The phenomenon of speech accommodation in dialogues has been studied and introduced into the framework of Speech Accomodation Theory (SAT) over two decades ago (Giles et al. 1987). This framework, that was later renamed Communication Accomodation Theory (CAT), proposes that accommodation of speech features (accent, speed, pause duration, lexical) occurs as a communication strategy (either conscious or unconscious), with specific social goals (integration into a social group, or identification with a member of the same group). In this section, a summary of the main ideas of SAT (and CAT) is given.
3.4.1 Convergence and divergence
SAT defined convergence as a linguistic strategy. Convergence refers to adaptation of an individual’s speech characteristics (pause duration, speech rate, utterance length, accent, etc.) in order to match those of a partner in dialogue more closely. Similarly, divergence refers to a tendency of the individual to maintain their distinct speech style by accentuating differences in the aforementioned characteristics of speech. SAT distinguishes between upward and downward
convergence (or divergence), the former refering to changing one’s style in order to match a valued social status profile, while the latter suggesting ‘shifting’ towards less valued social profiles, such as a language variant specific to ethnic/cultural/social groups or non-fluent speech/illiteracy.
Futher, SAT proposes further categorizations of convergence (and divergence) by relative movements or ‘shifts’ between two interlocutors (A,B). Thus convergence or divergence can be mutual (A B, A B ) or non-mutual ( A B, A B ), or one speaker might converge while the partner diverges ( A B).
Another distinction is introduced by the difference that lies between a manifest speech style and the
perception of that speech style that is biased by a stereotypical belief. (Giles et al. 1987) pointed out
that both convergence (or divergence) and its evaluation (how positively or negatively it is perceived) depend on one’s perception of the other’s speech, rather than their actual, manifest speech styles. A common example is imitation of a language variant by non-native speakers (such as a Dublin accent in Irish English): although characteristics of that accent might be prominent in the native speaker’s manifest speech style, they might be perceived as accentuated by the non-native speaker, therefore misleading them to converge towards a similar speech style. From the point of view of the native speaker, that might be perceived as mocking of their social group, or as a comical social integration attempt at best.
Convergence (or divergence) can be additionaly distinguished into total and partial. The former refers to near absolute matching of speech style metrics, e.g. two interlocutors exhibiting very similar speech rate. The latter signifies a clear movement towards matching, such as increasing one’s speech rate in order to converge to a higher rate of the interlocutor, but not to the extent of matching that speech rate.
Finally, convergence and divergence can be either unimodal, when accomodation occurs along only one characteristic of speech (such as speech rate, or accent), or multimodal, when two or more speech characteristics converge.
The central idea of SAT is that convergence (and divergence) is a strategy that humans engaging in dialogue use (either consciously or unconsciously) in order to achieve specific goals. In the landmark study of SAT (Giles et al. 1987), three such goals are proposed: social approval by the listener, serving communicational efficiency, and maintaining a positive social identity.
3.4.2 Communicative function of convergence
utilized to invoke similarity attraction by the listener. The latter is a sociological principle (Giles et
al. 1987), which states that attraction is more likely to occur towards individuals that display
similarity in behavior. Reduction of dissimilarities of dialects, convergence of native speakers towards grammatical errors of non-fluent speakers, interviewees adjusting their speech to match the style of the interviewer, and sales people matching speech styles of their customers, are only a few examples of such cases given in (Giles et al. 1987).
According to SAT, convergence will not always be the best communication strategy, as its effect is moderated by ‘situational constraints’. Such constraints are introduced by ‘sociolinguistic norms’ or, in simpler terms, what people believe ‘is right’ in a given situation. An example given in (Giles
et al. 1987) is that of interviews in Australian English, were interviewees were rated higher if they
were using a ‘refined’ rather than a ‘broad’ accent, regardless of the accent of the interviewer (who was switching accents between interviews). In this case, therefore, convergence of the interviewee towards the interviewer’s accent was not rated favorably. Also, “powerful” speech style was more often rated favorably as a response to a “powerless” speech style, whereas convergence to a ‘powerless’ speech style was more often negatively evaluated.
SAT further advocates that the evaluation of convergence of the interlocutor to an individual’s speech pattern is largely dependant on causal attribution. Listeners tend to evaluate the effort on the part of a converging speaker favorably, when they attribute that effort to the speaker’s desire for social integration and attraction. When convergence is forced by situational constraints, it is rated less favorably. As pointed out in (Giles et al. 1987), although SAT defines speech accommodation as a strategy, that does not necessarily mean that it is a conscious one. Rather, (Giles et al. 1987) points to evidence of spontaneity and autonomy for speech accommodation at various cognitive levels. SAT advocates that speech accommodation may well be “scripted” behavior (established behavioural routines) in many cases, but one can be simultaneously making conscious decisions on the appropriate choice of speech style.
3.4.3 Communication function of divergence
Similarly to convergence, SAT proposes a number of communication goals for divergence: the main goal proposed is social identity maintenance, or the desire of individuals or groups to maintain a positive social identity, cultural pride and distinctiveness. A series of studies in the review of (Giles et al. 1987) provides many examples of ethnic minority group members accentuating their distinct dialects or accents when their ethnic identity is made more salient, or they encounter ethnically “threatening” situations. Gender is also proposed as a socially identifying factor as, in
one of the studies reviewed, men talking to women were found to sound “more masculine” when their gender was made more salient.
(Giles et al. 1987) points out that a distinction between non-convergence and divergence would be superficial or unnecessary at best. Non-convergence is a passive behavior towards the dialogue and the interlocutor, and its most extreme form of intended accentuation of distinct speech features has been termed as speech dis-accommodation. Divergence (by definition) means shifting one’s speech style away from that of an interlocutor. According to (Giles et al. 1987), it is more likely that causal attribution plays a key role in the evaluation of divergence (similarly to convergence). After all, non-convergence may well be the result of repertoire constraints (as in the case of non-native speakers) or individual personality factors.
Another communicative function for divergence proposed by SAT is that of cognitive organization, i.e. to put the interaction (dialogue) in order, or to provide a ‘mutual basis for communication’. A series of studies reviewed in (Giles et al. 1987) provides various examples of this communicative behavior: speakers who are unfamiliar with the host social group or the situational context, tend to accentuate their accent or employ other divergent strategies in order to indicate their unfamiliarity. The expected result of this is tolerance on the part of the host community members towards violations of situational norms on the part of the speaker.
Another example given is that of speakers diverging from a speech style that is uncomfortable for them, in order to encourage the interlocutor to converge to a different speech style, such as when talking slowly in an attempt to “cool down” a rapidly speaking interlocutor. Similar examples include therapy sessions, where clients may be invited to talk more when therapists talk less.
In certain situations, dissimilarities in the interlocutors’ speech styles are expected, as is the case with interviewers and interviewees where the latter were more positively evaluated, when maintaining their ‘refined’ accent as opposed to converging downwards to the ‘broad’ accent exhibited by the former.
Finally, there are social norms that indicate a pattern of interaction where the interlocutors are expected to ‘complement’ each other’s speech. This is more often made obvious in interactions between doctors and patients, teachers and pupils, parents and children and so on. As pointed out in (Giles et al. 1987), this complementary nature of speech patterns does not exclude the possibility of simultaneous convergence, along a different dimension (such as speech rate). The complexity (and multi-modality) of convergence and divergence are also highlighted in the text:
diverge, and complement each other with regard to various verbal, vocal, and non-verbal forms simultaneously”.
In conclusion, SAT provides a theoretical framework that attributes communicative functions to convergence and divergence of speech style. Speech style is used as a broad term and can denote anything from speech rate and pause duration to choice of words, utterance length, accent, dialect, and even switching languages (in bilingual communities). Of particular interest are the definitions for mutual/non-mutual, partial/total, unidirectional/bidirectional and unimodal/multimodal convergence (or divergence). Additionally, SAT proposes that a genuine effort to converge to another’s speech style is likely to be evaluated positively, if the situational constraints do not suggest otherwise.