Convergence, Divergence, and Maintenance

Early CAT research focused primarily on objective speech variables and identified three basic ways in which people can adjust their communicative behaviors relative to one another: convergence, divergence, and maintenance (for a brief history of the historical development of CAT, see the Foreword, this volume). Convergence refers to adjusting one’s communicative behaviors to be more similar to another’s. Convergence has been studied extensively in laboratory as well as naturalistic settings. Recently, researchers have increasingly begun to examine convergence in mediated and online environments as well (e.g., Riordan, Markman, & Stewart, Reference Riordan, Markman and Stewart2013). For instance, Danescu-Niculescu-Mizil, Gamon, and Dumais (Reference Danescu-Niculescu-Mizil, Gamon and Dumais2011) investigated adjustment on Twitter and found that users regularly converged to one another in their tweets on a wide range of linguistic features, despite the limitations posed by this particular social medium. Interestingly, people have been shown to converge not only toward human, but also toward computer conversational partners (but see Beckner, Ràcz, Hay, Brandstetter, & Bartneck, Reference Beckner, Ràcz, Hay, Brandstetter and Bartneck2016). For instance, when interacting with computer-animated personas, children have been shown to converge in terms of both speech amplitude (Coulston, Oviatt, & Darves, Reference Coulston, Oviatt, Darves, Hansen and Pellom2002) and response latencies (Darves & Oviatt, Reference Darves, Oviatt, Hansen and Pellom2002).

Divergence refers to adjusting one’s communicative behaviors to be more dissimilar to another’s. For example, Bourhis and Giles (Reference Bourhis, Giles and Giles1977) found that when an English speaker described Welsh as a “dying language with a dismal future,” Welsh participants overwhelmingly broadened their Welsh accents and some even introduced Welsh vocabulary into their responses. Finally, maintenance refers to sustaining one’s “default” level of communicating, without adjusting for others. For example, Bourhis (Reference Bourhis1984) had a female confederate approach bilingual (English-French) pedestrians in downtown Montreal (Quebec) to ask for directions in either English or French. When the request was voiced in French, nearly half of the Anglophone pedestrians nonetheless responded in English (i.e., maintenance of default language).

Convergence and divergence can each take multiple forms (Gallois & Giles, Reference Gallois, Giles, Palmer and Barnett1998), depending on the social value, degree, symmetry, modality, and duration of the behavior. Below we discuss each of these distinctions.

Upward/Downward. When the dimension of adjustment has some social value, adjustment can be conceptualized as upward or downward (Giles & Powesland, Reference Giles and Powesland1975). Upward adjustment refers to shifts toward a more prestigious variety of speech, whereas downward adjustment refers to shifts toward a less prestigious, or even stigmatized, variety. For instance, so-called standard accents (e.g., Standard American English) are typically judged as more prestigious than “nonstandard” accents, which include most regional (e.g., southern accent in the U.S.) ethnic (e.g., African-American Vernacular English) and foreign varieties in a given society (e.g., Spanish accent in the U.S.) (Dragojevic, Giles, & Watson, Reference Dragojevic, Giles, Watson, Giles and Watson2013). Accordingly, a nonstandard speaker matching another’s standard accent is an example of upward convergence, whereas a standard speaker matching another’s nonstandard accent is an example of downward convergence (see Willemyns, Gallois, Callan, & Pittam, Reference Willemyns, Gallois, Callan and Pittam1997). Conversely, accentuating one’s own nonstandard accent with a standard-accented speaker is an example of downward divergence, whereas adopting a standard-accent with a nonstandard-accented speaker is an example of upward divergence.

Full/Partial. Adjustment can also be described as either full or partial (Bradac, Mulac, & House, Reference Bradac, Mulac and House1988). For instance, a speaker initially exhibiting a rate of 100 words per minute may increase his speed to match exactly another speaker’s rate of 200 words per minute (full convergence) or may increase his rate to 150 words per minute to only partially match her rate (Street, Reference Street1982). Similarly, interactants may diverge from one another to varying degrees, ranging from partial (e.g., code-switching for a few words) to full divergence (e.g., speaking an entirely different language).

Symmetrical/Asymmetrical. Sometimes adjustment is symmetrical and one person’s communicative moves are reciprocated by the other. For instance, Nelson, Dickson, & Hargie (Reference Nelson, Dickson and Hargie2003) reported how both Catholic and Protestant children in Northern Ireland avoided sensitive topics (e.g., religion, politics) during interreligious conversations, citing this as a way to avoid conflict and promote group harmony. At other times, however, one person’s communicative moves are not reciprocated by the other. Indeed, convergence is often directed toward those with greater power (without reciprocation by the high-power speaker); such shifts tend to asymmetrical. For example, van den Berg (Reference van den Berg1986) noted that salespersons in Taiwan were more likely to converge to shoppers than vice versa. A similar pattern of asymmetrical accommodation occurs frequently in male–female interactions, where women converge to men more often than men converge to women (e.g., Namy, Nygaard, & Sauerteig, Reference Namy, Nygaard and Saureteig2002). However, and as discussed later, in such cases asymmetrical accommodation may more accurately be described as “complementarity,” and be perceived positively by both parties (Giles, Reference Giles and de Silva1980).

Unimodal/Multimodal. Adjustment on some communicative features does not necessarily mean the speaker will adjust on all available variables and dimensions. Accordingly, CAT distinguishes between unimodal and multimodal adjustments. The former refers to shifts on only a single dimension (e.g., accent) whereas the latter refers to shifts on multiple dimensions simultaneously (e.g., accent, posture, eye gaze). For example, in the Bourhis and Giles (Reference Bourhis, Giles and Giles1977) study described earlier, Welsh participants who responded to the threatening English person by only broadening their Welsh accents were engaging in unimodal divergence, whereas those who broadened their Welsh accents as well as introduced Welsh vocabulary into their responses were engaging in multimodal divergence.

Owing to the fact that adjustment can take place on multiple dimensions, convergence and divergence are not mutually exclusive strategies and both may be enacted simultaneously (Gallois, Ogay, & Giles, Reference Gallois, Ogay, Giles and Gudykunst2005). For example, Bilous and Krauss (Reference Bilous and Krauss1988) found that women converged toward men’s utterance length, interruptions, and pauses, but diverged on backchannels and laughter. Relatedly, Zilles and King (Reference Zilles and King2005) showed how immigrant German women in Brazil simultaneously accommodated to host language features and emphasized their Germanic linguistic origins.

Short-term/Long-term. Adjustment can also vary in its duration. Sometimes adjustment toward a particular style is short-lived and occurs during only one or a few social interactions (short-term). Other times, adjustment toward a particular style is more sustained and occurs repeatedly over multiple interactions (long-term). For instance, Pardo, Gibbons, Suppes, and Krauss (Reference Pardo, Gibbons, Suppes and Krauss2012) examined accommodation among unacquainted male roommates over the course of an academic year and found that mutual convergence not only increased over that period but also was resistant to decay across breaks in exposure (see also Sancier & Fowler, Reference Sancier and Fowler1997).

The distinction between short- and long-term accommodation has been particularly useful in explaining dialect change (Trudgill, Reference Trudgill, Masek, Hendrick and Miller1981, Reference Trudgill1986). Specifically, whereas short-term accommodation toward a particular style may lead to transitory changes in a person’s habitual speech, long-term accommodation toward that style may ultimately result in permanent changes to a person’s speech. For instance, a young Russian-accented immigrant’s repeated convergence to a Californian accent may, over time, permanently change his or her habitual accent so that it becomes indistinguishable from other Californians. A similar process underlies community-level dialect change, wherein regional minorities typically engage in long-term accommodation to the language style of the majority (Nilsson, Reference Nilsson2015; Trudgill, Reference Trudgill1986). In this sense, then, (long-term) accommodation is a basic mechanism underlying language change.

Psychological, Subjective, and Objective Accommodation

Sometimes speakers’ motives and intentions to accommodate and their actual communicative behaviors are congruent. At other times, however, they are incongruent. To account for this potential discrepancy, Thakerar, Giles, and Cheshire (Reference Thakerar, Giles, Cheshire, Fraser and Scherer1982) distinguished between psychological accommodation – that is, speakers’ motives and intentions to adjust their communication – and linguistic accommodation – that is, speakers’ actual speech behavior. For example, in many role-discrepant situations, dissimilarities are not only acceptable but also expected (Grush, Clore, & Costin, Reference Grush, Clore and Costin1975). Thus, a job interviewee wishing to accommodate to her interviewer (i.e., psychological convergence) may do so by not assuming the directive, interrogative language of the interviewer (i.e., linguistic convergence), but rather by crafting a more tempered and cooperative communicative style (i.e., linguistic divergence). Similarly, psychological divergence can sometimes be enacted precisely through linguistic convergence. In this vein, Woolard (Reference Woolard1989) reported that when Castilian speakers converged to Catalan during conversations with Catalan speakers, they received replies in Castilian. Although both Castilian and Catalan speakers converged to one another in their respective choices of language, Catalan speakers’ convergent behavior (i.e., switching to Castilian) actually represented psychological divergence in an attempt to emphasize intergroup differences and boundaries (i.e., not allowing an outgroup to use Catalan).

Thakerar et al. (Reference Thakerar, Giles, Cheshire, Fraser and Scherer1982) further distinguished linguistic accommodation as being objective – that is, directly observable or measurable shifts in behavior – and/or subjective – that is, individuals’ perceptions of behavioral shifts. Like its psychological and linguistic counterparts, objective and subjective accommodation are not always aligned. For example, speakers may perceive their behavior as convergent when, in fact, it is objectively divergent. In this vein, Thakerar et al. (Reference Thakerar, Giles, Cheshire, Fraser and Scherer1982) observed that, in dyads characterized by status inequality, high-status participants slowed their speech rates and made their accents less standard, whereas low-status speakers increased their speech rates and made their accents more standard. Although both were objectively diverging from one another, each perceived that they were converging.

In the previous example, both high- and low-status speakers were likely adjusting their communication to their (status-based) expectations of their conversational partners. Indeed, people often adjust their communication toward where they believe others are communicatively, rather than were they actually are (Thakerar et al., Reference Thakerar, Giles, Cheshire, Fraser and Scherer1982). Although sometimes people’s expectations of how others will behave and their actual behaviors are one and the same, other times they may be incongruent. Such erroneous expectations are especially likely to occur during intergroup encounters – that is, situations in which people define one another primarily in terms of their social identities (i.e., social group memberships) rather than their personal identities (i.e., idiosyncratic characteristics) – because social categorization depersonalizes people’s perceptions of others and leads to stereotyped expectations (Dragojevic & Giles, Reference Dragojevic, Giles and Berger2014; Hogg & Reid, Reference Hogg and Reid2006). Expectations based on stereotypes can lead speakers to overadjust (i.e., overaccommodate) or not adjust sufficiently (i.e., underaccommodate) their communicative behaviors relative to their interlocutors (see Chapter 5, this volume). For instance, Bayard (Reference Bayard1995) found that women and men swore at similar rates during intra gender conversations, but that women swore more than men (i.e., overaccommodated) during inter gender conversations, presumably because they expected (in this case erroneously) men to swear more than women.

Accommodation Strategies

Adjustment can also be conceptualized in terms of its focus or goal relative to a conversational partner’s needs and characteristics (Coupland, Coupland, Giles, & Henwood, Reference Coupland, Coupland, Giles and Henwood1988), in at least five ways. First, when interactants focus their attention on their partners’ productive language and communication, they can employ approximation strategies, which involve (as earlier) adjusting their verbal and nonverbal behaviors toward (convergence) or away from (divergence) their interlocutor (see Giles & Wadleigh, Reference Giles, Wadleigh, Guerrero and Hecht2008; McGlone & Giles, Reference McGlone, Giles, Knapp and Daly2011). Most CAT research has focused on these strategies. Second, when interactants focus on their partners’ ability to comprehend what is being said, they can employ interpretability strategies, such as decreasing the diversity of their vocabulary, simplifying syntax, or becoming louder in order to increase clarity and comprehension. Third, when speakers are focused on their partners’ macro-conversational needs, they can employ discourse management strategies. These include regulating speaking turns and selecting or selecting conversational topics of mutual interest or concern. Fourth, when speakers are focused on role relationships within an interaction, they may adopt interpersonal control strategies, such as the use of interruptions or honorifics, to remind the partner of their relative status or role. Fifth and finally, when speakers are concerned about another’s feelings, they can employ emotional expressions, such as conveying reassurance and comfort (see Williams, Giles, Coupland, Dalby, & Manasse, Reference Williams, Giles, Coupland, Dalby and Manasse1990; Watson, Angus, Gore, & Farmer, Reference Watson, Angus, Gore and Farmer2015). Just as speakers can converge and diverge on different dimensions at the same time, speakers can adopt multiple strategies simultaneously – for example, one could simplify an explanation to aid interpretability and to remind a subordinate of their social position – and what goals or characteristics speakers attend to may vary over the course of an interaction (Gallois et al., Reference Gallois, Ogay, Giles and Gudykunst2005; Jain & Krieger, Reference Jain and Krieger2011).

Affective Motives

CAT is premised on the assumption that communication conveys not only referential, but also social and relational information. CAT also assumes, per social identity theory (SIT: Tajfel & Turner, Reference Tajfel, Turner, Worchel and Austin1986), that the self-concept consists of personal (i.e., idiosyncratic characteristics) and social (i.e., social group memberships) identity components, and that people want to create and maintain positive personal and social identities. Following from this, CAT posits that speakers can pursue positive personal and social identities by communicatively regulating social distance and, thus, signaling their attitudes toward each other as individuals and group members.

Cooperative accommodation (including convergence) is motivated by a desire for social approval from one’s interlocutors, as a means to positively reinforce one’s own personal and/or social identity. Following the similarity-attraction paradigm (Byrne, Reference Byrne1971; see also, Sprecher, Reference Dorjee, Giles and Barker2014), CAT posits that speakers can increase personal and social liking and gain others’ social approval by becoming communicatively more similar to them (i.e., converging) (see Wang & Fussell, Reference Wang and Fussell2010). For example, speakers may converge to their interlocutors’ idiosyncratic communicative behaviors (e.g., speech rate, gestures) so as to appear more similar to them and thus engender liking. Indeed, Natalé (Reference Natalé1975) found that speakers with a high need for social approval converged to their conversational partners’ vocal intensity and pause length to a greater extent than did those with a low need for social approval. That said, accommodative moves may also be fashioned by disingenuous motives, such as the desire to exploit one’s interlocutor (see Giles, Ota, & Foley, Reference Dragojevic, Giles, Watson, Giles and Watson2013). Speakers may also converge to their interlocutors’ socially marked communicative behaviors (e.g., accent, dialect) to signal that they belong to the same social group and, thus, secure potential social reward (cf. Tajfel & Turner, Reference Tajfel, Turner, Worchel and Austin1986). For example, Tamburrini, Cinnirella, Jansen, and Bryden (Reference Tamburrini, Cinnirella, Janse and Bryden2015) found that Twitter users converged to the language style of other members who belonged the same online social communities (e.g., Twilight fans), presumably to indicate common ingroup membership.

Non-cooperative accommodation (including divergence and maintenance) is generally motivated by a desire to emphasize distinctiveness from one’s interlocutors, as a means to differentiate oneself from relevant outgroups and positively reinforce one’s own personal and/or social identity (Giles, Coupland, & Coupland, Reference Giles, Coupland, Coupland, Giles, Coupland and Coupland1991). For example, Berger and Heath (Reference Berger and Heath2008) noted that people often diverged in terms of clothing and apparel from select others to avoid signaling socially undesirable group identities (e.g., geek). Additionally, members of ethnic and social minorities may emphasize features of their own (perhaps stigmatized) dialects when they become aware and proud of their cultural identity, as did the Welsh speakers in response to an English person derogating their language described earlier (Bourhis & Giles, Reference Bourhis, Giles and Giles1977). Speakers may also diverge from their interlocutors’ idiosyncratic communicative behaviors as a way to signal their disapproval of others as individuals. For instance, Putnam and Street (Reference Putnam and Street1984) found that when interviewees were instructed to act out being dislikeable, they diverged from their interviewers on a variety of non-content speech features.

Although these distinctions and patterns appear straightforward, the actual dynamics of conversation are often far more complex. People belong to many different social groups and, in a given situation, are likely to share some (e.g., ethnicity, age) but not all of these identities (e.g., gender) with their interlocutors. Furthermore, these different identities are likely to vary in salience across different encounters, as well as at different points within the same encounter, with accommodative moves following accordingly. For instance, Jones, Gallois, Barker, and Callan (Reference Jones, Gallois, Barker and Callan1994) found that, in an academic setting, ethnic group membership did not predict communicative behavior, but professional group membership did.

Cognitive Motives

In addition to the identity maintenance concerns outlined earlier, CAT posits that communicative adjustment may also be motivated by a desire to regulate comprehension and increase communicative efficiency (Thakerar et al., Reference Thakerar, Giles, Cheshire, Fraser and Scherer1982). Motivated as such, speakers can assess their interlocutors’ communicative needs and characteristics, and adjust their speech to be more (or less) intelligible, predictable, and comprehensible. Indeed, converging to a common linguistic style often improves communicative effectiveness and has been associated with increased predictability of the other and, in turn, reduced uncertainty, interpersonal anxiety, and increased mutual understanding (e.g., Gudykunst, Reference Gudykunst and Wiseman1995).

Comprehension can also be facilitated through divergent shifts (Street & Giles, Reference Street, Giles, Roloff and Berger1982). For instance, speakers may diverge from their interlocutors to encourage the latter to adopt a more effective communicative style – for instance, by slowing down one’s speech in order to re-calibrate an overly fast-talking partner (Brown, Giles, & Thakerar, 1982). Similarly, therapists may diverge from their patients by decreasing the amount of talking they do, to encourage patients to talk more (Matarazzo, Weins, Matarazzo, & Saslow, Reference Matarazzo, Weins, Matarazzo, Saslow, Schlier, Hung, Matarazzo and Savage1968).

Divergence can also be used to indicate that certain spheres of knowledge and behaviors may not be shared among interactants, with the goal of preventing misunderstandings or misattributions. For instance, non-native speakers sometimes deliberately “self-handicap” (Weary & Arkin, Reference Weary, Arkin, Harvey, Ickes and Kidd1981) by broadening their accent when talking to native speakers in their host community. Such a divergent shift signals that they are not members of or familiar with the host community, and that any norms they violate should be attributed to their ignorance and non-nativeness, rather than to malevolent intent. In some situations, speakers may also diverge from others intentionally with the goal of making communication problematic (Giles et al., Reference Giles, Coupland, Coupland, Giles, Coupland and Coupland1991).

Motivational Processes

CAT conceptualizes motivation as an emergent process that can dynamically change during the course of interaction. People enter a given communicative encounter with an initial orientation. As the interaction progresses, this initial orientation is transformed into a psychological accommodative stance, based on the salience of different identities and interactants’ perceptions of their own and others behaviors.

Initial orientation. How people initially adjust their communication is a function of their initial orientation, or their predisposition to construe one another in interpersonal or intergroup terms in conjunction with their initial intentions with respect to accommodation (Gallois et al., Reference Gallois, Ogay, Giles and Gudykunst2005). CAT proposes several macro-level factors that can influence interactants’ initial orientation, including interpersonal history, sociocultural norms and values, and the current and past state of relevant intergroup relations.

Interpersonal history. Interactants’ interpersonal history can vary in terms of duration – that is, from no contact (e.g., meeting someone for the first time) to a long-term relationship (e.g., a married couple) – and in terms of valance – that is, from predominantly negative to predominantly positive. When interactants share a positive interpersonal history, they are more likely to adopt an interpersonal orientation and converge toward one another. In contrast, when their interpersonal history is negative, they are more likely to diverge from one another (Gallois et al., Reference Gallois, Ogay, Giles and Gudykunst2005).

Sociocultural norms. Sociocultural norms and values specify with whom, when, and how it is appropriate to interact. As such, they not only circumscribe the available opportunities for intergroup contact, but also shape interactants’ behaviors. For instance, sociocultural norms often specify what (sort of) language is appropriate to speak in a given situation (Gallois & Callan, Reference Gallois, Callan, Giles, Coupland and Coupland1991). One such norm is the expectation that speakers will converge to those who speak the “standard,” or prestige variety of a language (e.g., Standard American English), particularly in status-stressing situations, such as a job interview (Giles & Marlow, Reference Giles, Marlow and Salmon2011). The tendency to treat others in interpersonal versus intergroup terms is also likely to vary culturally. Collectivist cultures tend to share strong beliefs about ingroup identification and loyalty, emphasize group identity over personal identity, and perceive relatively firm intergroup boundaries. In contrast, individualistic cultures tend to have weaker beliefs about ingroup identification and loyalty, emphasize personal over group identity, and perceive intergroup boundaries as relatively permeable. As a result, members of collectivistic cultures tend to be less receptive to convergence from outgroup speakers and are more likely to diverge from them than are members of individualistic cultures (e.g., Aritz & Walker, Reference Aritz and Walker2010). In this vein, Ross and Shortreed (Reference Ross and Shortreed1990) noted that when non-native speakers in Japan attempt to converge linguistically toward their Japanese interlocutors, they sometimes receive replies in English rather than Japanese. In other words, when cultural boundaries are strongly adhered to, attempts to cross them may be unwelcome.

Intergroup relations. Current and past relations between social groups can also be an important determinant of whether people initially construe one another in interpersonal or intergroup terms (see Dragojevic & Giles, Reference Dragojevic, Giles and Berger2014), and whether they are motivated to converge or diverge. When interactants belong to groups that have historically been involved in hostile or violent relations, they are more likely to construe the encounter in intergroup terms and to diverge from one another as a way to emphasize their valued ingroup identity. One important construct in the analysis of the relations between cultural and ethnic groups is ethnolinguistic vitality (Giles, Bourhis, & Taylor, Reference Giles, Bourhis, Taylor and Giles1977; Giles & Johnson, Reference Giles, Johnson, Turner and Giles1981). A group’s vitality is determined by three factors: its status (i.e., economical and sociocultural prestige), demography (i.e., number and distribution of speakers), and the degree of institutional support (i.e., representation in social institutions such as education or government) it enjoys. Although many dimensions of vitality can be measured objectively, interlocutors’ perceptions of their respective groups’ vitalities are better predictors of their attitudes during interaction (Giles et al., Reference Giles, Bourhis, Taylor and Giles1977). CAT posits that historically strong collectives (i.e., high vitality groups) are more likely to diverge in intergroup situations. Moreover, members who have a strong attachment and loyalty to their ingroup group (i.e., high ingroup identification) are more likely to diverge than those who have only a weak attachment. In this vein, Giles and Johnson (Reference Giles and Johnson1987) found that Welsh participants who were strongly identified with the ingroup diverged from a threatening English person even when their sense of group vitality was low. However, for those Welsh participants who only moderately identified with the ingroup, a sense of high ingroup vitality was required for divergence. CAT also suggests that divergence is more likely to occur when group members feel that their status in the intergroup hierarchy is illegitimate and unfair (see also, Vincze & Henning-Lindblom, in press).

Psychological accommodative stance. Once people begin to interact, their initial orientation is transformed into their psychological accommodative stance, or their immediate and ongoing intentions with respect to accommodation (Gallois et al., Reference Gallois, Ogay, Giles and Gudykunst2005). A speaker’s accommodative stance is shaped by their perception of the salience of personal and social identities in the interaction, as well as their perceptions of their partners’ motives and behaviors. In this respect, one’s stance is dynamic and has the potential to shift on an ongoing basis throughout the encounter (see Genesee & Bourhis, Reference Genesee and Bourhis1982), as interlocutors react and respond to perceptions of each other’s behaviors, needs, and motives. For instance, an initially accommodative stance can quickly become nonaccommodative when one of a speaker’s social identities becomes salient and they wish to positively differentiate themselves from their partner on this dimension. In this vein, Bourhis, Giles, Leyens, and Tajfel (Reference Bourhis, Giles, Leyens, Tajfel, Giles and Clair1979) found that when a French confederate asked trilingual (Flemish-English-French) Flemish students a content-neutral question in English, the students converged to English. However, when the French confederate diverged into French to voice an ethnically threatening question, the Flemish students overwhelmingly diverged into Flemish and vehemently disagreed with the French confederate’s statements. In other words, the French confederate’s threatening question changed the Flemish students’ initially accommodative orientation into a nonaccommodative one.

General Patterns

Convergence typically elicits favorable evaluations, particularly when it is symmetrical (Richmond & McCroskey, Reference Richmond and McCroskey2000), and has been shown to increase a speaker’s perceived attractiveness (e.g., Street, Brady, & Putnam, Reference Street, Brady and Putnam1983), intelligibility (Triandis, Reference Triandis1960), interpersonal involvement (LaFrance, Reference LaFrance1979), and perceived competence and credibility (Aune & Kikuchi, Reference Aune and Kikuchi1993), as well as to facilitate compliance (Buller, LePoire, Aune, & Eloy, Reference Buller, LePoire, Aune and Eloy1992), build rapport (Acosta & Ward, Reference Acosta and Ward2011; Crook & Booth, Reference Crook and Booth1997), and increase relational solidarity (Imamura, Zhang, & Harwood, Reference Imamura, Zhang and Harwood2011). Convergence has also been linked to increased agreement between coalition partners during online multiparty negotiations (Huffaker, Swaab, & Diermeier, Reference Huffaker, Swaab and Diermeier2011), more successful negotiations between police negotiators and hostage takers (Taylor & Thomas, Reference Taylor and Thomas2008), more positive attitudes toward and increased intention to purchase products (Run & Fah, Reference Run and Fah2006), improved polling figures for politicians converging to opponents in American Presidential debates (Romero, Swaah, Uzzi, & Galinsky, Reference Romero, Swaab, Uzzi and Galinsky2015), and more prosocial behavior in general (Kulezsa, Dolinski, Huisman, & Majewski, Reference Kulesza, Dolinski, Huisman and Majewski2014). Interestingly, the relatively consistent and broad positive implications of convergence have recently served as an impetus for the development of more “human” computer systems that are able to accommodate to their users. For example, Acosta and Ward (2001) developed a spoken dialog system they named “Gracie,” which is capable of recognizing users’ emotional state from speech and responding with appropriate emotional coloring. An evaluation of this system showed that, consistent with CAT’s propositions, respondents felt significantly more rapport with Gracie than with either of two controls.

In contrast, divergence and maintenance tend to be associated with negative relational outcomes and are often characterized as insulting, impolite, or hostile (Deprez & Persoons, Reference Deprez and Persoons1984; Sandilands & Fleury, Reference Sandilands and Fleury1979). For instance, speakers who deviate from the standard language by maintaining or diverging toward nonstandard varieties (e.g., a Birmingham accent in the UK; African American Vernacular English in the USA) are typically evaluated less favorably on status (e.g., intelligent) and solidarity (e.g., friendly) traits than those who converge (Giles & Watson, Reference Giles and Watson2013). Although this pattern is seen worldwide, the severity of negative evaluations can vary across different groups and cultures. In general, members of low vitality and stigmatized ethnolinguistic groups who fail to accommodate to the standard language tend to garner more negative evaluations than members of high vitality groups. However, when a high-vitality group is perceived to pose a threat to the majority, its members may suffer more extreme sanctions than members of low vitality groups due to the former’s relatively higher (perceived) influence within society (Ryan, Hewstone, & Giles, Reference Ryan, Hewstone, Giles and Eiser1984). Negative outcomes stemming from linguistic divergence/maintenance often go beyond mere speaker evaluations and can have important real-world consequences. Examples of these include discrimination in the workplace (Lippi-Green, Reference Lippi-Green1994) and housing (Purnell, Isdardi, & Baugh, Reference Purnell, Isdardi and Baugh1999); suitability for high-status employment (Giles, Wilson, & Conway, Reference Giles, Wilson and Conway1981); and even attributions of guilt and criminality (e.g., Dixon & Mahoney, Reference Dixon and Mahoney2004; Dixon, Mahoney, & Cocks, Reference Dixon, Mahoney and Cocks2002).

Multiple Meanings

Of course, the general patterns outlined earlier do not always hold. Communicative behaviors often have multiple social meanings and different individuals may have different perceptions of the same behavior. Accordingly, convergence and divergence can both entail costs as well as rewards. For example, although the tendency of members of linguistic minorities to converge to the language of the dominant majority may garner them social rewards (e.g., economic opportunities) in some settings, it may also lead to the potential loss of a valued aspect of their identity (see Marlow & Giles, Reference Marlow and Giles2010), as well as ridicule and social marginalization from ingroup members (Giles & Edwards, Reference Giles, Edwards and Malmkjaer2010; Hogg, D’Agata, & Abrams, Reference Hogg, D’Agata and Abrams1989).

Even when convergence is positively evaluated, full convergence may not always be appreciated by interactants. For example, Giles and Smith (Reference Giles, Smith, Giles and Clair1979) found that full convergence on pronunciation, speech rate, and message content was perceived as patronizing (i.e., overaccommodative) and evaluated negatively. Conversely, convergence only on speech content and speech rate was perceived more positively. Although divergence may be a blow to recipients’ self-esteem, full convergence may also make them uncomfortable. Giles and Smith (Reference Giles, Smith, Giles and Clair1979) suggested that people have different tolerance levels for convergence and that any shifts beyond a person’s desired (i.e., optimal) level will be evaluated negatively by recipients. A similar argument was put forth by Preston (Reference Preston1981), who noted that full convergence by foreign-language learners (i.e., native-like fluency) is often met with distrust and perceived as controlling by native speakers (see also, Ross & Shortreed, Reference Ross and Shortreed1990).

Speakers’ notions of what constitutes adequate and optimal levels of convergence or divergence are partially rooted in sociocultural norms for intergroup contact. For instance, during intergender conversations, mutual divergence on some speech characteristics (e.g., pitch) may be construed as socially, if not sexually, appealing and desirable by both parties. Indeed, when men and women interact, men often adopt more-masculine sounding voices by lowering their pitch (Hogg, Reference Hogg1985) whereas women adopt more feminine-sounding voices by raising their pitch (Montepare & Vega, Reference Montepare and Vega1988). Although these are, objectively, instances of mutual divergence, they may actually more accurately be labeled as “speech complementarity,” given that they may involve psychologically convergent motives, with both parties aiming for a nonverbal stance that conveys their respective gender identity and appeal (Giles et al., Reference Giles, Coupland, Coupland, Giles, Coupland and Coupland1991).

Perceptions and Attributions

How convergence and divergence are evaluated is partly based on the attributions recipients make about those behaviors – that is, the motives and intentions that they think caused it. Simard, Taylor, and Giles (Reference Simard, Taylor and Giles1976) found that convergence was evaluated more favorably when it was attributed to a speaker’s desire to break down cultural barriers (i.e., attribution of deliberate, positive intent), rather than to situational pressures (i.e., not intentional on the part of the speaker). Conversely, speakers who diverged were evaluated less negatively when the behavior was attributed to external pressures, rather than their own malevolent intent (see also Gasiorek, Reference Gasiorek2013; Gasiorek & Giles, Reference Gasiorek and Giles2012; Chapter 5 this volume).

Attributional processes are susceptible to a range of biases, particularly during intergroup encounters. People tend to favor ingroup over outgroup members and make differential attributions about their behavior (Hewstone, Rubin, & Willis, Reference Hewstone, Rubin and Willis2002; Howard & Rothbart, Reference Howard and Rothbart1980). In particular, they are more likely to attribute socially desirable behaviors to positive dispositions of ingroup than outgroup members, and socially undesirable behaviors to negative dispositions of outgroup than ingroup members (Hewstone, Reference Hewstone1990; Hewstone & Jaspars, Reference Hewstone, Jaspars and Tajfel1984).

Other studies have suggested additional factors that may mediate the relationship between speakers’ behaviors and recipients’ evaluations (see Dorjee, Giles, & Barker, 2010), as well as other outcomes, such as attributed intent (Gasiorek & Giles, Reference Gasiorek and Giles2012; Giles & Gasiorek, Reference Giles, Gasiorek, Forgas, László and Orsolya Vincze2013). Myers, Giles, Reid, and Nabi (2009) found that intergroup sensitivity partially mediated the relationship between police officers’ perceived accommodation and participants’ perceptions of those officers’ competence and social attractiveness. Additionally, other studies of police-civilian encounters in China, Taiwan, Korea, Japan, Guam, Canada, and the United States have shown that trust mediates the relationship between perceived accommodation and compliance (Barker et al., Reference Barker, Giles, Hajek, Ota, Noels, Lim and Somera2008; Hajek et al., Reference Hajek, Giles, Barker, Lin, Zhang and Hummert2008; see also, Scissors, Gill, Geraghty, & Gergle, Reference Scissors, Gill, Geraghty and Gergle2009).