26)
(Cambridge, Mass.: The M.I.T. Press).
Kielski, Boleslaw
1957 Struktura jqzyköw francuskiego i polskiego w swietle analizy porownawczej
[A comparative analysis of the structure of French and Polish] (Lodz:
Lodzkie Towarzystwo Naukowe).
Kohler, Klaus
1971 "On the adequacy of phonological theories in contrastive studies", in:
Gerhard Nickel (ed.), Papers in contrastive linguistics (Cambridge: Cam-
bridge University Press), 83 — 89.
Kohn, Kurt
1974 Kontrastive Syntax und Fehlerbeschreibung (Kronberg, Taunus: Scriptor).
Kolbert, Jack —Harry Goldby
1958 The first French handbook for teachers in elementary schools (Pittsburgh).
Konderski, Stefan
1973 "Some surface and deep aspects of case in Polish and English", PSiCL 1:
65-77.
König, Ekkehard
1970 "Transformational grammar and contrastive analysis", PAKS 6 (Stuttgart),
43-58.
Kopczyhski, Andrzej
1968 English and Polish consonant phonemes [unpublished doctoral dissertation.
Uniwersytet im. A. Mickiewicza, Poznan],
1973 "The Polish and English fricatives a problem in phonological equiva-
lence", PSiCL 1: 9 3 - 9 8 .
1980 Conference interpreting: some linguistic and communicative problems (Poz-
nan: Uniwersytet im. A. Mickiewicza. Seria filologia angielska nr. 13).
Koutsoudas, Andreas
1972 "The strict order fallacy", Language 48: 8 8 - 9 6 .
258 References

Kramsky, Jiri
1978 "Quantitative analysis of near-identical phonological systems", PSML 6:
9-38.
Kryk, Barbara
1987 On deixis in English and Polish. The role of demonstrative pronouns (Frank-
furt am Main: Lang).
Krzeszowski, Tomasz P.
1967 "Fundamental principles of structural contrastive studies", Glottodidactica
2: 3 3 - 4 0 .
1968 An outline of American English phonetics (Warszawa: Panstwowe Wydawn-
ictwo Naukowe).
1970 Teaching English to Polish learners (Warszawa: Panstwowe Wydawnictwo
Naukowe).
1971 "Equivalence, congruence and deep structure", in: Gerhard Nickel (ed.),
Papers in contrastive linguistics (Cambridge: Cambridge University Press),
37-48.
1974 Contrastive generative grammar: theoretical foundations (Lodz: Uniwersytet
Lodzki). [1979] [Reprinted: Tübingen: Narr.]
1976 "On some linguistic limitations of classical contrastive analyses", PSiCL 4:
8 9 - 9 5 [reprinted in: Jacek Fisiak (ed.) 1980, 1 9 3 - 1 9 9 ] .
1980 Gramatyka angielska dla Polakow [English grammar for Poles] (Warszawa:
Panstwowe Wydawnictwo Naukowe).
1981a "Tlumaczenie j a k o czynnosc pragmatyczna" [Translation as a pragmatic
activity], in: Materialy ζ IV Sympozjum Instytutu Lingwistyki Stosowanej
Uniwersytetu Warszawskiego, Jachranka 3 — 5 listopad 1976 (Warszawa:
Wydawnictwa Uniwersytetu Warszawskiego).
1981b "The problem of equivalence revisited", IRAL 19: 1 1 3 - 1 2 8 .
1981c "What do we need lexical contrastive studies for?", PSiCL 3: 1 3 3 - 1 4 8 .
198Id "Quantitative contrastive analysis", Studia Linguistica 35: 43 — 51.
1984 "Tertium comparationis", in: J. Fisiak (ed.), 301 —312.
1985 "The so-called 'sign theory' as the first method in contrastive linguistics",
in: Ursula Pieper —Gerhard Stickel (eds.), Studia linguistica diachronica et
synchronica (Berlin —New York —Amsterdam: Mouton de Gruyter), 485 —
501.
1986a "An Elizabethan contrastive grammar of Spanish and French", in:
D. K a s t o v s k y - A . Szwedek (eds.), 1 3 0 3 - 1 3 1 0 .
1986b "Prototypes and equivalence", PSiCL 21: 5 - 2 0 .
1989 "Towards a typology of contrastive studies", in: Wieslaw Oleksy (ed.), 55 —
72.
(in prep.) "Early contrastive studies in England"
Kudzinowski, Czeslaw
1978 Gramatyka jgzyka finskiego [Grammar of the Finnish language] (Poznan:
Uniwersytet im. A. Mickiewicza).
Kufner, Herbert L.
1962 The grammatical structures of English and German (Chicago: University of
Chicago Press).
Kühlwein, Wolfgang
1983 "Sociosemiotics across cultures", in: Proceedings of a symposium held at St.
Patrick's College, Drumcondra, Dublin, 8 — 9 July 1983: 5 — 19.
Kurylowicz, Jerzy
1971 "Podstawowe kategorie morfologiczne" [Fundamental morphological cat-
egories], Biuletyn PTJ 28: 3 - 1 2 .
 References 259

Lado, Robert
1957 Linguistics across cultures (Ann Arbor: University of Michigan Press).
1964 Language teaching: a scientific approach (New York: McGraw-Hill).
Lakoff, George
1968 "Instrumental adverbs and the concept of deep structure", Foundations of
language 4.1: 4 — 29.
1970 "Global rules", Language 46: 6 2 7 - 6 3 9 .
1971 "On generative semantics", in: D. Steinberg —L. Jacobovits (eds.). Seman-
tics (Cambridge: Cambridge University Press), 232 — 296.
1973 "Fuzzy grammar and the performance/competence terminology game"
Papers from the ninth regional meeting of the Chicago Linguistic Society
(Chicago: University of Chicago Press), 271—291.
1977 "Linguistic gestalts", Papers from the thirteenth regional meeting of the
Chicago Linguistic Society (Chicago: University of Chicago Press), 236 —
287.
1982 Categories and cognitive models. L.A.U.T. Series A. no. 96.
1987 Women, fire, and dangerous things: what categories reveal about the mind
(Chicago: University of Chicago Press).
Lakoff, George —Mark Johnson
1980 Metaphors we live by (Chicago —London: University of Chicago Press).
Lakoff, Robin
1972 "Language in context" Language 48: 907 — 927.
Lamminmäki, Reijo
1979 "The discrimination and identification of English vowels, consonants, junc-
tures and sentence stress by Finnish comprehensive school pupils" Jyväs-
kylä Cross-Language Studies 7: 165 — 231.
Langacker, Ronald W.
1987 Foundations of cognitive grammar 1 (Stanford: Stanford University Press).
Langendoen, Terence D.
1970 Essentials of English grammar (New York: Holt, Rinehart and Winston).
Lawendowski, Boguslaw
1974 A hypothesis concerning emotive aspects of lexis in English and Polish
[Ms., Uniwersytet Warszawski].
Lee, William R.
1968a " H o w can contrastive linguistic studies help foreign language teaching?",
YSC EC Ρ Β. Studies 5 7 - 6 6 .
1968b "Thoughts on contrastive linguistics in the context of language teaching"
MSLL 21 (Washington D.C.: Georgetown University Press), 1 8 6 - 1 9 4 .
Leech, Geoffrey
1974 Semantics (Harmondsworth: Penguin).
1983 Principles of pragmatics (London —New York: Longman).
Leech, Geoffrey —Jan Svartvik
1975 A communicative grammar of English (London: Longman).
Lees, Robert B.
1963 The grammar of English nominalizations (The Hague: Mouton).
Lewandowska, Barbara
1975 "On some properties of action nominals in Polish and their English equiv-
alents", PSiCL 3: 1 6 7 - 1 7 6 .
Lindner, Susan
1981 A lexico-semantic analysis of verb-particle constructions with up and out
(University of California, San Diego Ph.D. dissertation) (L.A.U.T. Series
A, no. 101).
260 References

1982 "What does go up doesn't necessarily come down: the ins and outs of
opposites", in: Proceedings of the eighteenth regional meeting of the Chicago
Linguistic Society (Chicago: University of Chicago Press), 305 — 323.
Lipmska, Maria
1972 Basic sentence patterns in English and Polish [unpublished ms., Uniwersytet
im. A. Mickiewicza, Poznan].
1973 "Some differences between English and Polish on the level of basic sentence
patterns", PSiCL 1: 3 9 - 4 6 .
1975 "Contrastive analysis and the modern theory of language", PSiCL 3: 5 — 62.
Lipinska-Grzegorek, Maria
1977 Some problems of contrastive analysis: sentences with nouns and verbs of
sensual perception in English and Polish (Edmonton, Canada —Champaign,
USA: Linguistic Research).
Liston, Jerry L.
1970 "Formal and semantic considerations in contrastive analysis", Y SC EC Ρ Β.
Studies 2: 2 7 - 4 9 .
Ludkiewicz, Anna
1985 American and Polish address system [unpublished M.A. thesis, Uniwersytet
Gdanski].
Ludskanov, Alexander
1972 Mensch und Maschine als Übersetzer (Halle (Saale): Niemeyer).
Lyons, John
1968 Introduction to theoretical linguistics (Cambridge: Cambridge University
Press).
Mackiewicz-Krassowska, Halina
1974 "Intonation of English and Polish declarative sentences", PSiCL 2: 137 —
144.
Mansion, J. E.
1928 French reference grammar (London — Edinburgh).
Manczak, Witold
1970 Ζ zagadnien jgzykoznawstwa ogolnego [Problems in general linguistics]
(Wroclaw, Warszawa, Krakow: Ossolineum).
Marek, Boguslaw
1974 "Intonation of imperative sentences and requests in Polish and English",
PSiCL 2: 1 6 1 - 1 7 9 .
Markkanen, Raija
1985 "Directives in English and Finnish", Jyväskylä Cross-Language Studies 11:
15-42.
Marton, Waldemar
1968a Noun modification in Polish and English [unpublished doctoral dissertation,
Uniwersytet im. A. Mickiewicza, Poznan],
1968b "Equivalence and congruence in transformational contrastive studies", SAP
1: 5 3 - 6 2 .
1970 "English and Polish nominal compounds: a transformational contrastive
study", SAP 2: 5 9 - 7 3 .
1976 Νowe horyzonty nauczania jgzykow obcych 2 [New horizons in teaching
foreign languages] (Warszawa: Wydawnictwa Szkolne i Pedagogiczne).
Matisoff, James A.
1979 Blessings, curses, hopes and fears: psycho-ostensive expressions in Yiddish
(Philadelphia: Institute for the Study of Human Issues).
McCawley, James D.
1968 "The role of semantics in a grammar", in: E. Bach —R. T. Harms (eds.),
124-169.
 References 261

Meech, S. B.
1935 "Early applications of Latin grammar to English" PMLA 1: 1012 — 1032.
Mel'chuk, Igor A.
1963 "Machine translation and linguistics", in: O. S. Akhmanova et al. (eds.),
Exact methods in linguistic research (Berkeley: University of California
Press), 4 4 - 7 9 .
Miege, Guy
1687 The grounds of the French tongue or a new French grammar according to the
present use, and modern orthography (London).
Mieszek, Aleksandra
1974 "Intonation of compound sentences in English and Polish", PSiCL 2: 181 —
188.
Milewski, Tadensz
1965 Jgzykoznawstwo [Linguistics] (Warszawa: Panstwowe Wydawnictwo Nau-
kowe).
Montague, Richard
1974 Formal philosophy: selected papers (edited by R. H. Thompson) (New Ha-
ven: Yale University Press).
Morgan, Bayard Q. —F. W. Strothmann
1952 Shorter German reading grammar (New York —Boston: Ginn).
Morris, Charles W.
1938 Foundations of the theory of signs (Chicago: University of Chicago Press).
Moulton, William G.
1962 The sounds of English and German (Chicago: University of Chicago Press).
Nemser, William
1971 "Approximative systems of foreign language learners" IRAL9: 115 — 123.
[Reprinted 1974 in: J. C. Richards (ed.), Error Analysis. London: Longman,
55-63],
Newmeyer, Frederick J.
1980 Linguistic theory in America (New York — London — Toronto — Sydney — San
Francisco: Academic Press).
Nickel, Gerhard
1971 "Variables in a hierarchy of difficulty" The Pacific Conference on Contras-
tive Linguistics and Language Universals Papers (Honolulu: University of
Hawaii), 1 8 5 - 1 9 4 .
Nickel, Gerhard —Karl Heinz Wagner
1968 "Contrastive linguistics and language teaching", IRAL 6: 233 — 255.
Nilsen, Don Lee Fred
1972 Toward a semantic specification of deep case (The Hague: Mouton).
Nowakowski, Miroslaw
1977 "The lexicon and contrastive language studies" PSiCL 6: 25—42.
Oleksy, Wieslaw
1977 "Tags in English and equivalent constructions in Polish", PSiCL 7: 9 5 -
109.
1983 "Pragmatic equivalence in contrastive studies: requests in Polish and Eng-
lish", Jyväskylä Cross-Language Studies 10: 79 — 93.
Oleksy, Wieslaw (ed.)
1989 Contrastive pragmatics (Amsterdam/Philadelphia: Benjamins).
Olsson, M.
1972 Intelligibilty: a study of errors and their importance (Gothenburg School of
Education Research Bulletin 12, Göteborg).
Ozga, Janina
1974 "Stress in English and Polish — an introduction to a contrastive analysis",
PSiCL 2: 1 2 3 - 1 3 5 .
262 References

1976 "The relevance of the notion 'basis of articulation' to contrastive phonetics",

PSiCL4: 6 1 - 7 4 .
Pasanen, Maija-Liisa
1978 "Finnish and English verbs of vision", Jyväskylä Contrastive Studies 6: 96 —
122.
Peters, Stanley (ed.)
1972 Goals of linguistic theory (Englewood Cliffs, N.J.: Prentice-Hall).
Pickbourne, James
1789 A dissertation on the English verb (London).
Pike, Kenneth L.
1947 Phonemics: a technique for reducing language to writing (Ann Arbor: Uni-
versity of Michigan Press).
Polanski, Kazimierz
1966 "Glowne typy struktur zdaniowych w j?zyku polskim" [Main types of
sentences structures in the Polish language], Zeszyty Naukowe Wyzszej
Szkoly Pedagogicznej w Katowicach 31 (Prace j?zykoznawcze 11), 83 — 99.
1971 "Gramatyki generatywne a metoda transformacyjna" [Generative grammars
and the transformational method], in: A. M. Lewicki (ed.), Problemy skladni
polskiej [Problems of Polish syntax] (Warszawa: Panstwowe Wydawnictwo
Naukowe).
Politzer, Robert L.
1968 "An experiment in the presentation of parallel and contrasting structures",
Language Learning 18: 35—43.
Postal, Paul
1968 Aspects of phonological theory (New York: Harper and Row).
1970 "On the surface verb 'remind'", Linguistic Inquiry 1: 37 — 120.
Preston, Dennis R.
1975 "Contrastive analysis — the outlook from modern grammar", PSiCL 3:
63-72.
1979 "Introduction", in: D. R. Preston —R. W. Shuy (eds.), Varieties of American
English. A reader (Washington D.C.: International Communication
Agency), 1—13.
Puppel, Stanislaw—Jadwiga Nawrocka-Fisiak — Halina Krassowska
1977 A handbook of Polish pronunciation (Warszawa: Panstwowe Wydawnictwo
Naukowe).
Puzynina, Janina
1969 Nazwy czynnosci we wspolczesnym jgzyku polskim [Names of activities in
contemporary Polish] (Warszawa: Panstwowe Wydawnictwo Naukowe).
Quirk, Randolph
1962 The use of English (London: Longmans).
Ravem, Roar
1974 "The development of w/i-questions in first and second language learners",
in: J. C. Richards (ed.), Error analysis (London: Longman], 134 — 155.
Riley, Philip
1981 "Towards a contrastive pragmalinguistics", in: J. Fisiak (ed.), 121 — 146.
Robinson, Jane J.
1970 "Case, category, and configuration", Journal of Linguistics 6.1: 57 — 80.
Rosch, Eleanor
1973 "Natural categories", Cognitive psychology 4: 328 — 350.
Rosch, Eleanor —Β. B. Lloyd (eds.)
1978 Cognition and categorization (Hillsdale, N.J.: Lawrence Erlbaum).
Ross, John Robert
1973 "Nouniness", in: Osamu Fujimura, Three dimensions of linguistic theory
(Tokyo: TEC Corporation), 137-258.
 References 263

1975 "Clausematiness", in: Edward Keenan (ed.), Formal semantics of natural

language (London: Cambridge University Press), 422 — 475.
1981 "Nominal decay", Department of Linguistics, M.I.T.
Rubach, Jerzy
1983 "Applied contrastive linguistics: in search of a framework", Jyväskylä Cross-
Language Studies 9: 141 — 156.
Rülker, K.
1973 "Zur pragmatischen Invarianz bei der Translation" in: A. Neubert —
O. Kade (eds.), Neue Beiträge zu Grundfragen der Übersetzungswissenschaft
(Leipzig: Athenäum), 29 — 35.
Rusiecki, Jan
1976 "The development of contrastive linguistics", ISB 1: 12 — 44.
Salmon, Nicolas
1797 Grammaire angloise comparee avec la grammaire franfoise (London: W and
S Spilsbury).
Sampson, Geoffrey
1970a Towards a linguistic theory of reference [unpublished Ph.D. dissertation,
Yale University],
1970b Natural language and the paradox of the liar [ms.].
Sanders, Carol
1981 "Recent developments in contrastive analysis and their relevance to lan-
guage teaching", in: J. Fisiak (ed.), 21 —32.
Sapir, Edward
1921 Language: an introduction to the study of speech (New York: Harcourt,
Brace and World) [reprinted in 1949],
Saumjan, Sebastjan
1968 Osnovanija porozdajuscej grammatiki russkogo jazyka [Foundations of a
generative grammar of the Russian language] (Moskva: Nauka).
Schlyter, Suzanne
1980 Vagheit, Polysemie und Prototypentheorie [Ms. Universität Konstanz].
Schmitz, John Robert
1983 "Color words in English and Portuguese: a contrastive semantic analysis"
PSiCL 17: 3 7 - 4 9 .
Schwarze, Christoph
1978 "Sprachtheoretische Grundlagen der Sprachvergleichsgrammatik", in:
C. Schwarze (ed.), Kasusgrammatik und Sprachvergleich (Tübingen: Narr),
93-112.
1979 "Lessicologia contrastiva e apprendimento lessicale", in: C. Schwarze (ed.).
Italienische Sprachwissenschaft (Tübingen: Narr), 227 — 243.
Searle, John R.
1969 Speech acts: an essay in the philosophy of language (Cambridge: Cambridge
University Press).
Searle, John R. (ed.)
1971 The philosophy of language (Oxford: Oxford University Press).
Sehnert, James
1975 A contrastive gastronomic terminology in English and Polish [Ms. Uni-
wersytet im. A. Mickiewicza, Poznan],
Sehnert, James —Michael Sharwood-Smith
1974 "The verbalization of instrumentals in English", SAP 5: 37—44.
Selinker, Larry
1972 "Interlanguage", IRAL 10: 2 1 9 - 2 4 3 . [Reprinted 1974 in: J. C. Richards
(ed.), Error Analysis. London: Longman],
264 References

Sharwood-Smith, Michael
1975 "Aspects of future reference in English and Polish", PSiCL 3: 9 0 - 9 9 .
1976 "Pedagogical grammars", IS Β 1: 45 - 57.
Siewierska, Anna
1984 The passive. A comparative linguistic analysis (London —Sydney —Dover —
New Hampshire: Croom Helm).
Skorupka, Stanislaw
1967 Slownik frazeologiczny jgzyka polskiego [Phraseological dictionary of the
Polish language] (Warszawa: Wiedza Powszechna).
Snook, Roger L.
1971 "A stratificational approach to contrastive analysis", in: G. Nickel (ed.),
Contrastive linguistics, 17 — 36.
Spalatin, Leonardo
1969 "Approach to contrastive analysis", YSCECP B. Studies 1: 2 6 - 3 5 .
Stanislawski, Jan
1950—1951 Gramatyka angielska dla zaawansowanych. Szczegölowa analiza rozbieznosci
zachodzgcych migdzy gramatykg angielskg a polskg [English grammar for
advanced learners. A detailed analysis of divergencies between English and
Polish grammars], 1 —2. (Warszawa: Ksi^zka i Wiedza).
Stieber, Zdzislaw
1966 Historyczna i wspolczesna fonologia j§zyka polskiego [Historical and contem-
porary Polish phonology] (Warszawa: Panstwowe Wydawnictwo Naukowe).
Stock well, Robert P.—J. Donald Bo wen
1965 The sounds of English and Spanish (Chicago —London: University of Chi-
cago Press).
Stockwell, Robert P . - J . Donald B o w e n - J o h n W. Martin
1965 The grammatical structures of English and Spanish (Chicago — London:
University of Chicago Press).
Stockwell, Robert P.— Paul Schachter — Barbara Hall Partee
1973 The major syntactic structures of English (New York: Holt, Rinehart and
Winston).
Suomi, Kari
1980 Voicing in English and Finnish stops (Turku: Publications of the Department
of Finnish and General Linguistics of the University of Turku).
Sussex, Roland
1976 "The measurement of contrast in contrastive linguistics", PSiCL 5: 5 — 17.
Szwedek, Aleksander
1981 Word order, sentence stress and reference in English and Polish (Edmonton,
Alberta: Linguistic Research) [first published Bydgoszcz: Wyzsza Szkola
Pedagogiczna, 1975].
Tomaszczyk, Jerzy
1976 "On establishing equivalence between lexical items of two languages",
PSiCL 5: 7 7 - 8 1 .
1979 "Dictionaries: users and uses", Glottodidactica 12: 103 — 119.
Trager, George L. — Henry Lee Smith
1957 An outline of English structure (Washington D.C.: American Council of
Learned Societies).
Tran-Thi-Chau
1975 "Error analysis, contrastive analysis, and students' perception: a study of
difficulty in second-language learning", IRAL 13: 119 — 143.
Trier, Jost
1931 Der deutsche Wortschatz im Sinnbezirk des Verstandes. Die Geschichte eines
sprachlichen Feldes (Heidelberg: Winter).
 References 265

Twaddell, W. Freeman
1963 "On defining the phoneme", in: Martin Joos (ed.), Readings in linguistics 3
(New York: American Council of Learned Societies), 55 — 80.
Valdman, Albert —John S. Phillips
1975 "Pidginization, creolization and elaboration of learner systems", Studies in
Second Language Acquisition 1: 21—40.
Van Buren, Paul
1971 "Some proposals concerning contrastive analysis" [Paper read at the 2nd
Conference of Polish-English contrastive studies at Karpacz, December
16-18],
1974 "Contrastive analysis", in: J. P. B. Allen —S. P. Corder (eds.), Techniques in
applied linguistics (Edinburgh Course in Applied Linguistics 3) (Oxford:
Oxford University Press), 2 7 9 - 3 1 2 .
1976 Review of Tomasz P. Krzeszowski, Contrastive generative grammar- theo-
retical foundations, ISB 1: 2 5 0 - 3 2 9 .
Van Dijk, Teun
1972 Some aspects of text grammars: a study in theoretical linguistics and poetics
(The Hague —Paris: Mouton).
Varga, Laszlo
1975 A contrastive analysis of English and Hungarian sentence prosody (Budapest:
Linguistic Institute of the Hungarian Academy of Sciences; Washington,
D.C.: Center for Applied Linguistics).
Welna, Jerzy
1977 "Deceptive words. A study in contrastive lexicon of Polish and English",
PSiCL 7: 7 3 - 8 4 .
Widdowson, Henry
1975 "The significance of simplification", Studies in Second Language Acquisition
1: 1 1 - 2 0 .
Wierzbicka, Anna
1980 Lingua mentalis. The semantics of natural language (Sydney —New Y o r k -
London—Toronto—San Francisco: Academic Press).
Wilkins, John
1668 An essay towards a real character and a philosophical language (London:
Printed for SA: Gellibrand and for John Martyn. Printer to the Royal
Society).
Wittgenstein, Ludwig
1953 Philosophical investigations, translated by G. Ε. M. Anscombe (Oxford:
Blackwell and Mott).
Woloszyk, Alicja
1974 "Intonation of interrogative sentences in English and Polish" PSiCL 2:
145-159.
Wojcik, Tomasz
1975 Gramatyka jgzyka rosyjskiego. Studium kontrastywne [Russian grammar: A
contrastive study] (Warszawa: Panstwowe Wydawnictwo Naukowe).
Yngve, V. H.
1975 " H u m a n linguistics in face-to-face interaction", in: A. Kendon et al. (eds.),
Organisation of behaviour in face-to-face interaction (The Hague: Mouton),
47-62.
Zabrocki, Tadeusz
1976 "On the so-called 'theoretical contrastive studies'", PSiCL 4: 9 7 - 1 0 9 .
Index of subjects

Figures in italics indicate detailed treatment of a topic.

Abstract domain 215 Antonyms 81—82

Aceusativus cum infinitive 39, 205 Antonymity 79
Acronymy 73 Approximative systems 192 — 193, 196,
Action 109, 185, 197 236, 237
completed and incompleted, in Polish fossilized 235
152-159 Archaism 246
momentary 110 Arguments
Active as noun phrases 152
voice 111 Articles 16, 37, 135
sentences 246 comparison of 44 —45
Actors 142 erroneous a in front of present partici-
Address 94, 246 ple 194
Addressee 1 8 0 - 1 8 1 in congruent strings 135
Adjective 71, 75, 79, 84, 87, 109, 115, 130, in reduced registers 196
169, 173 in Spanish and English 133 — 134
derived from nouns 136 Articulation 111
, prototypical 216 of vowels 112
Adverb 71, 84, 109, 115, 143, 173 Articulatory basis 127
Adverbial phrase 66 Artificial intelligence 214
Affixation 18, 73 Artificial languages
Affricates 123, 124, 125, 247 see Language
in Polish 1 2 3 - 1 2 5 Aspect 140, 141, 145, 152, 165
see also Consonants coalescence of simple and perfect 194
Age completed vs. incompleted in Polish
see Sociolinguistic factors 152-155
Agent 110, 143, 144, 163, 164, 169, continuous 183
178-187 durative 1 5 4 - 1 5 5
in the original subconfiguration 178 iterative 155, 248
unspecified 163, 164 perfective (perfect) 1 5 4 - 1 5 5 , 183, 184
Agentive case 142 Aspiration 32, 53, 125
Agreement 197 Association in pragmatic contrastive stud-
subject-verb 249 ies 101 -103
see also Concord Auxiliaries
Aliteration 18 in congruent strings 135
Allophones 55, 56, 120, 126, 241 Auxiliary Verbs
Ambiguity 173 — 174 see Verbs
American English 76, 98, 114
American structuralism Baby-talk 196
see Structuralism Back-formation 73
Analogy 49, 195 Background framing 231
Anthropology 213, 214 Base 137, 140, 142, 149
268 Index of subjects

semantic as tertium comparationis 152, grammar

176-177 see Grammar
universal rules of 141 Cognitive linguistics
Basic sentence patterns 63, 128, 130 see Linguistics
in Spanish and English 131 — 132 Cognitive sciences 213, 214
Basque 39, 246 Collocations 87, 102
Bilingual —, more and less conventional 102
competence 161, 168, 202 Colour terms 69, 75, 82
informant 104, 148, 230-231 Commands 186, 187
Black English 190 Comment 154
Borrowings 75, 90 Communicative acts 92, 93, 97, 217
Bracketed strings 180, 181, 182 Communicative competence 91
Branching trees 180-181 Comparison 83, 131, 203, 245
Bulgarian 239 accuracy of 242
of constructions 28, 37, 38-39, 133,
Case 109, 118, 140, 247 245
accusative in Polish 224, 225, 229 of languages 245
in case grammar 142 — 144 of phonemic inventories 112
in Polish and English 135 of rules 28, 37, 39-40, 133
instrumental in Polish 224 of speech sounds 11, 122 — 127
Case grammar of systems 28, 37, 41, 42-44, 133, 245
as a framework for contrastive studies Comparison proper 11, 35, 37—45, 124,
141-146 132, 141
see also Grammar Competence 128, 147, 148, 190, 225
Categories 241 bilingual 202
classical 214 frozen 235
complex vs. simple 214 grammatical 91
discrete 214 transitional 191, 192, 235
natural 214-216 vs. performance 128, 249
Categorial component 169, 170 Complement 63, 110
Chinese 9, 38, 109, 128, 173 object 205
Classical categorization 214 Complementary distribution
Clauses 137 of plural endings 119
as complements, as objects, as objects Complication 196, 197
of prepositions, as subjects, nominal, routes 196, 202
relative 64 see also Elaboration
concessive, relative, resultative 145 Concepts
dependent and independent 137 basic, subordinate and superordinate
main 204 215
subordinate 204, 205, 245 Concord 197
Clusters subject-verb 192
of consonants 57, 58, 120, 247 see also Agreement
of pragmatic, semantic and syntactic Condensation
properties 217 of subordinate clauses 205, 245
Coda 58 Configuration 177, 183, 187, 198
Cognate object as input to derivation (conversion
see Object rules) 177
Cognitive - , initial 179, 181
 Index of subjects 269

nuclear 179 see also Contrastive studies

original 178 Contrastive Generative Grammar 3, 12,
Congruence 136, 217, 218, 247 31, 37, 169-187, 196, 197, 200, 201,
Congruent strings 202, 217, 218, 248, 249
definition 135 — 136 five postulates 171
Conjunction 109 Contrastive grammar
Connotation 77, 81, 82, 87 see Grammar
Consonants 32, 37, 55, 56, 122, 1 2 3 - 1 2 5 , Contrastive linguistics 1, 9—13, 68
126 and descriptive linguistics 10
affricates 123 "theoretical" vs. "applied" 10
fricatives 52, 53, 54, 1 2 2 - 1 2 3 vs. contrastive analysis and contrastive
laterals 1 2 2 - 1 2 3 grammar 11
nasals 122—123 see also Contrastive analysis and Con-
sibilants 49, 1 2 2 - 1 2 3 trastive studies
stops 49, 53, 122, 125 Contrastive studies 48, 140, 242, 245, 247,
Constructions 28, 147, 241 250
avoidance of identical 240 and cognitive linguistics 213 — 231
comparisons of 28, 35, 218, 245 and linguistic models 107—145
congruent 27, 204, 240 and prediction of errors 189—191
coordinative 216 autonomous 24
differences in frequencies across lan- classical (three steps) 35 — 45
guages 206 classification of 23—33
equivalent 147, 203, 204, 205, 218, corpus restricted 26, 34
219, 249 directional 23, 33, 40, 41
impersonal vs. passive 250 generalized 24, 108
infinitive as imperative in Polish 249 generative transformational 107, 108,
interrogative 37 127- 146
negative 37 grammatical 4, 140
ambiguity of in English and Polish history of 1 —3
151 lexical 6, 24, 2 9 - 3 0 , 69-90, 245
passive 246, 250, see also Passive of intonation 246
relative 37, 216 of speech sounds 120—127
semantic and communicative values of operational 24
238 paradigmatic 29
Contact 247 phonological 21, 2 9 - 3 0 , 50-63, 245,
Contrast 2, 243, 247 247, 249
categorial 65 pragmatic 32, 33, 90-105
explicated by Contrastive Generative projective (systematic) 25, 28, 34, 67,
Grammar 170 203, 204, 205, 211, 249
functional 66, 67 qualitative 26, 34
impossible 243 quantitative 26, 34, 203-212, 249
phonemic 1 2 0 - 1 2 7 structural 107, 108, 113-127, 130,
pragmatic 101 131, 132
socio-cultural 101, 103 subphonemic 11, 122, 125
structural 66 syntactic 16, 18, 20, 31, 63-68, 69,
see also Difference 101, 140, 146, 147
Contrastive analysis 11, 35, 45, 234, 235, taxonomic 24, 35, 147
236, 240, 241, 242, 243, 246 - , text-bound 25-26, 31, 34, 162, 203
270 Index of subjects

"theoretical" vs. "applied" 23, 33, 41, Difference 64, 147, 189, 237, 239, 247
48, 5 0 - 5 1 , 233-243 categorial 64, 65 — 66
traditional 107, 108-113 degree (scale) of 218, 239, 240
see also Contrastive analysis and con- functional 64, 65 — 67
trastive linguistics in form 118, 120
Conventionality scale 103 structural 64 — 65
Conversational greasers 98 see also Contrast
Co-occurrence 71, 78 Difficulty 172, 239
restrictions 148-151, 161, 163 hierarchy of 172, 189, 239
Cooperative principle 91 learning 238
Coreference 45 pattern of 246
Creole 196 scale of 246
Czech 2 0 4 - 2 0 5 , 245 Diphthongs 37, 57, 120, 126
Disambiguating context 174
Dative case 111, 142, 143 Discourse 100
Deceptive words analysis 45
see False friends definition 93
Decomposition 86, 87, 89 Distinctive features 121, 122, 125, 214
Deep structure 136, 141, 168, 196 Distinctive stress 73
as input to transformations 137 Distribution 120
as tertium comparationis 161, 245 limited 205
identical deep structure hypothesis Distributionalism
152-168 see Structuralism
in case grammar 142 —146 Diversification 169, 170, 218
in the standard theory 148—151, 152, Doubt 185
161, 168 Drills
Degree 109 in transfer of training 193 — 194
Definiteness 44 — 45 Dutch 239
Deictic words in reduced registers 196
Denial 185 Educational terms
Denotation 80, 81, 82 see Semantic fields
Derivational Elaboration 196, 197
contrast as a structural signal 115 Ellipsis
history 172, 218 in answers to questions 167
Description (as a step in contrastive stud- Embedding 69
ies) 11, 3 5 - 3 6 Emotions
Descriptive adequacy 242 pleasant and unpleasant 82
Determiner 63, 117, 132, 133, 134 English 1, 2, 9, 36, 38, 39, 40, 41, 42, 43,
in Spanish and English 133—134 44, 45, 4 7 - 1 0 5 passim, 108, 111, 112,
Devoicing 125 116, 117, 118, 119, 128, 1 2 9 - 1 3 4 ,
Dictionary 69-70, 80, 83, 90, 182, 215, 1 3 7 - 1 3 9 , 141, 143, 1 4 9 - 1 5 1 ,
247 1 6 9 - 1 8 7 passim, 189, 194, 1 9 7 - 2 0 3 ,
bilingual 79, 8 8 - 9 0 , 155 2 0 5 - 2 1 2 , 217, 220, 2 2 2 - 2 2 5 , 227,
context in 8 9 - 9 0 228, 230, 241, 242, 245, 246, 248, 249
English-Polish 225 modern standard southern 122
in Contrastive Generative Grammar phonemes 120—127
182 standard 192
—, monolingual 155 — word order 63 — 64
 Index of subjects 271

Entailment 85 Errors 8, 187, 190, 191, 201, 202, 240

Equivalence 5, 16, 21, 24 — 33, 67, 123, frequencies of 190
126, 133, 136, 167, 191, 204, 217, 218, of abundance 43, 204, 205, 206, 212
219 of avoidance 204, 206, 208, 211, 212,
contraints on 156, 162, 165, 168 240
construction 29, 34, 38-39, 149-151 overgeneralization as a source of 191
contextual 17 performance vs. competence 191
derivational-semantic 25 potential 202
empirical reality of 162 prediction of 189, 190, 191, 206, 236,
formal 25, 100 243
functional-communicative 25 translation 28
pragmatic 18, 30, 32, 34, 99, 100, Estonian 239
1 0 4 - 1 0 5 , 218, 220, 221, 245, 250 Ethnography 213
prototype theory of 222, 230 Event 142, 185, 249
rule 34, 39-40, 129 Exercises 193
semantic 16, 17, 21, 121, 131, 217 in transfer of training 193 — 194
semanto-syntactic 19, 27, 28, 29, 30, remedial 190
34, 6 7 - 6 8 , 104, 147-168, 169, 170, translational 2
177, 218, 245, 248, 250 Explicitness 147
substantive 32, 34
statistical 26 - 28, 34, 203
system 29, 34, 38, 211 Facilitation 189
textual 147 Factitive case 142
translational 6, 17, 18, 28, 34, 67, 131, False friends 76
147, 245 Family resemblance 87, 88, 214, 216, 223,
Equivalents 71, 148, 165, 176, 202, 203, 227
228, 230, 241 Features 245
congruent 204 articulatory 111
phonemic 54, 57 contextual 84, 186
lexial 74, 76, 79, 80, 81, 8 5 - 9 0 , 155, decomposition into 86
211, 223 distinctive 121, 122, 125, 214
matching of in traditional contrastive inherent 84
studies 110 phonological 50, 214
pragmatic 98, 9 9 - 1 0 0 , 218, 220, 250 relational 84, 85
prototypical 222, 225 — 230 relevant and redundant 120, 121 — 127
semanto-syntactic 162, 165, 168, 205 selection 7
situation 99 semantic 84, 85, 8 6 - 8 8
socio-cultural 101 syntactic 84, 85
statistical 165 Fijan 38
structural description of 171 Finnish 77, 1 5 9 - 1 6 5 , 194, 239
textual 89 passive sentences in 159 — 165
translation 18, 89, 100, 160, 165, 166, Fixed expressions 71
203 Flap 126
Equivalent sentences Focus 69, 154, 187
see Sentences Foreign language 239, 240
Error analysis 189, 190, 201, 202 acquisition 236
annihilation of 236 learning 8, 236, 237, 243
— vs. contrastive studies 190 — 191 - teaching 2, 13, 233, 242
272 Index of subjects

Foreign learners 187, 189, 190, 193, 235, "Perfect" tenses in comparison with
238, 239, 249 English 111
Foreigner talk 196 transfer from 195
Formal correspondence 16, 17, 148 Gerund 205, 245
Formality 93 regular 207, 208, 209
Formulaic expressions 98 — 99, 101, 103, Gestalts 215, 221, 228, 230, 250
104-105 Glides 247
Fossilization 193 Global rules
Free variants 121 see Rules
French 38, 39, 40, 76, 9 4 - 9 6 , 110-111, Glottalization 125
112, 164, 189, 2 0 4 - 2 0 5 Government 197
vowels compared with English 112 Gradience 216
Fricatives 122, 124, 247 Grammar 48, 70, 83, 84, 194, 202, 216
in English 122 as a diversifying device 169
in Polish 1 2 2 - 1 2 3 case 141
Friction 54 cognitive 47
Fronting 125 comparative 247
Function words 9, 18, 69, 109, 115, 194, contrastive 3, 26, 146
197, 246 conversion 177
as a structural signal 115 core 217
descriptive 108
auxiliaries, coordinators, interrogators,
foreign language 107
prepositions, qualifiers 115
fuzzy 213
in congruent strings 135
generative 3, 127, 171, 201, 214, 217,
in reduced registers 196
235
noun determiners 115, 116
generative format of contrastive stud-
omission of 194, 201
ies 235
Future 249
relational 35, 107
Fuzzy grammar
stratificational 107
see Grammar
text 45, 128, 174
Fuzzy phenomena 109
vs. sentence grammar 173
boundaries of categories 231 traditional 107
boundaries of phonemes 216 transformational generative 35, 47, 71,
127-146, 213, 217, 247
Gender 43, 133, 140 universal 24, 68, 108, 109
in Polish and English 135 Grammatical rules
in Spanish and English 133 see Rules
virile vs. non-virile 42 Graphs 177
Generative grammar Greek 38, 39
see Grammar scholars 108
Generative phonology
see phonology Hearer 94, 128
Generative semantics 47, 141, 213 Historical linguistics 9
Generative transformational models 107, Homonymy 79, 80
147 Hopi 109
Georgian 247 Hungarian 194, 212
German 38, 39, 65, 76, 111, 116, 117, Hyperonymy 84
118-119, 143, 164, 217, 239 Hyponymy 84, 85
 Index of subjects 273

Icelandic 240 Knowledge

Idealized Cognitive Models 215, 248 encyclopedic vs. linguistic 85, 90
Idioms 73 Korean 165 — 166
Illocutionary
acts 93 Labialization 53
force 97 Landmark 222, 228, 229
structure 93 — 96 Language 3, 217, 235, 243
Image schema 222, 228, 229 artificial 128
Immediate constituents 144 as a hierarchical structure 15, 31
Imperatives 249 aspects of in comparing 245
see also Sentences behaviorist model of 237
Indefiniteness 44 categories of variation 246
Infinitive 205, 249 communicative function of 48
—, marked in English 205 etalon 241
Inflection 9, 18, 69, 73, 154, 197, 246 genetically close and remote 239
as a noun marker 109 modular view of 91
as a source of errors 8 native 191, 240
as a structural signal 115 second, learning, hierarchies of diffi-
plural in German and English culties 172
118-119 source 107, 189, 190, 201, 202, 236
Input structure 169, 171, 172, 187, 248 patterns 195
see also Semantic structure transfer from 195
Instrumental target 90, 107, 187, 189, 190, 191, 193,
case 142, 143 196, 201, 202, 204, 206, 212, 235, 236
inflection in Polish 149 — 151 deviations from the norms of
Interference 498, 195 201-202, 242
Interjections 109 strategy of learning in forming in-
Interlanguage 189, 193, 1 9 6 - 1 9 7 , terlanguage 194
2 0 0 - 2 0 2 , 206, 212, 2 3 4 - 2 3 7 Langue 25, 127
Polish-English 211 Laterals 37
Interlude 58 in English 122
Intonation 154, 246 in Polish 123, 124, 125
cognitive function 63 Latin 1, 2, 39, 40
rising in occasional questions 249 Lexical
Italian 40, 76, 143, 170 congruity 78 — 79
entry 70, 84, 90, 155, 173
Japanese 39, 98 insertion 140, 170, 197, 198, 200, 202
Juxtaposition 11, 35, 36-37, 60, 89, item 69, 71, 72, 75, 77, 79-90, 148,
116-117, 119-120, 122-124, 157, 182, 197, 198, 201, 202, 246
1 3 2 - 1 3 3 , 141, 155 equivalence of 85 — 90, 155,
of basic sentence patterns 131 — 132 158-159
of phonemes 122—123 representation 148
of lexical items 89, 117, 155 Lexicalist hypothesis 247
Lexicology 15, 70, 71, 83, 84
Kernel sentences Lexicon 47, 48, 91, 112, 137, 140, 154, 155.
see Sentences 246
Keys 82 Linguistic models 107
Kinship terms 69, 75, 82 —, choice for contrastive analysis 241, 242
274 Index of subjects

generative transformational 107, 147 Modality 75, 221

relational 107 in case g r a m m a r 141, 143 — 144
stratificational 107 - , sentence 142, 145, 182, 247
structural 107 Modifier 63, 110
Linguistic universals 9, 68, 140, 236, 239, M o o d 109, 182
240 M o r p h e m e 70, 83, 214, 245
Linguistics 91, 214, 234 negative 184
applied 234 pre-sentence Q 1 3 7 - 1 3 9 , 198,
broader and narrower sense 91 200-201
"code" vs. " h u m a n " 91 tense 198
cognitive 213 — 231 well-formed and ill-formed combina-
theoretical 234 tions of 83
typological 9 - 1 0 , 68 M o r p h o l o g y 70, 112, 216
vs. contrastive analysis 241 nominal 246
Listening 239 —, simplified in reduced registers 196
Lists 84 Morphophonological nets 246
Locative case 142
Locus 1 7 9 - 1 8 7 Names 143
in the original configuration 178 Narrowing 247
Lover's talk 196 Nasals 37, 122
in English 1 2 2 - 1 2 5
Macedonian 239 in Polish 1 2 2 - 1 2 5
Manner 1 8 0 - 1 8 7 Native language
as a universal semantic category 180 see Language
Markedness 82, 1 5 6 - 1 5 7 N a v a h o 248
M a r k i n g convention 157 Necessary and sufficient conditions 214
Marking polarity 156 — 159 Negation 145, 154, 184
conflicting 1 5 7 - 1 5 9 pure vs. modal 185, 186
Meaning 132, 148 Negative constructions
figurative 217 see Constructions
in establishing equivalence 117, 119, Nomina
166 agentis and loci 75
in the deep structure 137 actionis 75, 208
in linguistic analysis 113 see also Nominals
structural vs. lexical in American Nominals 206
structuralism 247 action and gerundive 206 — 211
vs. form 132 average frequency of occurrence in
Melanesian 38 English and Polish 210, 2 4 9 - 2 5 0
Mental scheme 215 Nominative 130
Metalanguage 172 Nominativus cum infmitivo 205
M e t a p h o r 217 N o o t k a 248
Metaphorical concepts 63 Norwegian 239
Metaphorical extension 217, 223 N o u n 71, 72, 84, 109, 111, 115, 116, 130,
of spatial prepositions 227 134, 145, 154, 173, 217, 248
Metaphorization 84, 246 as a universal category 116
Military ranks 82 full 207, 211
M o d a l s 173 prototypical 216
see also Verbs —, structural definition 116
 Index of subjects 275

verbal vs. regular 209 in Finnish 164

Nouniness 2 0 8 - 2 1 0 see also Constructions and Sentences
Noun phrase 63, 65, 66, 84, 130, 132, 138, Path 223, 228, 229
141, 144, 145, 169, 170, 173, 174, 175, Patient 163, 164, 169, 179-187
198 in the original configuration 178, 179
Nucleus 58 Pattern matching 229, 230
Number 109, 111, 132, 133, 140 complete (full) 228, 230
in Spanish and English 133 partial 2 1 7 - 2 2 1 , 241
of nouns 109, 118-119, 132, 194 Pedagogy 234
Numerals 38 Perfekt
Nursery rhymes in German 164
translation of 250 Performance 128, 147, 190, 243
in interlanguage 201, 202
Object 63, 109, 116, 128, 198 native 206, 212
cognate 66 non-native 203, 206, 211, 212
direct 66, 187 translation as 147
in case grammar 142 vs. competence 128, 249
indirect 66 Performative 186
, prepositional 111, 116 complex 185, 187
Objective case 142, 143, 144 negative 185
Obstruction 247 verb 184, 185
Obstruents 141, 142 Perlocutionary effect 2 1 8 - 2 2 1
Occlusion 32, 124, 247 Persian 165
Onset 58 Person 109, 135
Overgeneralization 211 in Polish and English 135
as a source of errors 191 Philosophy 213, 214
in forming interlanguage 193 Phoneme 70, 113, 126, 127, 214, 241, 245
as a natural category 216
Palatalization 53 comparison of 52 — 57
Paralinguistic equivalent 52 — 57, 242
expressions definitions 120—122
body movement, gestures, grunts, variants of 113
features, mimicry, noises 92 — 93 Phonemic inventories 112, 241
features 246 comparisons of 112
Paraphrase 84, 159, 162 Phonemic overlap 126, 216
Parody 203, 249 Phonetic representation 50
Parole 25, 127 Phonetic similarity
Parsing 109-110 see Similarity
Participant states 93 Phonetic substance 220
Participle 205, 245 Phonological interface 29
past 38 Phonological representation 50
active 204 Phonological rules
present 194 see Rules
Particle 109, 222, 223, 224 Phonology 32, 48, 70, 91, 114, 214, 242
Parts of speech 109, 114-116 classical 216
Passe compose in French 164 generative 121, 126, 127
Passive 3 8 - 3 9 , 110, 111, 129, 137, generative vs. taxonomic 127, 241, 242
159-160, 163, 165, 194, 216, 246, 250 —, segments in 121, 247
276 Index of subjects

Phrase 70, 71, 75 Pronoun 29, 37, 84, 109

Phrase structure rules comparison of 42—43
see Rules definite and indefinite 44
Phraseological frequency in texts 203
collocations 72, 73, 77 in reduced registers 196
fusions 73, 77 personal 16, 28, 41 - 4 3 , 65, 176, 203
units 7 8 - 7 9 , 83, 246 Pronunciation 75
Phraseology 70 Preposition 187
Pidgin 196 in case grammar 141 — 144
Place Prototypes 241
as a universal semantic category Prototypical effects 216
180-187 Psychological reality 237
Plural 111 Psycholinguistics 11, 237
inflection in German and English 118 Psychology 213, 214, 234
of regular gerunds 209, 210 vs. contrastive analysis 237
Polish 36, 3 8 - 4 5 , 4 7 - 1 0 5 passim, 108,
119, 1 2 9 - 1 3 1 , 1 3 7 - 1 3 9 , 1 4 9 - 1 5 1 , Quantifiers 38
1 6 9 - 1 9 4 passim, 195, 2 0 3 - 2 1 2 , 220, Question 49, 69, 137, 145, 154
222-226, 229-230, 240-242, ambiguity of in English and Polish
245-249 150-151
interlanguage of English 202 answers to negative in Korean
phonemes 122, 127 166-167
word order 63 — 64 dependent and independent 139
Politeness 91, 221 echo 137
Polyglots 240 general 187
Polysemy 79-80, 87, 88 in Polish and English 137—140
Possessive attribute 66 intonation of 61 —62
Pragmalinguistics 92 negative in Persian 165
Pragmatics 18, 45, 48, 85, 97, 99, 112 occasional 200, 249
contrastive 91, 92 positive 186, 187
definitions 90-92 rhetorical as a speech act 218
general vs. sociopragmatics 91 tag 137
"small" facts vs. "large" facts 97, 100, wh- 97, 137, 187, 197-201, 249
101 Question word
Prague School 121 czy in Polish, optional omission of 138
Predicate 63, 110, 169, 170, 181, 248 ij]whether in English 139
calculus 152
phrase 144, 145 Reading 239
Prefix 77, 2 2 2 - 2 2 6 Received
"Premature" lexicalization 199-200, 201, Pronunciation 122
202, 249 Standard 122
Preposition 84, 109, 164, 173, 2 2 2 - 2 2 4 , Receptive skills 239
226 see also Listening and Reading
prototypical 216 Reduced
with as an equivalent of instrumental codes 200, 201
inflection 1 4 9 - 1 5 1 registers 172, 196
Prepositional phrase 65, 204 Reductionism 214
Presupposition 69, 84, 152, 154, 185 Reference 84, 174, 177, 248
 Index of subjects 277

and coreference 174, 176—177 passivization 37

composite character of 174 phonological in Contrastive Genera-
Referential index 178, 180-187 tive Grammar 170
Referentials 175-176 phrase structure 47, 128 — 136, 141,
Referents 175, 249 145
Regionalisms 246 post-lexical 177
Relational grammar projection 140, 144, 149
see Grammar semantic categorical 177, 179
Request 2 2 0 - 2 2 1 subcategorization 72
Resident 179-187 subject-raising 37, 39
in the original configuration 178, 179 syntactic categorical 177, 180, 182,
Retraction 125 198, 202
Rheme 110 syntactic transformational 177
Rhyme 18, 250 transformational 127, 128, 129, 134,
Rhythm 18, 59, 135,137,148
see also Suprasegmental phenomena wh- formation 198-199
Roles 7, 48, 49, 169, 170, 178, 198, 204 word formation 247
complex 180, 183 Russian 190, 240
coreferential 187
elementary 179, 186 Samir 38
in case grammar 142 Sanskrit 53
in Contrastive Generative Grammar Saxon genitive 246
169-187 Second language learning
nuclear 201 see Language
social 93 Segment 50
transfer of 211 in phonology 121, 247
Roman scholars 108 Selection
Rule 7, 28, 31, 68, 83, 84, 147, 169, features 7
171 - 1 8 7 , 192, 193, 198, 201, 216, 217, restrictions 72, 84, 148-151, 160, 161,
218, 245, 247 163
adjective placement 37 Self-embedding 177, 178
aspect 154 Semantic
comparisons of 28, 35, 245 anomalies 72
complication 196 fields 79
congruent 40, 135 being at a place, colors, educa-
conversion 177—178 tional terms, emotions, gastro-
equivalent 40, 135 nomical terms, kinship terms,
exceptions to, in traditional grammars leaving, military ranks, sports and
109 games, vehicles 82
identical and different 172 interface 29
in generative grammars 127, 128 — 146 interpretation 148, 248
input to 171-172 markers 72, 86
interrogative inversion 37 relations 73, 84
lexical insertion 177 horizontal and vertical 83, 84
morphographemic/morphophonemic representation 4, 6, 31, 32, 40, 152,
177 161, 169, 173, 182, 184, 197, 218
ordering 172, 200 as nuclear subconfigurations 182
, output from 171—172 — of wA-questions 198
278 Index of subjects

structure 146, 152, 159, 163, 164, 165, positive 184

168, 169 production of 249
as input to transformations 146, prototypical 216
152 simple 128, 135, 205
identical in equivalent sentences stress 45
155, 168, 169, 1 7 6 - 1 7 7 superordinate 184
Semantics 6, 85, 91, 92, 118, 136, 152 various stages of derivation 201
and deep structure 152 Sentoid 145
definition 92 Serbo-Croatian 194, 241
interpretative vs. generative 152 Set
in the standard theory 137 intersection 219
sentence vs. word 18, 47—48, 69, theory 214
1 5 4 - 1 5 5 , 159, 162 Sex see Sociolinguistic factors
vs. pragmatics 85, 92 Shortening
Semanto-syntactic gaps 164 in word formation 73
Semanto-syntax 47, 100 Sibilants 122, 124
Semivowels 55 in English 122
Sense 223, 226, 230 in Polish 1 2 2 - 1 2 3
metaphorical 217, 222, 227 Signs 214
prototypical 223, 225, 229 simple see Morphemes
spatial 227, 228 Similarity 189, 247
Sentence 69, 70, 128, 130, 137, 145, 207, as tertium comparationis 141
210, 214 degree of 172, 2 1 8 - 2 2 1 , 230
active 1 5 9 - 1 6 0 , 163 formal 131
affirmative 93, 128, 135 index 171
ambiguous in Finnish 159—160, 171, in lexical studies 75 — 77
185 of usage 131
as output of generative grammar 169 phonetic 52 — 53, 57
average length of, in English and Po- pragmatic 221
lish 207 structural 131
complex 129, 135, 146, 182 syntactic 218
compound 129, 135, 146, 182 Simplification 194, 196
congruent 135 strategy 195
connection 154 Simplified codes 196, 197
embedding 154 Situation 93
enumeration of, in generative gram- Situational setting
mars 128 see Tertium comparationis
equivalent 6, 31, 65, 66, 136, 147, 148, Slovak 170, 239
1 4 9 - 1 5 1 , 157, 169-187, 218-222, Social conditions
246 see Sociolinguistic factors
hierarchical structure of 109—110 Sociolinguistics 11
imperative 69, 142 Sociolinguistic factors
interpersonal function of 18 age, sex, temperament, social condi-
interrogative 69, 129, 137, 142, 154 tions, verbal skills 104
intonation of 61 Spanish 36, 53, 54, 65, 76, 77, 1 3 1 - 1 3 4 ,
kernel 128, 129, 135 143, 170, 217
negative 129, 137, 185, 195 Speaker 94, 128, 181, 185
- , passive 129, 137, 1 5 9 - 1 6 0 , 163, 165 Specialization 76
 Index of subjects 279

Speech act 92, 99, 100, 180, 181, 218, 221 -anie, -enie, -cie in Polish 208 — 211
Spelling 75 -ing in English 207—211
Sports and games Sumerian 39
see Semantic fields Suppletion 7
Standard theory 7, 48, 136-137, 140, 144, Suprasegmentals
148, 152, 161, 180, 213, 218 see Suprasegmental phenomena
and contrastive studies 247 Suprasegmental phenomena
Statement 69 as a structural signal 115
as a speech act 218 intonation 59, 60-63, 69
State pauses 60
in the semantic structure of passive rhythm 59
sentences 164 stress 59-60
Stimulus Surface structures 4, 8, 127, 137, 138,
in stimulus-response relation 237 1 4 1 - 1 4 6 , 148, 185, 245
Stops 122, 124, 125, 247 Swahili 143
in English 122, 125 Swedish 9 4 - 9 6 , 239
in Polish 1 2 2 - 1 2 3 , 125 Syllable 37, 58
Strategies of communication Symbolic logic 152
in forming interlanguage 194 Synonymity 79, 84, 85
Stratificational grammar of constructions 149
see Grammar Synonyms 80 — 82
Stress 240, 247 Syntactic
congruity 78
heavy in occasional questions 249
frames 173
in compound nouns 136
markers 246
special 187
primes 146
see also Sentence
names, verboids 143 — 144
Structural
structure 181, 201, 245
description 171
and aspect of verbs 154
models 107
as output of transformations 146
signals 115-117
Syntagm 245
Structuralism 36, 48
Syntax 68, 70, 73, 85, 91, 112, 216
American 4, 113 — 127
Systems 28, 147
vs. cognitive linguistics 213
comparison of 28, 35, 245
Style 203
of phonemes 120
individual 212
in quantitative contrastive studies 203 Tautology 87
Subconfiguration 178 — 187 Telegraphese 196
nuclear 181 - 1 8 5 , 1 9 7 - 2 0 2 Temperament
Subject 63, 65, 66, 109, 110, 116, 128, 249 see Sociolinguistic factors
extraposed 206 Tense 69, 109, 138, 142, 152, 153, 173, 182,
in case grammar 142 198, 200
intransitive 246 future 182, 249
of the passive 111 "dangling" 1 3 8 - 1 3 9
Substantia past 69, 144, 182, 192, 198, 249
deverbalia 208-211 perfective forms of 164, 165
verbalia 208 - 211 present 182, 198, 249
Sufficient and necessary conditions 88 present continuous 110, 194
Suffix 117 —, present perfect 165
280 Index of subjects

Tertium comparationis 4, 15 — 21, 24 — 33, agreement 49

37, 40, 60, 69, 8 6 - 8 8 , 1 0 0 - 1 0 1 , 110, and the meaning in the standard the-
113, 1 1 6 - 1 1 7 , 119, 126, 1 3 0 - 1 3 2 , ory 137
1 4 0 - 1 4 1 , 221, 250 "cosmetic" (post-lexical) 173, 197, 198,
acoustic 52, 124 249
articulatory 5 2 - 5 3 , 123 do-support 49, 138
auditory 52, 124 impersonal 164, 170
deep structure as 161 interrogative 137 — 139
formal and semantic 16, 20 — 21, 37,
lexical insertion 83
132, 146
passive in English 170
immediately relevant 31 — 33, 123 — 125
syntactic, in Contrastive Generative
phonetic similarity as 52
Grammar 170, 197, 198
semantic features as 86
Translation 1 7 - 1 8 , 26, 28, 30, 147, 204,
situational setting as 100—101
241, 248, 250
substantive 2 9 - 3 0 , 123, 126, 245
as primary data for quantitative con-
ultimately relevant 32 — 33, 123 — 124
trastive studies 203
universal base as 127
Text 173, 1 7 4 - 1 7 5 , 177, 190, 203, 205, as performance 161, 168
211, 220 erroneous 30, 161 — 162
as a translation 203 of legal texts 221
as object of parody 249 of nursery rhymes 250
contrastive study of 206 statistical data 245
in interlanguage 212 vs. equivalence 18 — 19, 147, 161, 164,
Text grammar 165, 167, 171
see Grammar word-for-word 19, 20, 27, 162, 163,
Theme 110 165, 246, 250
Theories Transforms 135
choice of for contrastive analysis 240 Tree
expert and folk 215 as representation of structure 181
linguistic 233, 2 3 6 - 2 3 7 , 243 Trill 126
Time 145, 1 8 0 - 1 8 7 Truth conditions 92
as a universal semantic category 180 2-text 25-26, 28, 30, 34, 203
Topic 154 Typological linguistics
comment relations 69, 154 see Linguistics
Traditional grammars
see Grammars
Underlying structure 4, 24, 127, 160, 161,
Trajector 222, 2 2 8 - 2 2 9
165, 170, 187, 201
Transfer 212, 237
Universal
lexical 211
concepts 112
negative 2, 97, 172, 189, 210, 242
principles and parameters 247
of training, in forming interlanguage
Universal base
193-194
positive 172, 189, 204 as input to Contrastive Generative
unpredictability of, by contrastive Grammar 172, 177
studies 212 as tertium comparationis 127
Transformational generative grammar hypothesis 6 - 7 , 144, 146
see Grammar Universal grammar
Transformation 47, 128, 134, 136, 145, 146 see Grammar
 Index of subjects 281

Variants prototypical 216

of phonemes Stative 164
prototypical vs. non-prototypical transitive 38, 84
216 two-word 77
free, of Polish nadjponad 225 Verb phrase 63, 66, 130, 132, 173
Variphones 120 as complement 151
see also Free variants as predicate 151
Vehicles Verbal particles 77
see Semantic fields Verbal skills
Verb 2, 37, 38, 71, 72, 75, 84, 109, 115, see Sociolinguistic factors
141, 142, 144, 145, 150, 155, 163, 173, Verboids 143
192, 198, 200, 208, 217, 248, 249 Vocal tract 32, 247
auxiliary 3 8 - 3 9 , 173 Voicing 32, 125
auxiliary, be 38, 145 Vowels 37, 54, 55, 126, 242, 247
auxiliary, do 249 stressed and unstressed 216
causative 205
copula, absence of, in reduced registers Word 7 0 - 7 1 , 7 3 - 9 0 , 205, 210, 211, 248
196 complex 74 — 75
defective in Polish 157 composition of 73
equivalent in Polish and English compound 71, 74 — 75, 78
156-159 content 79
finite 128, 131 definition of 70 - 71, 248
imperfect 156 formation 73 — 77, 83
inflection of 65 generator 83
intransitive 84, 160 lexical 135
in Finnish 160 order 9, 10, 4 4 - 4 5 , 63, 69, 115, 154
linking 134 as a structural signal 115
main 65, 138 in reduced registers 196
modal 65, 138, 142, 247 , transfer from Polish 195
non-finite 205 polymorphic 73
of communication 131 polysemous 80
perceptive 205
, perfect 156 Yiddish 98
Index of authors

Aarts, Flor 75, 251 Cordcr, S. Pit 191, 192, 196, 234, 235, 236,
Aclfric 1 251, 253, 265
Agard. Frederick B. 54, 59, 107, 251 Cygan, Jan 248, 253
A k h m a n o v a , Olga S. 261
Danilewicz, Tadeusz 41, 253
Allen, J. P. B. 251, 265
De C a m p , David 196, 253
Anderson, J o h n M. 247, 251
De Geest, W. 107, 253
Anderson, Stephen R. 125, 251
Delattrc, Pierre 54, 253
Apresjan. Jurij Y. 48, 113, 248, 251
De Saussure, Ferdinand 48
Arabski, Janusz 190, 251 Dillon, George L. 217, 253
Arnold. I. W. 70, 74, 78, 82, 252 Dingwall, William O. 11, 129, 130, 135,
Awcdyk. Wieslaw 51, 58, 59, 252 253
Dinneen, Francis P. 246, 254
Bach, E m m o n 6, 252, 254, 260 Di Pietro, Robert J. 1, 23, 24, 54, 59, 86,
Banathy, Bela H. 190. 252 107, 108, 141, 143, 144, 145, 146, 152.
Bartsch, Renate 47, 154, 252 169, 170, 172, 218, 233, 251, 254
Baudouin de Courtenay, Jan 121 Dluska, M a r i a 60, 254
Becka. J. V. 30, 204, 245, 252 Duczmal, Stanislaw 75, 82, 254
Berault, Peter 112, 252 D u s k o v a , Libusa 205, 254
Biedrzycki, Leszek 55, 59, 252
Eliasson, Stig 255
Bloomficld, Leonard 113, 252
Enquist, Niels Erik 190, 254
Boas, H a n s Ulrich 141, 252
Bogacki. Krzysztof 210 Ferguson, Charles A. 196, 254
Bölingen Dwight 17. 72, 86, 252 Filipovic, Rudolf 241, 254
Borges, Jorge Luis 248 Fillmore, Charles 6, 7, 90, 91, 97, 98, 99,
Borkowski. G . 129, 252 100, 101, 103, 104, 141. 142. 143, 144,
Bouton, Lawrence F. 1. 152, 153, 155, 159, 152, 213, 246, 254
160, 161, 163, 165, 167, 168, 252 Fisiak, Jacek 3, 10, 11, 15, 23, 41, 43, 50,
Bowen, J. D o n a l d 54, 59, 120, 238, 264 51, 59, 60, 61, 64, 65, 67, 84, 107, 108,
Boyd, Julian 249, 252 113, 121, 126, 130, 137, 140, 233, 234,
Briere, Eugene J. 189, 252 236, 246, 254, 255, 258, 262, 263
Bross, J. S. 129, 252 F o d o r , Jay A. 255
Brown, Roger 197, 200, 201, 249, 252 Francis, W. Nelson 59, 109, 115. 116, 121,
Brugman. Claudia 213, 222, 223, 252 247, 255
Burgschmidt. Ε. 253 Fräser, Bruce 250, 255
Fries, Charles C. 2, 113, 114, 117, 255

C a t f o r d , John C. 6, 147, 233, 253 G a z d a r , Gerald 92, 255

Chesterman, Andrew 1, 202, 249, 253 G i m s o n , A. C. 56, 58, 62, 247, 255
C h o m s k y , N o a m 7, 47, 48, 68, 71, 72, 91, Ginzburg, R. S. 78, 255
126, 128, 129, 130, 135, 136, 137, 141, Gleason, Henry A. Jr. 59, 82, 121, 255
143, 148, 151, 168. 171, 213, 247, 253 G n i a d e k , Stanislaw 107, 255
Coles, Elisha 2, 253 Goldby, H a r r y 112, 257
Comrie, Bernard 1, 253 G ö t z , D. 253
284 Index of authors

Graczyk, Ireneusz 206, 255 Keenan, Edward 263

Grzegorczykowa, Renata 250, 255 Kendon, A. 265
Gussmann, Edmund 50, 51, 58, 242, 255 Khidekel, S. S. 255
Kielski, Bolestaw 107, 257
Halle, Morris 83, 125, 126, 247, 253, 255, Knyazeva, G. J. 255
256 Köhler, Klaus 127, 257
Halliday, Michael A. K. 6, 17, 18, 19, 35, Kohn, Kurt 257
256 Kolbert, Jack 112, 257
Hall-Partee, Barbara 264 Konderski, Stefan 141, 257
Harms, Robert T. 252, 254, 260 König, Ekkehard 141, 257
Harris, Zcllig 113, 114, 256 Kopczyhski, Andrzcj 18, 54, 125, 257
Hewes, John 2, 256 Koutsoudas, Andreas 172, 257
Hill, L. A. 77, 256 Kramsky, Jiri 247, 249, 258
Hiz, Henry 176, 256 Krassowska, Halina 262; see also Mack-
Hjelmslev, Louis 18, 29, 38, 256 iewicz-Krassowska, Halina
Hockett, Charles 38, 248, 256 Kryk, Barbara 41, 258
Holmes, J. 256 Krzeszowski, Tomasz P. 3, 7, 9, 18, 20, 27,
Howel(l), James 2, 256 38, 54, 76, 77, 95, 107, 120, 121, 125,
Hudson, Richard A. 47, 256 135, 140, 141, 152, 161, 175, 176, 177,
Hymes, Dell 48, 91, 256 184, 187, 198, 218, 245, 246, 247, 248,
249, 250, 258, 265
Ivir, Vladimir 16, 17, 256 Kudzinowski, Czeslaw 164, 258
Kufner, Herbert L. 107, 116, 119, 120, 258
Jacobovits, Leon A. 259 Kühlwein, Wolfgang 12, 24, 25, 258
Jacobs, Roderick A. 255 Kurytowicz, Jerzy 249, 258
Jaeger, Jeri 216, 256
Lado, Robert 16, 52, 57, 60, 75, 76, 189,
Jakobson, Roman 121, 125, 256
190, 239, 246, 259
James, Carl 1, 11, 15, 18, 19, 91, 190, 202,
Lakoff, George 47, 63, 88, 109, 148, 149,
233, 234, 235, 236, 237, 238, 239, 241,
151, 213, 214, 215, 216, 217, 221, 231,
242, 245, 256
248, 250, 259
Janicki, Karol 32, 104, 140, 220, 256
Lakoff, Robin 1, 259
Jassem, Wiktor 52, 55, 56, 58, 120, 121,
Lamminmäki, Reijo 52, 259
125, 247, 256
Langacker, Ronald 47, 85, 109, 213, 215,
Jerome, K. Jerome 245
223, 259
Johnson, Mark 63, 217, 259
Langendoen, D. Terence 143, 259
Jones, Daniel 121, 257 Lawendowski, Boguslaw 82, 259
Joos, Martin 265 Lee, William R. 233, 259
Leech, Geoffrey 69, 91, 97, 259
Kade, O. 263 Lees, Robert B. 207, 208, 209, 210, 211,
Kakietek, Piotr 40, 41, 257 250, 259
Kalisz, Roman 75, 82, 217, 218, 219, 250, Lewandowska-Tomaszczyk, Barbara 40,
257 249, 250, 259
Kalogjera, Damir 41, 257 Lewicki, Andrzej Maria 262
Karcewski, Sergej 121 Lewis, Mark 2
Karttunen, Kimmo 40, 257 Lindner, Susan 213, 259
Karttunen, Lauri 92 Lipinska-Grzegorek, Maria 3, 7, 15, 16,
Kastovsky, Dieter 257, 258 23, 107, 130, 255, 260
Katz, Jerrold J. 148, 149, 159, 255, 257 Liston, Jerry L. 16, 260
 Index of authors 285

Lloyd, Β. Β. 215, 262 Pickbourn, James 2, 262

Ludkiewicz, Anna 246, 260 Pieper, Ursula 258
Ludskanov, Alexander 248, 260 Pike, Kenneth L. 121, 262
Lyons, John 70, 84, 210, 246, 256, 260 Poe, Edgar Allan 220
Polanski, Kazimierz 129, 262
Mackiewicz-Krassowska, Halina 59, 60, Politzer, Robert L. 16, 189, 262
61, 260 Postal, Paul 47, 56, 148, 149, 159, 257, 262
Madarasz, P. H. 190, 252 Preston, Dennis R. 7, 8, 68, 104, 246, 262
Mansion, J. A. 110, 260 Pride, J. B. 256
Manczak, Witold 89, 260 Puppcl, Stanislaw 55, 262
Marek, Boguslaw 59, 61, 62, 260 Puzynina, Janina 250, 262
Markkanen, Raija 41, 260
Martin, J. W. 238, 264 Quirk, Randolph 81, 262
Marton, Waldemar 40, 129, 135, 136, 218,
233, 247, 260 Ravem, Roar 200, 201, 202, 262
Matisoff, James A. 98, 260 Richards, Jack C. 253, 261, 262, 263
McCawley, James D. 47, 72, 84, 152, 213, Riley, Philip 92, 93, 94, 95, 96, 97, 99, 100,
260 101, 262
Mcintosh, Angus 256 Robinson, Jane J. 141, 262
Meech, S. B. 3, 260 Rosch, Eleanor 215, 262
Mel'chuk, Igor A. 241, 260 Rosenbaum, Peter S. 255
Mickiewicz, Adam 250
Ross, John Robert 209, 213, 216, 262
Micklesen, L. 129, 252
Rubach, Jerzy 127, 263
Miege, Guy 112, 261
Rülker, Κ. 18, 263
Mieszek, Aleksandra 59, 61, 62, 63, 261
Rusiecki, Jan 245, 263
Milewski, Tadeusz 39, 261
Montague, Richard 92, 261
Sajaavara, Kari 254
Morgan, Bayard Q. I l l , 261
Salmon, Nicolas 247, 263
Morris, Charles W. 48, 261
Sampson, Geoffrey 175, 248, 263
Mostowska, Katarzyna 246
Moulton, William G. 54, 261 Sanders, Carol 233, 263
Sankin, Α. Α. 255
Sapir, Edward 113, 114, 263
Nawrocka-Fisiak, Jadwiga 262
Saumyan, Sebastjan 83, 263
Nemser, William 192, 193, 235, 261
Schachter, Paul 264
Neubert, Albert 263
Schlyter, Suzanne 217, 263
Newmeyer, Frederick J. 213, 261
Schmitz, John Robert 75, 82, 263
Nickel, Gerhard 107, 233, 238, 257, 258,
Schwarze, Christoph 12, 76, 263
261, 264
Searle, John 48, 263
Nilsen, Don Lee Fred 141, 247, 261
Sehncrt, James 75, 82, 263
Nowakowski, Miroslaw 71, 83, 84, 155,
Selinker, Larry 193, 194, 195, 235, 263
246, 261
Sharwood-Smith, Michael 41, 75, 233, 263.
Oleksy, Wiestaw 40, 99, 100. 102, 258, 261 264
Olsson, M. 190, 261 Shibatani, Masayoshi 253
Ozga. Janina 59, 60, 127, 200, 201, 261 Shuy, Roger W. 262
Sicwierska, Anna 1, 246, 264
Pasanen, Maija-Liisa 75, 262 Skorupka, Stanislaw 246, 264
Phillips, John S. 196, 265 Smith. Henry Lee 59, 264
Peters, Stanley 262 Snook, Roger L. 107, 264
286 Index of authors

Sobieniowski, Florian 162 Tusiani, Juseph 170

Spalatin, Leonardo 16, 18, 264 Twaddell, W. Freeman 121, 265
Stalnakcr, Robert C. 92
Stanislawski, Jan 80, 107, 264 Valdman, Albert 196, 265
Steinberg, Danny D. 259 Van Buren, Paul 1, 3, 12, 23, 107, 161,
Stickel, Gerhard 258 247, 265
Stiebcr, Zdzislaw 54, 264 Van Dijk, Teun 128, 265
Stock well, Robert 20, 36, 54, 59, 107, 120, Varga, Laszlo 59, 246, 265
130, 131, 132, 134, 238, 247, 264 Vennemann, Theo 47, 154, 252
Strevens, Peter 256
Wagner, Karl Heinz 233, 238, 261
Strothmann, F.W. I l l , 261
Wekker, Herman 75, 251
Suomi, Kari 125, 164
Welna, Jerzy 76, 265
Sussex, Roland 37, 264
Whitfield, Francis J. 256
Svartvik, Jan 97, 254, 259
Widdowson, Henry 194, 195, 265
Szwedek, Aleksander 44, 45, 257, 258, 264
Wierzbicka, Anna 89, 265
Wilkins, John 109, 265
Thomason, Richmond H. 92
Wittgenstein, Ludwig 88, 214, 265
Thorne, James 249, 252
Woloszyk, Alicja 59, 61, 62, 265
Tomaszczyk, Jerzy 82, 89, 264
Wojcik, Tomasz 107, 265
Träger, George L. 59, 264
Tran-Thi-Chau 238, 264 Yngve, V. H. 91, 265
Trier, Jost 82, 264
Trubetzkoy, Nikolaus S. 121 Zabrocki, Tadeusz 3, 8, 68, 255, 265
Jacek Fisiak (Editor)

Contrastive Linguistics
Prospects and Problems

1984.14.8 χ 22.8 cm. Χ, 449 pages. Cloth. DM 1 5 8 -

ISBN 90 279 3260 3
(Trends in Linguistics. Studies and Monographs 22)

This collection of papers covers various aspects of contrastive lin-

guistics. Among the topics treated are contrastive phonology, syntax,
text linguistics, pragmatics and applications of contrastive linguistics
to language teaching. The languages investigated include English,
German, Polish and Irish.

Contents:
Robert D. Borsley: Free relatives in Polish and English - Stefan Dyia:
A note on inversion/conjunct ascension constructions in Polish and
English - Nils Erik Enkvist: Contrastive linguistics and text linguis-
tics - Claus Fierch and Gabriele Kasper: Ja und? - og hva' sä? - a
contrastive discourse analysis of gambits in German and Danish -
Rudolf Filipovic: What are the primary data for contrastive analy-
sis? - Charles J. Fillmore: Remarks on contrastive pragmatics - Maria
Grzegorek: English sentences with introductory there and their
Polish counterparts - Edmund Gussmann: Abstract phonology and
contrastive analysis Raymond Hickey: Towards a contrastive
syntax of Irish and English - George M. Horn: Constraints on trans-
formations: evidence from contrastive analysis - Ewa Jaworska: On
the structure of adverbial subordinate constructions in English and
Polish - Andrzej Kopczynski: Problems of quality in conference inter-
preting - Tomasz P. Krzeszowski: Tertium comparationis - Wolfgang
Kühlwein: Pedagogical limitations of contrastive linguistics - Lewis
Mukattash: Contrastive analysis, error analysis and learning diffi-
culty - Wieslaw Oleksy: Towards pragmatic contrastive analysis -
Jerzy Rubach: Rule ordering in phonological interference - Kari Saja-
vaara: Psycholinguistic models, second language acquisition, and
contrastive analysis - Michael Sharwood Smith: Learnability and
second language acquisition - Aleksander Szwedek: Some problems
of contrastive analysis and text linguistics - Tadeusz Zabrocki: On
the nature of movement rules in English and Polish.

mouton de gruyter
Berlin · New York
m Niels Davidsen-Nielsen

m Tense and Mood in English

mj A Comparison with Danish
1990. X, 224 pages. Cloth.
m ISBN 311 0125811
(Topics in English Linguistics l)
m This monograph deals with the grammatical realization of
expressions of time and modality in English.
m
Tense is interpreted as a broad category with eight mem-
m bers, the perfect and future constructions being included.
Mood is assumed to comprise not only morphologically
signalled constructions like the subjunctive and the impera-
m tive, but also syntactically signalled constructions with
modal auxiliaries.
m
This work has both an analytical and a descriptive objec-
tive. One of the purposes is to provide a descriptive basis on
mj which reliable and efficient teaching material may be pre-
pared.
m
By contrasting English with Danish, a language in which the
m forms of expression of modality and time are relatively
similar, important differences are brought to light.

m
m
m
m

m
mouton de gruyter
m Berlin · New York