Conversation as a resource for in ̄uence: evidence for prototypical arguments and social identi®cation processes

Members of eight single-sex groups each consisting of three proand three anti-capital punishment adherents discussed their views for 30 minutes, and afterwards individually rated ingroup and outgroup members on social in ̄uence ranking, prototypicality, and social attractiveness. From the intragroup hypothesis that speaking turns are a resource for in ̄uence (Ng & Bradac, 1993), we predicted and found that turns were correlated strongly with in ̄uence in the intergroup context. Further, using self-categorization theory (SCT; Turner, 1985), we hypothesized that social identity processes would interact with turns, especially with turns obtained through interruptions. Interruptions encoded in prototypical utterances were more strongly correlated with social in ̄uence and prototypicality, but not social attraction, than interruptions encoded in non-prototypical utterances. Further, interruption attempts enacted in prototypical utterances were found to be more likely to be successful than unsuccessful in obtaining turns, while those enacted in non-prototypical utterances were more likely to be unsuccessful than successful. Additionally, interruption turns were longer when enacted in prototypical over non-prototypical utterances. Overall, the ®ndings suggest that the power/in ̄uence of language is interactively organized and constructed around salient self-categorizations. Copyright # 2000 John Wiley & Sons, Ltd. It has been demonstrated that group members who speak the most frequently, or have the greatest proportion of a group's turns or interruptions, subsequently emerge highest in in ̄uence ranking (Bales, Strotbeck, Mills, & Roseborough, 1951; Brooke & Ng, 1986; Ng, Bell, & Brooke, 1993; Ng, Brooke, & Dunne, 1995; Scherer, 1979). This ®nding may be explained by assuming that conversational turns function as a resource for establishing in ̄uence. Once a turn is gained, the speaker may establish CCC 0046±2772/2000/010083±18$17.50 Received 3 March 1998 Copyright # 2000 John Wiley & Sons, Ltd. Accepted 1 April 1999 European Journal of Social Psychology Eur. J. Soc. Psychol. 30, 83±100 (2000) *Correspondence to: Scott A. Reid, School of Psychology, University of Queensland, St Lucia, Queensland 4072, Australia. e-mail: [email protected] Contract grant sponsor: Victoria University of Wellington. conversational control by engaging others in a hearer role (Zajonc, 1960), maintain or change the topic, or allocate the following turn to a speci®c other person (Ng, 1996; Ng & Bradac, 1993). The present study addresses two related aspects of the resource± in ̄uence relationship that have been overlooked in existing research. First, studies conducted within the resource paradigm have without exception employed intragroup comparative contexts. They have most commonly employed task-oriented (e.g. Bales, 1955) or ad hoc discussion groups (e.g. Ng et al., 1995). In other words, these studies have investigated conversational behaviour as occurring between individuals qua individuals. The role of group norms and identities in the resource±in ̄uence relationship have not been investigated. This is of particular concern in light of theory (e.g. Social Identity Theory, Tajfel & Turner, 1979) and research ( for a review see Hogg &Abrams, 1988) that demonstrate the e€ects of social comparative context on social behaviours and evaluations. Given that social identity becomes salient in the context of other groups, it is important to consider the intergroup dimension to the resource±in ̄uence relationship. Second, research conducted within the resource paradigm has largely neglected conversational content in favour of conversational form. It is possible, depending upon content, that only particular turns are related to in ̄uence. Self-categorization theory (Turner, 1985; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987; Turner, Oakes, Haslam, & McGarty, 1994) suggests that turns with prototypical content should be more in ̄uential than turns with less prototypical content. Because prototypical utterances provide information about the contextually salient social identity, they are, by de®nition, in ̄uential (Abrams & Hogg, 1990; Turner, 1991). Thus, turns gained using utterances high in prototypicality (high ®t with social context) will bear a stronger relationship to social in ̄uence than turns gained through less prototypical utterances (low ®t with social context). CONVERSATION AS A RESOURCE FOR INFLUENCE The idea that language functions as a resource for in ̄uence or power has beginnings in the work of Bales and colleagues (e.g. Bales, 1955; Bales et al., 1951). Bales, although initially focusing upon language content, has ultimately lent conversational form a pivotal explanatory role, Bales (1970, pp. 76±77) argued: Who speaks how much and to whom in the group is a `brute fact' characterizing the actual present situation. Speaking takes up time. When one member speaks, it takes time and attention from all other members of the group, some of whom may want to speak themselves. To take up time speaking in a small group is to exercise power over the other members for at least the duration of the time taken, regardless of the content . . . Within the small group the time taken by a given member in a given session is practically a direct index of the amount of power he has attempted to exercise in that period [emphasis added]. Indeed, Bales based these assertions on the robust ®nding that a speaker's conversational in ̄uence, in terms of rated quality of ideas, is strongly and positively related to the number of turns taken (e.g. Bales, 1956; Bales et al., 1951; see also Scherer, 1979; Brooke & Ng, 1986). More recently Ng et al. (1993, 1995) found that post-discussion 84 S. A. Reid and S. H. Ng Copyright # 2000 John Wiley & Sons, Ltd. Eur. J. Soc. Psychol. 30, 83±100 (2000) in ̄uence ranking correlated strongly and positively with turns, and with both successful and unsuccessful interruptions. A notable feature of the Ng et al. (1993) study was that turns gained through interruption were a more powerful predictor of in ̄uence ranking than turns gained by non-interruptive means. This ®nding suggests that interruptions are more power-oriented than turns achieved through noninterruptive means. In explaining these ®ndings, Ng and colleagues have taken the position, consistent with Bales (1970), that conversation is a resource for gaining in ̄uence and power. They further assumed that by speaking the longest and/or interrupting the most, a conversationalist prevents others from speaking, and thus exercises conversational control (Ng, 1996; Ng & Bradac, 1993). Importantly, Ng and Bradac (1993) argue, contrary to Bales (1970), that utterance content plays a central role in the in ̄uence process. INTERRUPTIONS, UTTERANCE CONTENT, AND SCT Self-categorization theory (SCT) adds to social identity theory (e.g. Tajfel, 1978; Tajfel & Turner, 1979, 1986) by elaborating on the social cognitive underpinnings of social identi®cation. The process by which a social identity becomes salient depends upon an `accessibility' by `®t' interaction. Accessibility represents the individual's predisposition to any given categorization, and recognizes variability in the individual's experiences, expectations, motives, goals, and needs. Fit is broken into two separate components re ̄ecting comparative and normative aspects of the categorization process. Comparative ®t describes the in ̄uence of social structure, while normative ®t describes the associated content. Comparative ®t is de®ned by meta-contrast (Campbell, 1958). A collection of individuals will be more likely to be categorized as an entity to the extent that average inter-category di€erences exceed average within-category di€erences; that is, the metacontrast ratio exceeds one (see McGarty, Haslam, Hutchinson, & Grace, 1995). This aspect of ®t is simply concerned with the relatively mechanistic aspect of stimulus representation; it is the interaction with normative ®t that lends social relevance to the theory (cf. Tajfel, 1981; Oakes, 1996). Having categorized entities into distinct groups, normative ®t describes the consensually represented social meaning correlated with the categorization. Comparative and normative ®t jointly determine the typicality or representativeness (i.e. prototypicality) of a group. At the level of social identity, social comparisons between self and the group prototype determine self-prototypicality. That is, self is perceived in terms of the contextually salient ingroup prototype, is seen as interchangeable with other individuals who are also perceived to be prototypical of that group, and is simultaneously perceived as positively distinct from a relevant outgroup(s). Further, the ability (as contextually de®ned) and inclination (via category accessibility) to conform to the group prototype varies across group members. Roger, Bull, and Smith (1988) di€erentiate between successful and unsuccessful interruptions. Successful interruptions are coded when a speaker is prevented from completing an utterance, while the interrupter completes an utterance. Unsuccessful interruptions are coded when either the speaker who is being interrupted is not prevented from completing an utterance, or the attempting interrupter does not complete an utterance. Conversational in ̄uence 85 Copyright # 2000 John Wiley & Sons, Ltd. Eur. J. Soc. Psychol. 30, 83±100 (2000) Consequently, the individual highest in prototypicality is most representative of the ingroup because he or she is the most positively distinct (see Tajfel & Turner, 1986) from the outgroup. Importantly, prototypicality may be de®ned by normative consistency per se when comparative information is constant (e.g. Oakes, Turner, & Haslam, 1991). In Oakes et al.'s second experiment, participants viewed a six-person group consisting of three arts and three science students discussing attitudes to university life. Attributions to social category membership were strongest when the ingroup and outgroup were opposed (high comparative ®t), and the target speaker expressed attitudes consistent with the ingroup norm (high normative ®t). In short, social category attributions were most strongly a€ected by comparative information when it was accompanied by normatively consistent attitudes. This ®nding suggests that when comparative information is constant, the more normatively consistent a speaker is, the more prototypical that speaker will be perceived. Following the work of Oakes and colleagues (Oakes, 1987, Oakes & Turner, 1986, 1990; Oakes et al., 1991) we propose that the ability to gain conversational turns is determined, in part, by the degree of ®t between an individual's utterances and the group prototype. Prototypical utterances provide information that de®nes speakers and listeners within a given social context. Consequently attention will be paid to speakers who use prototypical utterances, hence lending them relatively more conversational turns than speakers whose speech is less prototypical. Speakers who use prototypical utterances will be more likely to be perceived as prototypical than other groupmembers, and will consequently emerge higher in social in ̄uence ranking. However, an underlying assumption in this analysis is derived from Tajfel's work on stereotyping (e.g. Tajfel, 1981; Tajfel & Forgas, 1981). Tajfel argued that people are motivated to maintain and con®rm their values and stereotypes. Applying this argument to SCT suggests that people are motivated to maintain and con®rm salient self-categorizations. Indeed, this hypothesis is supported by Moreland (1985), who found that the mere mention of a category distinction, even when not actually real, lead to the maintenance of the category over time. This suggests, further, that people are motivated to attend to information that de®nes self in a salient social context, but not information that con ̄icts with their self-de®nition (cf. Swann, 1990). In other words, people would attend to prototypical information because it serves to con®rm their social identity, while at the same time they would ignore or actively remove from the environment non-prototypical or aprototypical information.