Non-native/Non-native Conversations:

A Model for Negotiation of Meaning

EVANGELINE MARLOS VARONIS and SUSAN GASS
The University of Michigan

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

This paper builds upon the research investigating conversational interactions
between native speakers (NS) and non-native speakers (NNS) by focusing on inter-
actions among non-native speakers. We examine NS/NS, NS/NNS, and NNS/NNS
conversations, noting that negotiation of meaning is most prevalent among NNS/
NNS pairs. We propose an explanation for this phenomenon and set up a model
to describe it in terms of 'pushdowns' from and 'pops' to the main discourse.
Finally, we suggest the importance of NNS/NNS conversation and especially the
function of negotiation in second language acquisition.

1. INTRODUCTION1
Investigations of non-native speaker (NNS) discourse have focused primarily on
interactions between native speakers (NS) and non-native speakers (see Long 1983a
for a review). In support of the importance of these interactions for L2 acquisition,
Long (1983a and 1983b) has suggested that the modified interaction found in
conversations between native speakers and non-native speakers is the sine qua non
of second language acquisition.
This paper both departs from and builds upon research investigating the nature
of conversational interactions beween native speakers and non-native speakers, by
considering the nature of conversational interactions between non-native speakers.
The major issue we deal with in this paper is how conversations between non-
native speakers differ from those between native speakers on the one hand and
between native speakers and non-native speakers on the other hand.
We suggest that the types of linguistic activities that occur in NNS-NNS con-
versations differ from those in other types of discourse, particularly with respect
to the negotiation of meaning when there has been an actual or potential break-
down. With regard to NS-NS discourse, Schegloff, Jefferson, and Sacks (1977)
argue that adults prefer self-correction over other-correction, perhaps because if an
interlocutor understands the speaker's utterance well enough to correct it, she also
understands it well enough to continue the conversation without correcting it.
Other-correction is seen as embarrassing to interlocutors who are supposedly on
equal footing in the conversation. However, the authors also suggest that other-
correction may be more frequent in adult-child interaction, or in general where
interaction is between those 'not yet competent' in some domain. Extending their
hypothesis, we suggest that non-native speakers, being not yet competent in the
domain of the target language, would thus also be more likely to respond to other-
repair without embarrassment. Because the 'fault' of non-understanding may reside
with either the speaker or the hearer or both, the interlocutors have a shared
incompetence. Therefore, NNS-NNS discourse allows greater opportunity than
NS-NNS or NS-NS discourse for the negotiation of meaning.
In this paper, we first discuss features of NS-NNS discourse. Next, we discuss
aspects of NNS-NNS discourse, reporting data gathered for the present study.
Applied Linguistics, VoL 6, No. 1
72 NON-NATIVE/NON-NATIVE CONVERSATIONS

We set up a model to account for the form of meaning negotiation in non-native

discourse, suggesting the function of these negotiations in the discourse, as well as
their function as part of the acquisition process.
A number of studies have been conducted involving NS-NNS discourse, which
we will briefly summarize. In general, these studies focus on the particular dif-
ficulties that the native speaker has in understanding the non-native speaker, and
the native speaker's .use of what has been termed 'foreigner talk' to compensate for
these difficulties. In particular, native speaker responses to non-native speakers

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

involve more elaboration, more repetition, slower speech, more questions, more of
what may be considered linguistic correction, and greater willingness to allow
a topic shift than do native speaker responses to other native speakers (Long
1983b; Arthur et al. 1980; Chun et al. 1982). Our own work on this subject
(Varonis and Gass 1982 and Gass and Varonis 1984) has indicated that one of the
underlying issues, although clearly not the only one, involved in a native speaker's
use of foreigner talk is the comprehensibility of the input from the non-native
speaker. We further suggested that this comprehensibility involves a number of
factors, including linguistic features such as the non-native speaker's pronunciation
and grammar, as well as native speaker-related variables such as familiarity with
the topic of conversation and with non-native speech in general.

2. DATABASE
This study is designed to move beyond a discussion of NS-NNS discourse to an
examination of conversations where both interlocutors are non-native. NS-NS and
NS-NNS pairs are also examined as a control. The study reported on in this paper
comes from three sources: (1) fourteen conversational dyads between non-native
speakers; (2) four conversational dyads between a native and non-native speaker;
(3) four conversational dyads between native speakers. The dyads were composed
of speakers who had not previously met.2 All members of the NNS-NNS dyads
were students at the English Language Institute of The University of Michigan. The
members of these "dyads were of the same sex (seven male pairs and seven female
pairs). All were native speakers of either Spanish or Japanese. Each dyad was
audio-taped in an informal conversation with no instructions other than to talk
in English with each other. They were given hints such as 'introduce yourselves
and find out about each other'.
The members of the NS-NNS dyads were taped at their first meeting as con-
versation partners. In this case it was not possible to control for sex. The members
of the NS-NS dyads were two female pairs and two male pairs, all students at the
University of Michigan, and not previously acquainted with one another. For all con-
ditions, the first five minutes of each conversation were analysed for the purpose of
looking at the instances of non-understandings which occurred in the discourse.

3. NNS-NNS DISCOURSE
In most conversations (regardless of whether the participants are native or non-
native), the discourse progresses in a linear fashion, that we will be representing
in this paper by a horizontal line. When the interlocutors share a common back-
ground and language, the turn-taking sequence is likely to proceed smoothly,
reflecting what Jones and Gerard (1967) call a 'symmetric contingency', each
speaker responding to the utterance of the previous speaker, while maintaining
her own sense of direction in the discourse. However, in discourse where there
 EVANGELINE VARONIS AND SUSAN GASS 73

is not shared background, or in which there is some acknowledged 'incompetence',

the conversational flow is marred by numerous interruptions. These may be seen as
vertical sequences in a horizontal progression. Interlocutors are far more likely to
lose their conversational footing as a result of confusing aspects of the discourse
as a whole. As a result, they may learn to compensate by questioning particular
utterances and/or requesting conversational help. This is not necessarily the same
as what Jefferson (1972) calls 'side-sequences', in other words, breaks from the
main flow of conversation, but rather those side-sequences that specifically involve

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

the negotiation of meaning which is crucial to the success of the discourse. These
sequences allow the participants, in some sense, to maintain as well as possible
equal footing in the conversation. We define equal footing within a conversation
as one interlocutor's ability to respond appropriately to another interlocutor's
last utterance—in other words, to take a turn when it becomes available with full
understanding of the preceding turn and its place in the discourse. In what follows
we set up a model to account for the sequences, which we call non-understanding
routines, that help interlocutors regain their places in a conversation after one or
both have 'slipped'. We suggest that the conversational episode which is represented
by this model serves essentially one of two functions: (1) negotiation of non-
understandings, and/or (2) continuation of the conversation. Non-understanding3
routines are operationally defined as: those exchanges in which there is some overt
indication that understanding between participants has not been complete.
The sine qua non of a non-understanding routine is that within the exchange
there are embeddings of one or more clarifications. These can stem from an exchange
in which there was a mis-understanding, no understanding, or incomplete under-
standing.
An example of non-understanding is given in (1) below.4
1 Non-understanding
140 S When are you going to visit me?
140 J Pardon me?
We exclude exchanges such as (2) below, which we consider an example of a mis-
understanding which has gone unrecognized.
2 Mis-understanding (data from Varonis 1981)
Bank employee and customer, discussing NOW account
NNS: OK. Another thin' please . . m . . do you add another service charge
for theses kindes?
NS: Yes, there is a twelve dollar service charge—twelve dollar membership
fee per year of twelve dollars with VISA and Mastercharge.
In this example the word 'theses' was interpreted by the native speaker as VISA,
most likely because of the banking context.

4. A MODEL FOR THE NEGOTIATION OF MEANING

4.1. Presentation
We begin by noting that examples as in (1) above are extremely frequent in NNS-NNS
discourse, at times occupying the major portion of the conversational interaction at
the expense of moving the conversation along in a horizontal fashion. To deal with
these deviations we propose a model with four functional primes (see Figure 1).
74 'NQN-NATIVE/NON-NATIVE CONVERSATIONS

Trigger Resolution

Fig: 1: Proposed model for non-understandings

The first part of the model consists of a trigger, denoted by T. The second part of
the model we term the resolution. It consists of an indicator (which we denote by

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

I), a response (denoted by R), and a reaction to the response (denoted by RR).
Simply put, the trigger is that utterance or portion of an utterance on the part
of the speaker which results in some indication of non-understanding on the part of
the hearer. The hearer has essentially two immediate choices—to ignore the trigger
or to react to it in some way. Examples of these two options are presented in
(3a) and (3b) and (4a) and (4b) . s
3 Ignore
a. hoping for more information
NS: Do you think his research is monolithic?
NS: Well, it's hard to say.
b. considered irrelevant
Julie: I'm sorry to have called you so early, but I'm just on my way out
of the house for (indiscernible)
Martha: no response
4 Comment upon it
a. echo
140 S: When can you go to visit me?
-»• 140 J: visit?
b. overt indication
140 J: ...research
-*• 140 J: Research, I don't know the meaning.
In (3 a) the native speaker respondent did not know the meaning of the word
'monolithic' in that context and, as he explained later, tried 'every conversational
strategy I could think of to get more contextual information' so that he would
not have to admit that he had not understood the question. It is important to note
that there was no misunderstanding. Rather, the second native speaker had no idea
of that particular meaning of the word 'monolithic'. In (3b) Julie called Martha
long-distance early on a Sunday morning. Martha realized from something later on
in the conversation that the indiscernible portion was Toronto'. She also realized
that the reason for the call was more important than rinding out where she was
going that day, so she decided not to comment on the fact that she had not under-
stood, and let Julie go on with her explanation of why she was calling.
Commenting on a non-understanding is exemplified in two ways in (4a) and
(4b). In (4a), 140 J echoes what is probably the word which he has some question
about. In (4b) there is a direct appeal for assistance.
When a trigger is ignored, there is often no way for the investigator to recognize
that there was a breakdown in comprehension, although something later in the dis-^
course may indicate that in fact the listener had not understood (see (2) above).
What we are concerned within this paper is only those instances of non-understanding
 EVANGELINE VARONIS AND SUSAN GASS 75

in which there is some overt marker on the part of the hearer, indicating an inter-
ruption in the horizontal flow of the conversation.
The second part of the model is what we call the resolution, potentially encom-
passing the second, third, and fourth primes. The second prime is an indicator (I).
This is an utterance on the part of the hearer that essentially halts the horizontal pro-
gression of the conversation and begins the downward progression, having the effect
of 'pushing down' the conversation rather than impelling it forward.
The third prime is the speaker's response (R) to the indicator, acknowledging
the non-understanding in some way. The fourth prime, an optional element, is the

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

reaction to the response (RR).
The model in Figure 2 suggests that comprehension checks can occur at any
step of the way; we further claim that one can 'pop'-6 out of the conversation at
any point designated by an arrow. A pop before the response is possible, but not
cooperative in Grice's (1975) terms, indicating either that the first speaker does not
recognize the fact that the conversation has to be negotiated, or that she is not
willing to negotiate. Interestingly, we did not find any examples of this in our
data. We did, however, find instances in which the sequence of T, I, R, RR can
occur, followed by a second resolution on the original trigger. We will now illustrate
the primes of the model and then discuss its implications.

My father now retire? yes oh yeah

is retired

T • ^(CC)->I-»(CC)-»-R->(CC)-»RR->(CC)^

Fig. 2: Expanded model

4.1.1 Triggers. Triggers can initiate from any aspect of the discourse. In (5a) we
present an example in which it comes from a question.
5a Trigger as question
-* ULS: What is your name?
120 S: My name?
ULS: yeah
In (5b) there is an example in which the trigger comes from an answer to a
question.
5b UL J: yeah. How long . . will you be? will you be staying?
-»• 120 J: I will four months
ULJ: four months?
120 J: stay four months here until April
In (5c) the trigger comes from neither a question nor an answer.
5c Trigger as neither question nor answer
-»• 120 S: . . . and and the condition for uh bets my level in my company it
necessary my speaking English
UL S: hm you mean that English is important in your company to (indis-
cern.)
76 NON-NATIVE/NON-NATIVE CONVERSATIONS

In essence, a trigger is recognized only in retrospect, in other words, if it has

been reacted to by the hearer.

4.1.2 Indicators. Indicators signal that an utterance has triggered a non-understand-

ing. In Schachter's terms (in press), this is 'negative input', that is, 'information
provided to the learner that her utterance was in some way deviant or unacceptable
to the NS\ We present in (6) a number of ways in which one member of the con-
versational pair signals to another that something has gone wrong. In (6a) the sig-

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

nals are in the form of an echo. In our data these occurred most often with rising
intonation. In (6b) there is an explicit statement of non-understanding. In (6c)
there is no verbal response. In (6d) there is an inappropriate response (cf. Varonis
and Gass (in press) for a discussion of how such incorrect interpretations can lead to
a breakdown in conversations).

6 Examples of indicators
a. Echo
i Rising intonation
ULS: What is your name?
-»• 120 S: My name?
ULS: yeah
ii Falling intonation
140 S: But he work with uh uh institution
-*• 140 J: institution
140 S: Do you know that?
b. Explicit statement of non-understanding
UL J: Are you a student in your country?
120 S: in my class?
UL J: in your country?
-»• 120 S: Oh, I don't understand.
UL J: OK OK so what did you do in your country?
c. No verbal response
UL S: What is your purpose for studying English in Ann Arbor?
-»• 120 S: silence
UL S: What is your purpose for studying English?
d. Inappropriate response
UL J: Are you a student in your country?
-»• 120 S: in my class?
UL J: in your country

In Table 1 we provide a list and examples of the devices which we found to

function as indicators.

4.1.3 Responses. In (7) we focus on types of responses to the request for addi-
tional information which has been either implicitly or explicitly stated in the
form of an indicator. In (7a) we have an example of a repetition of the previous
utterance. In (7b) we have an example of an expansion, and in (7c) there is an
example of a rephrasing as a response. In (7d) an acknowledgement of the
indicator is given, and finally in (7e) we give an example of a reduction of the
input.
 EVANGELINE VARONIS AND SUSAN GASS 77

Table 1: Categories of Indicators

1 Explicit indication of non-understanding

ex. pardon1., what"! I don't understand
2 Echo word or phrase from previous utterance
3 Non-verbal response
silence or mmmm
4 Summary
ex. Do you mean"!
5 Surprise reaction

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

ex. ReaUyi, did she?
6 Inappropriate response
7 Overt correction
120 S: They say that here is no I no problem for this temperature OK?
ULS: You say you don't have
120 S: You don't have
UL S: I don't have
120 S: I don't have problem

7 Examples of responses
a. Repetition
140 J: This is your 2 term?
140 J: Pardon me?
-*• 140 J: 2 term, this is this term is term your 2 term
b. Expansion
UL J: yeah, How long . . . will you be? will you be staying?
120 J: I will be four months
ULJ: four months?
-»• 120 J: stay four months here until April
c. Rephrasing
140 S: You know heating?
140 J: So it is a heat exchanger
-»• 140 S: radiator
d. Acknowledgement
140 S: When can you go to visit me?
140 J: visit?
-»• 140 S: yes
e. Reduction
UL S: What is your purpose for studying English in Ann Arbor?
120 S: silence
-*• UL S: What is your purpose for studying English?

4.1.4 Reactions to Response. RR's are an optional unit of the routine, in some
way tying up the routine before the speakers pop back up to the main flow of
conversation. An example can be seen in (8).
8 120 S: My father now is retire.
120 J: retire?
120 S: yes
120 J: oh yeah
78 NON-NATIVE/NON-NATTVE CONVERSATIONS

4.1.5 Comprehension Checks. As mentioned earlier, comprehension checks can

optionally occur in any of four places, as can be seen in (9a-d) below: (a) after
the trigger (or potential trigger); (b) after the indicator; (c) after the response, and
(d) after the reaction to the response.
9 Comprehension checks
a. 140 J: I was born in Nagasaki.
-»• Do you know Nagasaki?
b. 120 S: I'm from Venezuela.

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

ULJ: Venezuela
-»• 120 S: Do you know?
c. 140 S: declares her ingress
140 J: Ingless
140 S: Yes, if for example, if you. when you work you had an ingress.
-> you know?
d. 140 S: and your family have some ingress
140 J: yes ah, OK OK
-»• 140 S: more or less OK?

4.2 Extended Example

What we have presented so far are common examples of the downward movement
of a conversation, not unlike what has been noted in the literature on NS-NNS
discourse. We turn now to situations which do not seem to occur in NS-NNS dis-
course with the same frequency and complexity which we have found in NNS-
NNS discourse. In NNS-NNS discourse we have found numerous examples which
consist of multiple layers of trigger-resolution sequences, at times with inter-
vening returns to the conversation (as in Figure 3).
I'm not really
I'm living in Osaka mean Osaka city
It's near city
T| • / I • R • R R \
yeah yeah Osaka I
Osjka/

Fig. 3: Embeddings in negotiations of meaning

We now present the most complex and lengthy (more than two minutes of
the transcnbed five-minute conversation) example of embedded non-understanding
routines which occurred in our data. In (10) below we present the entire negotia-
tion and in Figure 4 we model it.
10 140 J: And your what is your mm father's job?
140 S: My father now is retire. T
140 J: retire? I
140 S: yes R
140 J: oh yeah RR
 EVANGELINE VAR0N1S AND SUSAN GASS 79

140 S: But he work with uh uh institution T

140 J: institution I
140 S: Do you know that? The name i s . . . some thin like eh control
of the state CC/R/T
140 J: aaaaaaaaah RR
140 S: Do you understand more or less? CC
140 J: State is uh . . . what what kind of state? I
140 S: It is uhm R

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

140 J: Michigan State? I
140 S: No, the all nation R/T
140 J: No (back channel) government? I
140 S: all the nation, all the nation, Do you know for example is a the
the institution mmm of the state mm of Venezuela R
140 J: ah ah RR
140 S: had to declare declare? her ingress. T
140 J: English? I
140 S: No, English no (laugh) . . . ingress, her ingress R
140 S: Ingress? I
140 S: Ingress, yes, INGRESS more or less R
140 J: Ingless I
140 S: Yes, if for example, if you, when you work you had an
ingress, you know? R/CC
140J: uh huh an ingless? RR/I
140 S: yes R
140 J: uhhuhOK RR
140 S: yes, if for example, your homna, husband works, when
finish, when end the month his job, his boss pay—mm—him
something R
140 J: aaaah RR
140 S: and your family have some ingress R
140 J: yes ah, OK OK RR
140 S: more or less OK? and in this in this institution take care of
all ingress of the company and review the accounts CC/R
140 J: OK I got, I see RR
140 S: OK my father work there, but now he is old

The speakers are two women in an intermediate class at the English Language
Institute of the University of Michigan. Maria is a Spanish speaker, and Mieko
a Japanese speaker; thus they constitute an example of a dyad of the same pro-
ficiency, but different language backgrounds.
Mieko has asked Maria what her father does. Maria responds by saying 'My
father now is retire'. The statement (probably the word 'retire') serves as a trigger.
Mieko indicates such by echoing with rising intonation the word 'retire?', to which
Maria responds with 'yes'. Mieko reacts to the response with 'oh yeah'. The con-
versation then 'pops' to the original point and moves along horizontally until Maria
uses the word 'institution'. This apparently serves as a trigger for a second instance
of non-understanding. Mieko again indicates a problem by echoing 'institution',
which we assume is the word that caused the problem. This is followed by Maria's
comprehension check: 'do you know that?' and then an expansion: 'something like
 But he work with
My Tathcr now li retire uh uh institution
OK my Father work there, but now lie n o
piT—+f~T—* R —• R R " V ^ W\ •/~~i CC R RR
— I iclire yei oh y e a h / — I insillutlon Do you control aaah Do you understand 1
know thai' of (he Hale more or leis? J

hat kind It iiuhm )

fMaie-> J

~ T-\
( Michfgai
ichigin no, the all j
1 late* nation J
V
R
»H the nation

declare I.er
tngreii

pr ( Englisll1 No, Enjluh no injieil, I

he
\ V ""»"« 7
I 7
R X
Ingress lngre«, yet. \
INC R E S S y
more or \t\\J

I R CC RR
Ingless yes, if for example, you know* uh huh
if you when you work
you had an tngrets,

I R RR
an inglcis' yes uh huh OK

yes, if lot example

your lioiiinj htitbjitd
v»-ofki, when limit),
when end (he the
month his job, hit
boss pay him something

CC
and your tantlly have yes ah OK OK .He HI Ictt OK"
some ingress
Fig. 4: Model of negotiation of meaning in a NNS-NNS conversation

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

 EVANGELINE VARONIS AND SUSAN GASS 81

eh control of the state'. Mieko acknowledges with 'aaaah'. However, Maria is sensi-
tive to the fact that Mieko has probably not yet understood, despite her overt
response to the contrary. As a result, Maria offers a comprehension check: 'Do
you understand more or less?'
But Mieko doesn't understand. She asks 'what kind of state?' In the lengthy
exchange which follows, indicators were frequently in the form of guesses. For
example, while they are trying to negotiate the word 'state', Mieko provides an
'out' by saying 'Michigan State?', while Maria offers 'nation', which only pushes

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

the conversation further down. Mieko suggests 'government', indicating her con-
tinuing lack of understanding. The multiple pushdowns continue, with Maria still
trying to explain what her father did before his retirement. She offers 'had to
declare—declare?—her ingress'. Mieko indicates her lack of understanding by
echoing 'English?' which Maria responds to in the form of rejection, laughter, and
then repetition: 'No, English no (laugh) ingress, her ingress'. Mieko then echoes
this with rising intonation 'ingress?', this time more or less correctly. Maria's
response is a repetition: 'ingress' with an affirmation, 'yes' plus spelling of the word.
However, Mieko repeats 'ingless', which requires Maria to respond, again, this time
in the form of an example followed by a confirmation check. Mieko again repeats
the word with rising intonation 'uh huh an ingless?' signalling that comprehension
is not complete. However, Maria's response to this indicator is 'yes', which is a typi-
cal one when a speaker is ready to 'pop' to the original conversation. In fact, Mieko
realizes this and gives a typical RR 'uh huh OK', which is agreement to return to
the conversation. Maria, however, must have realized that her explanation was such
that no one could have understood what she meant, and continues with a new
response which essentially consists of another example, until Mieko finally reacts
with a true signal of understanding: 'OK I got, I see'. At this point the pushdown
pops and they return to the original conversation with an 'OK, my father work
there, but now he is old'.
As can be seen, this negotiation is not only lengthy, but also involves multiple
embeddings. In fact, there are four vertical levels of negotiation, involving nine
separate indicators and eleven responses before the problem is finally resolved. The
model we have proposed reveals the complexities involved in non-understanding
routines, allowing us to measure and compare the depth, i.e. the complexity of
these non-understandings before the non-understanding is resolved or the attempt
to resolve it is abandoned.
This model suggests that negotiations of meaning exist within, but separate
from, the main discourse, in effect putting the progression of the discourse on
'hold'. A similar phenomenon with regard to the organization of discourse has
been noted by Grosz (1977, 1979) who found that pronominal anaphora within
a pushdown is self-contained, whereas pronominal anaphora after a pop references
the text prior to the pushdown.7

4.3 Conversational Continuants

We now turn to the second category of exchanges which this model describes: conver-
sational continuants. The exchanges in this category function to keep the conversa-
tion going. In (11) below we present examples of exchanges in this category.
11 Surprise Indicators
a: 120 S: 11 have three weeks here
82 NON-NATIVE/NON-NATIVE CONVERSATIONS

-* ULJ: Oh really?
b. 120 J: I'm from Nagoya.
-»• UL J: Are you from Nagoya?
120 J: yeahhh
UL J: I'm from. eh. suburb Nagoya
c. 140 J: . . . but I. write. I write a letter to my husband every Friday every
Friday
-* 140 J: so you write four times.

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

140 J: yeah
In many instances a particular exchange is ambiguous with regard to whether it is
truly an example of a conversational continuant or whether it is an indicator of a non-
understanding. In fact, in some instances it may serve both functions simultaneously.
Chun et at (1982) suggest that confirmation checks in particular are both a gentle
means of indicating a problem in the conversation, and an act to encourage the
interlocutor to continue. Whenever an utterance in our data-base was ambiguous
with respect to classification, we did not include it in our count of indicators.
While we claim that our model serves two distinct functions, we do not want to
suggest that these functions are always separate. There are clearly instances when
one member of a conversational dyad tries to appear relevant and knowledgeable
by using continuing devices in just those situations when there has been no under-
standing, but when one is trying to hide the lack of understanding. Consider the
following example from Keller-Cohen (1980) which, while not taken from a wit-
nessed conversation, is none the less plausible and probably common in discourse.
12 Guest: I generally weave using warp way stripes.
-•• You: Oh, warp way stripes.
Guest: They have wonderful dynamism, yet such simplicity.
-* You: Yes. Dynamic yet simple.
Guest: The color possibilities seem almost limitless once you've mastered
the basic technique.
You: Limitless possibilities!

5. RESULTS AND DISCUSSION

We now turn to a discussion of the differences between NNS-NNS discourse and
other reported discourse involving non-native speakers, and suggest implications
for learning based on these differences.
For there to be successful communication, there must be attentiveness and
involvement in the discourse itself by all participants. Stevick (1976, 1980, 1981)
claims that it is precisely active involvement which is a facilitator of acquisition in
that it 'charges' the input and allows it to 'penetrate' deeply. Note that in Stevick's
framework, the charge comes from the hearer and not from the speaker-source.
Scarcella and Higa (1981), in a study comparing conversations involving native
speakers and non-native children and conversations between native speakers and
non-native adolescents, found that there was a greater amount of simplified input
to children than to adolescents, but that the adolescents 'worked' harder than
younger learners at sustaining the discourse. Because they were more involved
in keeping the conversation going, the input that they received was more 'charged'
in Stevick's sense. Assuming that comprehensible input (Krashen 1980) is essential
to successful acquisition, they suggest that simplified input is not as 'optimal' as
 EVANGELINE VARONIS AND SUSAN GASS 83

the input which comes as a result of negotiation work. We further suggest that
negotiation work is greater in NNS-NNS discourse than in other types of discourse
and that, hence, it is a good forum for obtaining input necessary for acquisition.

Table 2: Number of Pushdown Routines

NS-NS (A) NS-NNS (B) NNS-NNS (Q

n=4 n= 4 n = 14
x = .'5O x"=2.75 x = 10.29

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

sd = .58 sd = .50 sd = 4.79
t-tests(l-tafl)
A vs. B t = 6.08, df = 7, p < .0005
A vs. C t = 7.46, df = 17, p < .0005
B vs. C t = 5.78, df = 17, p < .0005
These three comparisons are simultaneously significant at the .005 level, using Bonfenoni
inequality.

In Table 2 we present the results comparing the incidence of non-understanding

routines8 among the three populations of this study. First, it is obvious that these
routines are not common in NS-NS discourse; of four dyads, there were only two
occurrences of such a routine, one in a male pair and one in a female pair for an
average of .50 (sd = .58).' Second, it is also obvious that NS-NNS dyads do evi-
dence a fairly regular use of this type of routine: there were eleven occurrences in
four dyads, for an average of 2.75 (sd = .50). Finally, NNS-NNS pairs show the
greatest incidence of the non-understanding routines, 144 occurrences in fourteen
dyads for an average of 10.29 (sd = 4.79). These three means were compared using
t-test procedures.10 Differences between the NNS-NNS dyads and each of the other
two groups was significant (t = 7.46, df = 17, 1-tail, p < .0005 [NNS-NNS vs.
NS-NS]; t = 5.78, df = 17, 1-tail, p < .0005 [NNS-NNS vs. NS-NNS]). In addi-
tion, the difference between the NS-NS and NS-NNS groups was significant (t =
6.08, df = 7, 1 tail, p < .0005). One can make a simultaneous statement, using
a Bonferroni inequality, asserting that these levels obtained simultaneously at
the .005 level of significance. We further point out that many of the routines
in the NNS-NNS dyads are actually pushdowns in a larger routine (31 of 144,
or 22 per cent), indicating that NNS-NNS pairs not only spend more time negotiat-
ing than the other pairs, but also that their non-understandings involve more
work in the resolution before the routine finally pops and the conversation
continues. Thus, the more involved non-native speakers are in a dyad, the more
time interlocutors will spend moving down, or in other words, in the negotiation
of meaning, rather than moving forward, in other words, in the progression of
the discourse.
In addition to noting these differences between dyads based on the amount of
involvement of non-native speakers, we also point out differences within the group
of NNS-NNS dyads. Within this group, speakers were either of the same or dif-
ferent language, and of the same or different proficiency. We predicted that those
who had the most in common would also have the least to negotiate. In fact, this
turned out to be the case. These differences are presented in Table 3.
The lowest incidence of non-understanding routines occurred in exactly those
dyads that shared a language and a proficiency level, for an average of 4.75 (sd = .50).
Note, however, that this average is still higher than that of the NS-NNS pairs
84 NON-NATIVE/NON-NATIVE CONVERSATIONS

Table 3: Number of Pushdown Routines in NNS Population

Group 1 Group 2 Group 3

Same language Same/Different Different Language
Same proficiency Language/Proficiency Different proficiency

4 dyads 7 dyads 3 dyads

x = 4.75 x = 11.00 x = 16.00
sd = .50 sd = 3.06 sd = 3.00

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

0 degrees of difference 1 degree of difference 2 degrees of difference

(x = 2.75). The next lowest incidence of routines occurred in those dyads that
shared either language or proficiency background, but not both. This group averaged
11.00 pushdown routines per dyad (sd = 3.06).11 The highest incidence of negotia-
tion routines was found in those dyads that evidenced the most differences, in
other words, those that shared neither a language nor a proficiency background.
This gToup averaged 16.00 routines per dyad (sd = 3.00). Results of t-tests are given
in Table 4.

Table 4. t-tests Comparing NNS Populations

Group 1 Group 2 Group 3

Group 1
Group 2 5.28*
Group 3 6.34** 2.40***
* P < .0005, df = 10
** p < .0005, df = 6

The first two of these comparisons are simultaneously significant at the .005 level of signifi-
cance, using Bonferroni inequality.

As can be seen, a comparison of the means yielded a significant difference in

two of the three cases (t = 5.28, df = 10,1-tail, p < .0005, Group 1 vs. 2; t = 6.43,
df = 6, 1-tail, p < .0005, Group 1 vs. 3;t = 2.40, df = 9,1-tail, p < .025, Group 2
vs. 3). These values of t are simultaneously significant beyond the .05 level when
alpha is adjusted using the Bonferroni procedure (.05/3 = .017). These results
support our hypothesis that the greater the degree of difference which exists in the
backgrounds of the conversational participants, the greater the amount of negotia-
tion in the conversation between two non-native speakers.
Given these results, we suggest that a greater amount of negotiation work takes
place in NNS-NNS discourse than in either NS-NS or NNS-NS discourse. This
negotiation serves the function of providing the participants with a greater amount
of comprehensible input. As mentioned above, non-native speakers, in conversa-
tions with other non-native speakers, may feel that as learners they have little to
lose by indicating a non-understanding, because they recognize their 'shared incom-
petence'. In fact, in the NNS-NNS dyads, participants frequently commented on
bow bad their English was. We claim that it is the shared incompetence in the
domain of English which allows them to 'put the conversation on hold' while they
 EVANGELINE VARONIS AND SUSAN GASS 85

negotiate meaning. As seen, this type of deviation occurs much less frequently
in NS-NNS discourse, where the non-native speaker recognizes the inequality
of the conversational situation. We did, however, find an example which
approaches the levels of complexity witnessed in our data in an exchange
between two friends —a native speaker and a non-native speaker—from Schu-
mann's 1975 transcript, cited by Schachter (in press). We present this conversation
in (13) below.
13 Data from Schumann 1975 transcripts—taken from Schachter (in press)

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

NS: What's the movie tonight? (referring to TV)
NSS: I don't know.
NS: What was it last week?
NNS: Yesterday?
NS: yeah.
NNS: Em, ah, no me no, no looked, no?
NS: You didn't look at it?
NNS: No, eh, eh, I look play.
NS: You play?
NNS: No, I look play hockey. The game.
NS: You play hockey? You play the game?
NNS: No! In the television.
NS: Uhhuh?
NNS: I'm looking one game.
NS: At a game, you looked at a game on television. What kind of a game?
NNS: Hockey.
In this case it may be the fact that the participants are friends which allows the
toleration of the suspended conversation. In support of this hypothesis, we note
that Day, Chenoweth, Chun and Luppescu (1984) report that in conversations
between NS-NNS friends, the native speakers produced on-record corrections more
often than off-record corrections, despite a prediction by Brown and Levinson
(1978) that the opposite would occur. Day et al. suggest that this is because the
native speakers felt secure in their relationship with their NNS friends. It is to be
further noted that there may not have been any forward movement in the above
conversation, in that the topic under consideration (what is on television) may have
been an isolated exchange, rather than forming part of a continuing conversation.
Thus, the suspension may not have been a major interruption of the conversation
at all.
We have argued that non-native speakers do not lose face by negotiating meaning
in the same way they might with native speakers. Because all our non-native
speakers were learners, enrolled in an intensive language institute, they had nothing
invested in trying to come across as totally competent speakers of English. We
might then predict that non-native speakers who did not consider themselves
language students might be less willing, given equivalent language ability, to admit
a language difficulty. As an example, two native speakers were once escorted all
the way across campus at UCLA by a Japanese 'visiting scholar" whom they had
asked for directions. He did not want to leave them, as visitors obviously unfamiliar
with the area, yet could not relate the complicated directions well enough even to
attempt an explanation which they then might have negotiated. He escorted them,
despite the fact that the trip took him well over a mile out of his way.
86 NON-NATIVE/NON-NATIVE CONVERSATIONS

6. CONCLUSION
In conclusion, we have proposed a model to deal with negotiation of meaning in
NNS-NNS discourse, a model which provides a means for dealing with the com-
plexities of negotiation. Communication between any two interlocutors is facilitated
when they share a common background, linguistic and otherwise. Even so, as
a number of investigators note (Grosz 1979; Hobbs and Robinson 1978; Labov and
Fanshel 1977; Marios 1981; Varonis and Gass in press), interlocutors need to share
also an understanding of each other's goals in the interaction for the conversation

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

to proceed with maximum ease. Among non-native speakers, however, even when
interlocutors realize each other's goals, other factors, for example linguistic, social,
or cultural factors, may lead to a breakdown in communication. Such breakdowns
must be negotiated before exchange of information can proceed. In addition,
negotiation may strengthen the social or interpersonal dimension of an inter-
action by acting as a conversational continuant.
We have established that negotiations of meaning occur with greater frequency
in NNS-NNS dyads than in dyads that include native speakers. Among NNS-NNS
pairs, this need for negotiation is probably due to the lack of shared background
between non-native speakers. This is true even for non-native speakers of the same
ethnic background, because the medium of communication—English—is foreign
to both. Among NS-NNS pairs, there is also a lack of shared background that
would presumably lead to an amount of negotiation of meaning which is greater
than or equal to what is found in NNS-NNS pairs. However, we suggest that the
inequality in the status of the participants (with regard to the language medium)
actually discourages negotiation, because it amplifies rather than masks the
differences between them. As a result, there is a greater tendency for conversa-
tion to proceed without negotiation. Among NS-NS speakers, shared back-
ground is maximized, and, as a result, there is little need for linguistic nego-
tiation.
However, this model may also be used to describe conversations among native
speaker interlocutors where a lack of shared background, linguistic or cultural,
or a lack of shared values may lead to the need to negotiate meaning. Further-
more, echoing a response with rising intonation is used in psychotherapy to
focus a patient's attention on an aspect of the discourse which may reflect a source
of problem for the patient. As an example, consider (14) below:
14 From Zemon-Gass (personal communication)
Patient: My husband won't allow me to work.
-*• Therapist: Allow?
Another extension of this type of conversational device can be found in a class-
room situation between student and teacher. Consider the following example.
15 Example from Eric Rabkin (personal communication)
Teacher: Compare the openings of Absalom, Absalom and A Farewell to
Arms. In Absalom, Absalom, there is a greater use of hypotaxis.
-* Student: Hypotaxis?
In this case, the student was questioning the use and meaning of the word 'hypo-
taxis'. StiD another use of this type of conversational device comes in 'consciousness
raising', when one person is attempting to make the other aware of the significance
of the words they use and the ways in which words reflect their political views.
 EVANGELINE VARONIS AND SUSAN GASS' 87

This is not unlike the psychotherapy example. In (16) below, the woman attempts
to make the man aware that 'girl' is not an appropriate word for an adult female.
16 Male (speaking to a woman in her late twenties):
I saw a girl today who looked just like you.
Female: Girl?
In all these examples, interlocutors indicate their need to negotiate meaning before
the conversation can progress.

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

We suggest that the discourse resulting from NNS-NNS interactions serves an
important function for non-native speakers. First, it allows them a non-threatening
forum within which to practise developing language skills. Second, it provides them
with an opportunity to receive input which they have made comprehensible
through negotiation. In fact, we propose that this type of interaction facilitates
the second language acquisition process, agreeing with Schwartz (1980) who claims
that 'second language learners of English can learn more from one another than
they think they can'.
Like other types of discourse situations, NNS-NNS discourse is not monolithic.
We have dealt with negotiations of meaning in NNS-NNS discourse in general and
compared them to similar negotiations in dyads involving at least one native
speaker. However, there is much more to investigate, particularly the kinds of dis-
course situations that non-native speakers are involved in, and the demands made
upon the interlocutors to the quantity and complexity of negotiations in NNS-NNS
discourse. Gass and Varonis (in press) show that the control of information is an
important variable to consider; interlocutors seeking particular information are
more likely to initiate negotiation than interlocutors holding the information. Other
important sociolinguistic variables to consider include ethnicity, native language,
the role of the participants and their respective status, their sex, their age, the num-
ber of participants in the discourse, and the topic of discourse, leaving this a rich
area for future research.
(Received June 1983, revised December 1983)

NOTES
1 This is a revised version of a paper presented at TESOL 1983, Toronto. The order of the
names does not reflect the contribution of the authors, but rather our commitment to alternate
names in our ongoing research. We thank Craig Chaudron for insightful comments on an earlier
version. We also acknowledge Ken Guire of the University of Michigan Statistical Research Lab,
Evelyn Hatch, and Carolyn Madden for helping us clarify our thinking on statistical concepts.
As always, Josh Ard and Orestes Varonis provided intellectual and moral support from begin-
ning to end and beyond. We alone are responsible for all mis-understandings and non-under-
standings which remain.
2 One NNS-NNS dyad was composed of two speakers who had been in the same class for
one week prior to the taping.
3 Non-understanding is perhaps an infelicitous term, since we include confirmation checks in
this category. These, while indicating some potential of incomplete message transmission or
reception, are not non-understandings in the same sense as a response such as 'What did you
say?' or 'I don't understand*. The issue of confidence in one's ability to comprehend may be
relevant with regard to confirmation checks, although a discussion of this issue goes beyond
the scope of this paper (cf. Varonis and Gass (in press) for a further discussion of this issue).
4 The following abbreviations will be used in this paper.
S = Native Spanish speaker
88 NON-NATIVE/NON-NATIVE CONVERSATIONS

J = Native Japanese speaker

120 These refer to proficiency levels at the English Language Institute of the Univer-
140 sity of Michigan. 120 is a beginning level, 140 an intermediate level and UL (upper
UL level is an advanced level.
5 Not all of the examples come from the data collected specifically for this study. Many
are ones which we have collected over the years.
6 The terms 'pushdown' and 'pop' axe borrowed from the computer science literature. (See
Grosz 1977, Linde and Goguen 1978, and Marios 1981 for further examples of the use of this
term in discourse analysis.)

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

7 We thank Joyce Friedman for pointing out to us the relevance of Grosz's work to our
own work.
8 We exclude responses such as 'really', which seem to serve the function of keeping a con-
versation going, as opposed to requesting clarification.
9 There is also an example of a pushdown in the NS-NS data that we did not include
because there was no indication of non-understanding. One of the female pairs happened to
consist of two Jewish women, Sharon and Barbara. They began by discussing university classes,
and then Sharon brought up her planned trip to Israel in the summer.
Sharon: . . . see my father's like really scared . . . and—Have you been there?
Barbara: Israel, no.
Sharon: Are you Jewish?
Baxbaxa: uh hunh
Sharon: and he's like really scared . . .
Sharon's sudden question 'Have you ever been there?' in the middle of her own discourse
seems out of place, as if she were interrupting herself. We suggest, however, that this is an
example of a different type of negotiation, namely, negotiation of identity. Sharon wanted
to know if Barbara was an in-gioup member, namely, a Jewish-American woman, with whom
she could speak about in-group subjects, such as dating and marrying Jewish men. In fact,
having discovered the ethnic tie, Sharon quickly carried the conversation to a more per-
sonal level and interrogated Barbara about her Jewish fiance'. Such pushdowns are impor-
tant in establishing -social footing in the conversation, and certainly are a ripe area for
research.
We also point out that our results would probably have even greater support than these
data suggest. Five minutes of NS-NS and NS-NNS conversation produce more talk than five
minutes of NNS-NNS conversation, so that the differences we found among the three groups
would probably be greater had a unit of measurement such as a conversational turn, rather than
five minutes of conversation, been used. We thank an anonymous Applied Linguistics reader
for pointing this out to us.
10 The question of which statistic to use for these data is a problematic one. The appropriate-
ness of t-tests is questionable, since this test assumes normal distribution. In the case of the
NS-NS category, the range of values is limited so that the assumption here is violated. On the
other hand, a discrete approach such as that of a x2 test is difficult, given the small sample
size. Because the discrete categories which we would have had to use for a x J test were not
consistent with the way the experiment was set up, we decided not to use this latter statistic.
Nonetheless, we have included the statistical analyses, because we believe the t-test to be
a reasonably robust procedure. We also point out that the differences between the groups were
sufficiently great to make inferential procedures unnecessary.
11 We further subdivided this group into two groups, each representing one of two possible
differences (language or proficiency). The average number of pushdowns in the same language/
different proficiency group is 10.67 (sd = 3.79), while the average number in the different
language/same proficiency group is 11.25 (sd = 2.99). The difference between these two groups
is not significant This suggests that the important fact is that there is a difference. The kind
of difference is less important.
 EVANGELINE VARONIS AND SUSAN GASS 89

REFERENCES
Arthur, B., R. Weiner, M. Culver, Y. Lee and D. Thomas. 1980. The register of
impersonal discourse to foreigners: verbal adjustments to foreign accent' in
D. Larsen-Freeman (ed.). Discourse Analysis in Second Language Research.
Rowley, Mass.: Newbury House.
Brown, P. and S. Levinson. 1978. 'Universals in language usage: politeness pheno-
mena' in E. N. Goody (ed.). Questions and Politeness: Strategies in Social
Interaction. Cambridge: Cambridge University Press.
Chun, A., R. Day, N. A. Chenoweth and S. Luppescu. 1982. 'Types of errors

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

corrected in native-nonnative conversations.' TESOL Quarterly 16/4: 537-47.
Day, R., N. A. Chenoweth, A. Chun and S. Luppescu. 1984. 'Repair techniques in
native-non^iative conversation.' Language Learning 34/2: 19-45.
Gass, S. and E. Varonis. 1984. 'The effect of familiarity on the comprehensibility
of non-native speech.' Language Learning 34/1: 65-90.
Gass, S. and E. Varonis (in press). 'Negotiation of meaning in non-native/non-native
interactions-' in S. Gass and C. Madden (eds.). Input in Second Language Acquisi-
tion. Rowley, Mass.: Newbury House.
Grice, H. H. 1975. .'Logic and conversation' in P. Cole and J. L. Morgan (eds.).
Syntax and Semantics, Vol. 3. Speech Acts. New York: Academic Press.
Grosz, B. 19?7. 'The representation and use of focus in dialogue understanding.'
Technical Note #151, Artificial Intelligence Center, SRI International, Menlo
Park, California.
Grosz, B. 1979. 'Utterance and objective: issues in natural language communica-
tion.' Technical Note #185, Artificial Intelligence Center, SRI International,
Menlo Park, California.
Hobbs, J. and J. Robinson. 1978. 'Why ask?' Technical Note #169, Artificial
Intelligence Center, SRI International, Menlo Park, California.
Jefferson, G.- 1972. 'Side sequences' in D. Sudnow (ed.). Studies in Social Inter-
action. New York: Free Press.
Jones, E. E. and H. B. Gerard. 1967. Foundations of Social Psychology. New York:
Wiley.
Keller-Cohen, D. 1980. Language Development, Grammar and Semantics. Arling-
ton: Centre for Applied Linguistics.
Krashen, S. 1980. 'The input hypothesis' in J. E. Alatis (ed.). Current Issues in
Bilingual Education. Washington, D.C: Georgetown University Press.
Labov, W. and D. Fanshel. 1977. Therapeutic Discourse. New York: Academic Press.
Linde, C. and J. Goguen. 1978. 'Structure of planning discourse.' Journal of Social
Biological Structures 1:219-51.
Long, M. 1983a. 'linguistic and conversational adjustment to non-native speakers.'
Studies in Second Language Acquisition 5/2: 177-93.
Long, M. 1983b. 'Native speaker/non-native speaker -conversation and the negotia-
tion of comprehensible input.' Applied Linguistics 4/2: 126-41.
Marios, E. 1981. "Why answer? A goal-based analysis of a speech event' in D. San-
koff and H. Cedergren (eds.). Variation Omnibus. Carbondale, 111.: Linguistic
Research Inc.
Scarcella, R. and C. Higa. 1981. 'Input, negotiation and age differences in second
language acquisition.' Language Learning 31 /2: 409-37.
Schachter, J. (in press). 'Negative input' in W. Rutherford (ed.). Second Language
Acquisition and Language Universals. Amsterdam: John Benjamins.
Schegloff, E., G. Jefferson, and H. Sacks. 1977. The preference for self-correction
in the organization of repair in conversation.' Language 53/2: 361-82.
Schumann, J. 1975. 'Second Language Acquisition: the Pidginization Hypothesis.'
Ph.D. dissertation, Harvard University.
Schwartz, J. 1980. The negotiation for meaning: repair in conversations between
90 NON-NATIVE/NON-NATIVE CONVERSATIONS

second language learners of English' in D. Larsen-Freeman (ed.)- Discourse

Analysis in Second Language Research. Rowley, Mass.: Newbury House.
Stevick, E. 1976. Memory, Meaning and Method. Rowley, Mass.: Newbury House.
Stevick, E. 1980. Teaching Languages; A Way and Ways. Rowley, Mass.: Newbury
House.
Stevick, E. 1981. 'The Levertov machine' in R. Scarcella and S. Krashen (eds.).
Research in Second Language Acquisition. Rowley, Mass.: Newbury House.
Varonis, E. 1981. 'A Sociolinguistic Analysis of Cross-cultural Misunderstandings.'
Paper presented at N-WAVE X, Philadelphia.

Downloaded from https://academic.oup.com/applij/article/6/1/71/171724 by guest on 09 February 2021

Varonis, E. and S. Gass. 1982. 'The comprehensibility of non-native speech.'
Studies in Second Language Acquisition 4/2: 114-36.
Varonis, E. and S. Gass (in press). 'Miscommunication in NS/NNS Interactions.'
Language in Society 14/2.