Lang Soc. 2, 177-192. Printed in Great Britain
Towards a theory of interpersonal accommodation through
language: some Canadian data1HOWARD GILESUniversity of Bristol
DONALD M. TAYLOR AND RICHARD BOURHISMcGill
University2ABSTRACT
The study was designed to investigate the process of speech
accommodation between bilinguals from two ethnolinguistic groups.
It was hypothesized that the greater the amount of effort in
accommodation that a bilingual speaker of one group was perceived
to put into his message, the more favourably he would be perceived
by listeners from another ethnic group, and also the more effort
they in turn would put into accommodating back to him. Eighty
bilingual English-Canadians were divided into four groups and
individually tested. 5s heard on tape a French-Canadian describe a
picture and they were required to sketch this while listening. 5s
were made fully aware that their speaker had a choice of language
for his description. The four groups heard the same male speaker
describe the drawing but each in a different guise: (1) French, (2)
Mix-mix, (3) Fluent English and (4) Nonfluent English. 5s were then
requested to rate their speaker and his performance, and to record
a description of another picture themselves for that same
French-Canadian to draw later. From analyses of the ratings and the
5s' tapes, the hypotheses were confirmed and different types of
accommodation noted. A theoretical framework for these results and
other forms of interpersonal accommodation was suggested.
(Bilingualism; social interaction; accommodation theory; Canadian
French, Canadian English.) Much research has accrued recently which
demonstrates that an individual's speech patterns are in part
dependent on the person to whom he is talking, the topic of the
discourse and the setting in which it takes place (reviewed by
Giles &[1] The present investigation was supported by a grant
from the Canada Council. The authors wish to express their
gratitude to Marc Leroux for his excellent voice recordings and to
Professor Dell Hymes for his extremely useful comments on an
earlier version of this paper. An abridged version of this paper
was delivered at the Annual Conference of the British Psychological
Society in April 1972, at Nottingham. [2] Howard Giles and Richard
Bourhis are now at University College, Cardiff, South Wales,
U.K.
177
LANGUAGE IN SOCIETY
Powesland, in preparation). The bilingual's and bidialectal's
choice of language and dialect respectively have been shown to be a
function of these three factors (e.g. Herman 1961; Rubin 1962; Blom
& Gumperz 1972) as has the monolingual's choice of speech style
(e.g. Brown & Gilman i960; Labov 1966; Giles 1973). Indeed,
'Fishman, Hymes and Gumperz also stress the further point that both
bilingualism and diglossia are best not considered in isolation,
but rather as salient examples of a capacity for code variations
which is also found among monolinguals' (MacNamara 1967). It is,
according to Hymes (1972), this notion of code variation and speech
diversity that has been singled out as the hallmark of
sociolinguistics. The models of speech dynamics that have emerged
in sociolinguistics (ErvinTripp 1964; Hymes 1967) have relied on a
descriptive approach in terms of presenting a taxonomy of factors
influencing code variations, such as topic, setting and so forth.
This initial work has been extremely important since 'the work of
taxonomy is a necessary part of progress towards models (structural
and generative) of sociolinguistic description, formulation of
universal sets of features and relations, and explanatory theories'
(Hymes 1972). However, since our taxonomies have changed little
over the past few years, the present writers have felt the need to
develop a tentative, explanatory sociolinguistic theory to account
for at least certain specific types of speech diversities. The
strategy at this initial stage was to formulate a theory which
focused on one taxonomic level, in this case, the interpersonal
aspects of speech diversity. Most of the research at the
interpersonal level in multi- and bi-lingual societies has been
concerned with suggesting that social norms are operative in the
choice of a language or code depending on the characteristics of
the participants such as sex and status (e.g. Herman 1961;
Sechrest, Flores & Arellano 1968; Kimple, Cooper & Fishman
1969). Certain studies of monolinguistic code variation can also be
understood within a framework of normative behavior. For instance,
it has been shown that a speaker makes linguistic adjustments
depending on such perceived characteristics of the listener as
social status (Slobin, Miller & Porter 1968), sex (Benney,
Reisman & Star 1956), age (Granowsky & Krossner 1970) and
presumed knowledge about the conversational topic (Ratner &
Rice 1963). Therefore, certain shifts in speech style can be the
result of complying with social norms. In other words, it is
expected that we should speak in a particular manner to a certain
type of person. However, a great deal of work on monolingual speech
diversity at the interpersonal level has been more concerned with
demonstrating that often speakers adapt or accommodate their speech
towards that of their interlocutors when social norms in all
probability are not operative. Such accommodative behavior has been
termed 'response matching' by Argyle (1969) and may be a subtler,
and perhaps more unconscious speech shift than those hitherto
mentioned. This phenomenon has been demonstrated in a number of
studies on at least one member 178
INTERPERSONAL ACCOMMODATION THROUGH LANGUAGE
of an interactive dyad where a speaker tends to adopt the speech
patterns of the person to whom he is talking. It has been studied
on a range of linguistic levels including speech rate (Webb 1970),
vocal intensity (Black 1949), regional accent (Giles 1973), speech
durations (Matarazzo, Weins, Matarazzo & Saslow 1968) and
speech silences (Jaffe & Feldstein 1970). Other studies
(Lennard & Bernstein i960; Welkowitz & Feldstein 1970) have
shown that speech accommodation between members of a dyad can often
be a mutual process increasing each time the participants interact.
Unfortunately, however, little work has been conducted into the
process whereby members of entirely separate ethnolinguistic groups
attempt to accommodate to each other. Giles (1971; 1973) has
attempted to formulate a model for accent change in social
interaction and has termed such accommodative speech strategies
'convergent' behaviour. But he has pointed to the fact that
accommodative speech is unlikely to occur in all interpersonal
situations. For instance, Giles hypothesized that if a speaker does
not require this interlocutor's social approval and finds the
latter somewhat distasteful (maybe because of his personality or
attitudes), he may modify his accent in a direction opposed to that
of his partner 'divergence'. Little work has been conducted as yet
demonstrating the existence empirically of divergent speech
patterns either bi- or monolingually (Bourhis, Giles & Lambert
1972). Nevertheless, work on the language and code loyalty of a
minority group who retains its code as an expression of group or
national identity in the face of the majority culture's language
could be regarded as a form of divergent behavior (e.g. Gumperz
1964; Fishman et al. 1966). Thus, one may identify at least three
forms of speech modifications in interpersonal situations; namely
normative, accommodative and divergent code variations. But it is
clear that even this may be too broad an area to initiate
model-building at this stage, and so our emphasis will be on only
one form of interpersonal code shift - accommodation. The aims of
this paper then are twofold. First, we shall attempt to provide a
theory for interpersonal accommodation in speech hoping that, like
the previous taxonomies, it may be flexible enough to cope with
both mono- and bilingual accommodation. Secondly, as part of
ongoing research into the theory, another aim of the paper was to
test it in one very specific bicultural setting with its own
peculiar history - Quebec, Canada, The essence of the theory lies
in the social psychological research on similarityattraction (see
review of Simons, Berkowitz & Meyer 1970). This work suggests
that an individual can induce another to evaluate him more
favorably by reducing certain dissimilarities between them. The
process of speech accommodation of course operates on this
principle and as such may be a reflection of an individual's desire
for social approval. In exchange theory terms (Homans 1961), it
seems likely that the accommodative act may involve certain costs
for the speaker, that is in terms of identity change and expended
effort, and so such behavior may only 179
LANGUAGE IN SOCIETY
be initiated if potential rewards are available. Thus, if one
can accept the notion that people find social approval from others
rewarding, it would not seem unreasonable to suppose that there may
be a general set to accommodate to others in most social
situations. This, set is of course not insensitive to the demands
of the specific occasion as can be inferred from the reference to
speech divergence earlier. Moreover, it can be proposed that the
amount of accommodation a person exhibits may in part be a function
of the strength of his need for social approval from the other.
Many other factors could affect the intensity of this need
including the probability of future interactions, the extent of the
accommodatee's social power and the perception of prior
accommodation from that person. Nevertheless, not all examples of
response matching (or convergence) can be understood within this
accommodation framework and hence it seems important to consider a
distinction made by Giles (1971) between 'positive' and 'negative'
response matching. The studies cited so far have all been examples
of the former and may be explainable in terms of the model
proposed. Negative response matching on the other hand, was the
term introduced to denote certain types of modelling which appear
more plausibly explainable within a fabric of social retaliation.
Such behaviour may be exemplified in situations where one person
reciprocates the other's use of interruptions (Argyle & Kendon
1967) and verbal aggression (Mosher, Mortimer & Grebel 1968).
As previously stated, little work has been conducted into the
process whereby members of entirely separate ethnolinguistic groups
attempt to accommodate to each other. What then are the likely
speech modifications or adjustments (if any) that a person from one
language background will make when interacting with a
representative from another linguistic milieu? Is it likely that
bilinguals from one ethnic group will use both languages when
interacting with bilinguals of another ethnolinguistic group? Or
alternatively, must accommodation with regard to bilingual behavior
be an all-or-none process? If the latter were to be the case, is it
possible that an English-Canadian (EC) bilingual who had just been
provided with information in English by a French-Canadian (FC)
bilingual would accommodate back to that person by replying in his
second language (French)? The present empirical study was designed
in an attempt to answer these questions in the context of a
bicultural setting, Quebec, and also to test certain assumptions of
the general accommodation model suggested above. These assumptions
relate to the listener's evaluations of the accommodator and also
to a greater desire to accommodate to those who have previously
shown willingness to accommodate to oneself. It must be stressed
that the authors are fully aware of the fact that the study can in
no way be thought of as a valid test in itself of the theory.
Rather, the study is conceived of as an attempt to establish
preliminary support for hypotheses derived from the theory in one
of many possible cultural contexts available. The following
hypotheses were proposed: 180
INTERPERSONAL ACCOMMODATION THROUGH LANGUAGE
I. The greater the amount of effort in accommodation that a FC
bilingual is perceived to put into his message, the more favorably
he will be perceived by EC bilingual listeners. II. When
accommodation by a FC speaker is perceived by EC listeners, the
latter will respond in an accommodating manner. III. The greater
the amount of effort that a FC speaker is perceived to put into
accommodating towards an EC bilingual listener, the more effort he
in turn will put into accommodating back to him.METHOD
Subjects
The Ss for this experiment consisted of 80 ECs (43 female and 37
male) who claimed to possess at least a 'working knowledge' of
French. Ss were students at McGill University, Montreal, and
volunteered for the study thinking it was concerned with the
effectiveness of communication between French- and
EnglishCanadians. Their self-evaluation scores for their own French
skills (using 9-point rating scales where 1 indicated native usage)
were 4.21 for comprehension and 4.95 for speaking fluency. 5s were
randomly assigned to one of four experimental conditions while at
the same time equating the subgroups for sex. There were no
significant differences (one-way analyses of variances) between the
four groups in their self-rated French skills. Materials The
materials for this experiment included two scenic pictures, paper
and pencil, tape-recorded descriptions, tape-recording equipment
and series of rating scales. Tape recordings. A 380-word English
description of a very simple harbour scene was composed in such a
manner that another person could draw it while listening to the
description; verbal repetitions of the objects in the picture and
their positions were abundant. A male FC bilingual student
tape-recorded four versions of this passage. (1) Totally in French
(duration 2 mins 47 sec). (2) Mixed French and English (duration 2
mins 56 sec). To make this mixture sound realistic, the body of the
passage was spoken in French whilst the repetitions were in
English; English accounted for about 1/3 of the passage length. (3)
Fluent English (duration 3 mins 12 sees). The description was
spoken in fluent, grammatically-perfect English but with a distinct
FC accent. (4) Non-fluent English (duration 5 mins 10 sees). This
passage was produced by the speaker as though he was not really
fluent in English and it contained many filled and unfilled pauses,
speech disturbances and a few grammatical errors (e.g. misuse of
tense markers). Nevertheless, the description was fully
comprehensible. It was thought that descriptions (i)-(4) would be
considered by EC bilinguals as reflecting a series of messages
increasing with regard to their effort in perceived accommodation.
Each of these four versions was preceded on the tapes by exactly
the same 40181
LANGUAGE IN SOCIETY
second recording. In this extract, the 5 apparently hears the FC
speaker at PUniversite de Montreal (a French-speaking institution)
being given his instruction for describing the picture by the
experimenter. The substance of this predescription recording was as
follows. The speaker could be heard being told (in French) that his
recipient would be an EC bilingual. The FC was then heard to
enquire of the experimenter which language he should use for his
audience. The experimenter asked (in English) whether he could
speak English - the reply was affirmative (also in English) - and
then told him he could speak in the language of his choice. The
purpose of this procedure was to ensure that EC listeners were made
fully aware of certain features of the situation even though these
were made explicit and stressed in the listening instructions.
These features were (a) that the 5s' speaker was known to be a FC
bilingual, (b) that 5s were aware of the fact that their speaker
knew that they themselves were bilingual, and (c) that 5s were
aware that their speaker's language form on tape was the result of
a conscious choice on the part of the FC. Rating scales. There were
three separate sets of scales used in the study, all of which
consisted of nine rating points. Four pre-instruction scales were
administered to 5s on arrival in the testing situation. These were
concerned with 5s' reported skills in speaking and understanding
both French and English. Seventeen post-drawing scales were used
and these were concerned with the Ss' assessments of the FC speaker
and his descriptive performance. They related to (a) the quality of
the speaker's description, (b) ease in reproducing the picture, (c)
task enjoyment, (d) likelihood that the speaker would be a friend,
(e) desire to have the same speaker as a future experimental
partner, (f) speaker's attachment to FC values, (g) speaker's
effort in bridging the cultural gap between FCs and ECs. From the
data of a study by Aboud and Taylor (1971), three scales strongly
stereotypical each of FCs (colourful, emotional and talkative) and
ECs (logical, stable and egotistic) were used such that 5s were
required to rate to what extent these traits represented their
particular speaker. In addition, 5s were required to rate how
likeable, friendly, considerate and narrow-minded they considered
their speaker to be. 5s were also given an opportunity of providing
any further comments they might care to make. Finally, 5s were
required on seven post-speaking scales to rate how they felt in the
role of the speaker. They were asked to rate how much they enjoyed
giving their description and to what extent they had felt
unsociable, relaxed, uncomfortable, happy, active, weak and
unfriendly when recording their description. In addition, 5s were
asked to state which languages they had used and why they had
chosen them. Again, 5s were asked if they would like to make any
further comments. 182
INTERPERSONAL ACCOMMODATION THROUGH LANGUAGE
Procedure
The testing procedure lasted about 25 minutes per 5 and was
conducted in English by a FC experimenter. All 5s were thus
individually-tested and before they were read the standard
instructions, they were required to rate their ability in French
and English on sheets provided. 5s were told that it had been hoped
that a face-to-face situation could have been created but because
of the enormous difficulties involved in transporting FCs to McGill
University and EC students to l'Universitd de Montreal at
convenient times for all, taped messages were being used. They were
told that bilingual FC students had been recorded at PUniversite de
Montreal in which they had been describing a picture for bilingual
EC students. 5s were told that they would each hear one of these FC
bilinguals, and from his taped description were expected to draw
the picture he was talking about. Later, they were informed, they
would be given the opportunity of recording a description for that
same FC to draw. 5s were assured that the quality of their drawing
was completely unimportant as we just wanted to see if the message
had been effectively communicated. 5s were then provided with paper
and pencil and told to draw the picture as the speaker was
describing it. The experimenter turned on the tape recorder when 5s
were ready, but apparently did so a little too far back on the tape
(and apologized) as the FC speaker was heard receiving his
instructions in French. During this 'mistake', 5s heard the FC
speak English and be given the opportunity of choosing the language
in which to describe the picture for an EC bilingual. Members of
any one subject group heard the same description by the same male
FC stooge but in one of the four different guises mentioned above.
After drawing the picture from the F C s description, 5s were
presented with a questionnaire on which they were required to rate
their reactions to their speaker and his performance. The 5s were
then presented with a similar picture to the one they had just
drawn and asked to describe this one for the same FC bilingual whom
they were told would draw it the following week; 5s were handed a
microphone and were then tape-recorded giving their description.
This completed, 5s were asked to fill in another short
questionnaire on which they were required to give reasons for
choosing the language(s) they had, and also to rate how they felt
giving their message.RESULTS
In order to determine whether perceived effort in accommodation
by the FC speaker affected listeners' evaluations of him and how
they felt talking to him, one-way analyses of variance were
computed from the questionnaire data. Very few of the 5s' ratings
yielded significant trends, but those that did can be found in
Table 1. It can be seen from the mean ratings that a speaker's
perceived considerateness and effort in bridging the cultural gap
are a function of his perceived effort in accommodation. Table 1
also shows that 5s consider the mixed 183
LANGUAGE IN SOCIETY
French and English situation to be the most difficult decoding
and encoding condition. As expected, 5s in the four conditions did
not differ in their selfrated English language skills, nor as
reported earlier, in their French skills. Fourteen different
categories of speech accommodation were identified from the ECs'
tape-recorded messages and these appear together with their
frequencies of occurrence under the four conditions in Table 2.
From this table, it can be seen that various amounts of speaking
the two languages did in fact occurTABLE 1. Mean ratings of Ss and
their F values on four scales Rating scales I N = 80 Stimulus
conditions MixFluent French English mix(n = 20) (n = 20) (n =
20)
Nonfluent English(n = 20)
F values (df = 3, 76)
Ease in reproducing drawing Effort in bridging culturalgap made
by FC speaker FCs' considerateness
5-93 1.30 1.67 2.16
6.20 1.70
2.852.15 2.00
4.33.65 2.90
13.04** 9.71** 3-22* 4.23*
ECs' relaxation when giving description to FC
i-95 3.65
1.80
3.65
The higher the mean ratings the more ease, effort,
considerateness and relaxation was felt by 5s. **p