Different size,different language? Linguistic ingroup favoritism and outgroup derogation by majority and minority groups

Minority–Majority Effects on Intergroup Discrimination

Research on the effects of sociostructural characteristics of groups has highlighted that group size per se—independent of variables such as group status or group power—has important consequences for members’ behavior (e.g., Erb, Hilton, Bohner, & Roffey, 2015; Imhoff & Erb, 2009; Lücken & Simon, 2005; Simon et al., 2001). Minorities have consistently been found to show higher ingroup bias than majorities (Bettencourt, Miller, & Hume, 1999; Hewstone et al., 2002; Liebkind, Henning-Lindblom, & Solheim, 2008). For instance, Sachdev and Bourhis (1984) reported that minorities were less parity-oriented than majorities in intergroup allocations of symbolic resources, whereas majorities were more concerned about maintaining differentials between the groups. Minorities—but not majorities—also show ingroup bias in the allocation of negative outcomes (e.g., Otten, Mummendey, & Blanz, 1996), which can be considered as a more heinous means of discrimination than resource allocation, as it implies burdening the outgroup rather than just being more generous towards the ingroup¹ (Moscatelli & Rubini, 2013). For their part, majorities show lower intergroup bias than minorities, or even no bias at all (Liebkind et al., 2008; Mummendey & Simon, 1989).

According to a social identity account, minorities are more discriminatory than majorities as minorities need to compensate for their insecure position (Ellemers, Doosje, van Knippenberg, & Wilke, 1992; Tajfel & Turner, 1979). A cognitive explanation claims instead that numerical distinctiveness elicits a higher focus on the ingroup, which in turn leads to higher ingroup identification and stronger ingroup bias (Lücken & Simon, 2005; Otten et al., 1996). Conversely, optimal distinctiveness theory (Leonardelli, Pickett, & Brewer, 2010) maintains that minority membership meets individuals’ need of inclusiveness within the group, as well as the need for intergroup differentiation, better than membership in larger groups. Accordingly, discrimination by minority groups is driven by motives of ingroup support, whereas majority members are more concerned with enhancing ingroup distinctiveness.

Ingroup Favoritism and Outgroup Derogation in Spontaneous Language Use

Previous research has not examined whether and how relative group size influences intergroup discrimination conveyed through language. In this respect, studies on linguistic intergroup bias (Maass, 1999; Wigboldus & Douglas, 2007) have shown that individuals describe ingroup desirable behaviors and outgroup undesirable behaviors in more abstract terms than ingroup undesirable and outgroup desirable behaviors—thus facilitating the generalization of ingroup positive and outgroup negative, but inhibiting the generalization of ingroup negative and outgroup positive features across different contexts and time (Menegatti & Rubini, 2012, 2013, 2014; Semin & Fiedler, 1988). Although under specific conditions individuals can control linguistic bias (e.g., if they provide descriptions in a comparative framework; Douglas, Sutton, & Wilkin, 2008), neither communicators nor recipients are aware of such a strategic use of language (Maass, Castelli, & Arcuri, 2000).

Interestingly for our purposes, a series of minimal group studies have highlighted that even members of minimal groups described behaviors—resource or negative outcome allocations—performed by ingroup members with more abstract positive terms and more concrete negative terms than the same behaviors performed by outgroup members (Moscatelli & Rubini, 2011; Rubini, Moscatelli, & Palmonari, 2007). Moreover, overall groups having a superior/inferior position to the outgroup in terms of power or status, use language in a discriminatory fashion, whereas equal power/equal status groups do not (Moscatelli, Albarello, & Rubini, 2008; Rubini, Moscatelli, Albarello, & Palmonari, 2007).

Importantly, these studies have pointed out that through systematic variations of abstraction and valence of terms in free language use, individuals can express favoritism towards the ingroup or derogation against the outgroup in response to different intergroup conditions (Albarello & Rubini, 2015; Prati et al., 2015; Rubini et al., 2014). Indeed, individuals can focus on positive features of the ingroup by describing the ingroup with positive terms at a higher level of abstraction than negative terms, or they can emphasize negative features of the outgroup by describing it with negative terms at a higher level of abstraction than positive terms (Moscatelli & Rubini, 2011; Rubini et al., 2014). For instance, Moscatelli, Albarello, Prati, and Rubini (2014) found that relatively gratified groups expressed linguistic ingroup favoritism, whereas relatively deprived groups—possibly because of their stronger discontent—showed linguistic derogation in the descriptions of the relatively gratified outgroup. Thus, one may wonder how membership in groups who differ in their relative group size would affect the two facets of linguistic discrimination. Although a study by Guinote (2001) found that minority and majority members use linguistic abstraction as a means to convey perceived ingroup and outgroup homogeneity, to date no research has analyzed how abstraction and valence of language are used by members of majority and minority groups as means of linguistic intergroup discrimination.

The Present Experiment

The present study aims to test the effects of relative group size on linguistic ingroup favoritism and outgroup derogation. Members of equal-size, majority, and minority groups—created by modifying the minimal group paradigm (Tajfel, Billig, Bundy, & Flament, 1971)—freely described the choice made by either an ingroup or an outgroup member (i.e., the target) in distributing negative outcomes (cf. Moscatelli & Rubini, 2011). As mentioned, according to optimal distinctiveness theory (Leonardelli & Brewer, 2001) majorities are more motivated than minorities to obtain positive ingroup distinctiveness. On this basis, we expected that the language used by majority members would mainly express ingroup favoritism, with descriptions of ingroup members being characterized by positive terms at a higher level of abstraction than negative terms (Hypothesis 1a). We expected no outgroup derogation in this condition, that is, no difference in the abstraction of positive and negative terms used in outgroup descriptions (Hypothesis 1b).

As minorities are usually more discriminatory than majorities even when discrimination implies assigning negative outcomes to the outgroup (Mummendey et al., 1992; Otten et al., 1996), we might expect that minority groups would use linguistic ingroup favoritism (Hypothesis 2a) but also linguistic outgroup derogation in descriptions of the majority outgroup (Hypothesis 2b). This hypothesis is in line with social identity’s assumption that members of minority groups discriminate to compensate for their insecure position (Tajfel & Turner, 1979), as well as with the cognitive account of minorities’ discrimination (Simon et al., 2001). In line with previous evidence that members of mere categories do not show linguistic discrimination in descriptions of negative outcome allocations (Moscatelli et al., 2014; Moscatelli & Rubini, 2011), no ingroup favoritism or outgroup derogation was expected in the equal-size condition (Hypothesis 3).

Whereas the aforementioned hypotheses concerned the prevalent facet of linguistic discrimination expressed by groups of different size—and require a focus on the language used within each group condition—it is possible to advance predictions with respect to differences in descriptions of ingroup and outgroup members provided by majority, minority, and equal-size groups. First, in line with the hypothesis of outgroup derogation by minority groups, we expected that minorities would depict outgroup members less favorably—that is, with negative terms at a higher level of abstraction and/or positive terms at a lower level of abstraction—than majorities would (Hypothesis 4a). Although we expected ingroup favoritism by both majority and minority groups, it is possible that majorities—because of their higher motivation to establish positive ingroup distinctiveness (Leonardelli & Brewer, 2001)—would show more favorable descriptions of the ingroup than minorities would by using positive terms at a higher level of abstraction, and/or negative terms at a lower level of abstraction (Hypothesis 4b). Both majority and minority groups should show more favorable descriptions of the ingroup and less favorable descriptions of the outgroup, compared to equal-size groups (Hypothesis 5).

As a further test of group size effects, in the present study participants made an explicit rating of the extent to which they desired the ingroup to receive favorable outcomes or the outgroup unfavorable outcomes, in the allocation task. We expected that both majorities and minorities would report higher desire for ingroup favoritism than equal-size groups (Hypothesis 6a). Minorities should also report higher desire for outgroup derogation than the other groups (Hypothesis 6b). As minorities usually identify more with the ingroup than majorities do (e.g., Leonardelli & Brewer, 2001; Lücken & Simon, 2005; Otten et al., 1996), in this study we measured ingroup identification as a check on the efficacy of group size manipulation. Although most studies on the linguistic intergroup bias (LIB) have reported no correlation between linguistic abstraction and self-report measures (Maass, 1999; Maass et al., 2000), we also tested whether identification was related to ingroup favoritism and/or outgroup derogation under different group size conditions.

Method

Results

Linguistic Abstraction

For each participant, positive and negative abstraction indexes were computed. Considering positive and negative terms separately, weights of 1, 2, 3, and 4 were assigned to DAVs, IAVs, SVs, and ADJs, respectively. The summed weights were then divided by the total number of terms used. Scores on each index ranged from 1 (lowest abstraction) to 4 (highest abstraction). Scores were submitted to a 3 (relative group size) × 2 (target group membership) × 2 (valence of language) ANOVA, with the last factor within-participants. Since preliminary analyses showed no effect of gender or perceptual style on the dependent variables, they were not considered in the analyses. The analysis showed a significant Group Size × Valence of Language interaction, F(2, 136) = 3.10, p = .048, η² = .044, and a significant Target Group Membership × Valence of Language interaction, F(2, 136) = 27.92, p < .001, η² = .170. These interactions were qualified by a significant three-way interaction, F(2, 136) = 4.89, p = .009, η² = .067 (see Figure 1). The interaction was decomposed by Bonferroni pairwise comparisons to test the specific hypotheses.

OPEN IN VIEWER

Figure 1.

Positive and negative abstraction scores as a function of relative group size and target’s group membership.

In order to test whether majority and minority groups differed in the extent and type of discrimination that they show, we compared the abstraction of positive versus negative terms referring to ingroup or outgroup within each group size condition (i.e., where majorities rated minorities, and vice versa; whereas equal-size groups rated each other). All predictions were supported. Majorities showed ingroup favoritism, with positive terms used in ingroup descriptions being at a higher abstraction than negative terms, p < .001, d = 1.131 (Hypothesis 1a). No outgroup derogation occurred in the majority condition, p = .370 (Hypothesis 1b). Minorities, too, showed ingroup favoritism as revealed by the higher abstraction of positive versus negative terms in ingroup descriptions, p = .011, d = 0.981 (Hypothesis 2a). As expected, minorities also showed derogation against the majority outgroup by describing the latter with negative terms at a higher level of abstraction than positive terms, p < .001, d = 0.453 (Hypothesis 2b). Supporting Hypothesis 3, no ingroup favoritism, p = .215 nor outgroup derogation, p = .727 occurred in the equal-size condition.

With respect to the comparison between the majority and the minority conditions, as expected (Hypothesis 4a) minorities described the outgroup less favorably than majorities did by using negative terms at a higher level of abstraction, p < .001, d = 0.638. The two conditions did not differ with respect to positive terms referring to the outgroup, p = 1.00. Moreover, majorities described the ingroup more favorably than minorities did by employing positive terms at a higher level of abstraction, p < .001, d = 0.676 (Hypothesis 4b). No difference was found with respect to negative terms referring to the ingroup, p = .472.

Overall both majority and minority groups used language in a more discriminatory fashion than equal-size groups (Hypothesis 5). Majorities described the ingroup more favorably than equal-size groups did by using positive terms at a higher level of abstraction (M_{equal size} = 1.81, SD = 0.93; M_majority = 2.45, SD = 0.92), p = .009, d = 0.692. Majorities also described the outgroup with negative terms at a higher level of abstraction than equal-size groups (M_{equal size}= 1.40, SD = 0.62; M_majority = 1.84, SD = 0.90), p = .045, d = 0.569. The two conditions did not differ with respect to negative terms referring to the ingroup (M_{equal size} = 1.47, SD = 0.73; M_majority = 1.48, SD = 0.79),p = 1.00, or positive terms referring to the outgroup (M_{equal size} = 1.49, SD = 0.56; M_majority = 1.59, SD = 0.84), p = 1.00. Minorities depicted the outgroup less favorably than equal-size groups did by using negative terms at a higher level of abstraction (M = 2.72, SD = 0.94), p < .001, d = 1.658, whereas no difference was found with respect to positive terms referring to the outgroup (M_minority = 1.55, SD = 0.87), p = 1.00. The two conditions did not differ with respect to positive (M_minority = 1.84, SD = 0.87), p = 1.00 or negative terms (M_minority = 1.17, SD = 0.79), p = .538 referring to the ingroup. ⁴

Finally, correlation analyses were run to assess whether within each group size condition, positive and negative abstraction scores were related to ingroup identification. The analyses revealed that in the equal-size condition, ingroup identification was negatively correlated with the abstraction of negative terms employed in ingroup descriptions, r = −.482, p = .023. No other correlation was significant in this condition, ps > .140. Ingroup identification and the abstraction scores were not correlated either in the majority, ps > .070, or in the minority condition, ps > .462. Ingroup identification was also entered as covariate in a 3 (relative group size) × 2 (target group membership) × 2 (valence of language) ANCOVA. The analysis revealed no significant relationship between the covariate and the dependent variables, F(1, 135) = 0.812, p = .369.

Discussion

This study examined for the first time how variations in relative group size influence ingroup favoritism and outgroup derogation in free language use. Findings showed that the mere fact of belonging to a group that differs in its size from an outgroup leads members to use language in a biased way, whereas equal-size groups do not show linguistic discrimination. The most novel finding of this study was that majorities and minorities differed in terms of the linguistic discrimination they evidenced. Majorities showed ingroup favoritism by describing ingroup members with positive terms at a higher level of abstraction than negative terms. Minority groups, too, expressed linguistic ingroup favoritism. However, comparisons between the two conditions revealed that majorities provided more favorable descriptions of ingroup members by using positive terms at a higher level of abstraction than minorities did. These results are in line with predictions based on optimal distinctiveness theory (Leonardelli & Brewer, 2001) as majorities appeared more concerned than minorities with establishing a positive ingroup distinctiveness by maximizing the stability of positive ingroup features. Supporting this contention, in the present study majorities also reported a higher desire for ingroup favoritism than both minorities and equal-size groups. No linguistic outgroup derogation occurred in majority groups.

For their part, minority members did show outgroup derogation. Indeed, they described the majority outgroup with negative terms at a higher level of abstraction—implying a higher stability of its negative features—than positive terms. This is a novel finding in the literature on group size effect, and it is further supported by the higher desire for outgroup derogation reported by minorities compared to the other types of groups. Of course, the implicit nature of the linguistic measure employed in this study has likely allowed individuals to overcome the normative constraints that inhibit overt outgroup derogation. Namely, whereas individuals may refrain from showing outgroup derogation with more explicit measures, in this study participants could have intentionally chosen the valence of terms but were hardly able to monitor abstraction (cf. Douglas et al., 2008).

The linguistic outgroup derogation expressed by minority groups can be accounted for by a social identity account (Tajfel & Turner, 1979). Indeed, minorities might have tried to transmit unfavorable images of the outgroup as a way to compensate for their numerical disadvantage. This finding is also in line with predictions based on the cognitive account of minority discrimination (Lücken & Simon, 2005). In fact, the awareness that few individuals in the experimental setting belonged to the same group as themselves may have led minority members to be more concerned with their ingroup—as supported by the higher ingroup identification reported by minority groups compared to other groups—and to derogate the outgroup when they were presented with behaviors performed by majority group members. Conversely, following optimal distinctiveness theory, the higher sense of ingroup inclusiveness elicited by minority membership should actually reduce outgroup derogation (Leonardelli et al., 2010). Thus, overall optimal distinctiveness theory better explains ingroup favoritism shown by majority and minority groups, whereas the social identity and the cognitive account seem to provide a better explanation of outgroup derogation shown by minority groups.

Further studies should address this issue by examining the mechanisms underlying minorities’ outgroup derogation. Although in this study we measured ingroup identification, we were not able to assess whether this variable accounted for linguistic outgroup derogation, as identification and abstraction scores were not related. This absence of correlation is in line with previous evidence on the LIB, and it is generally attributed to the varying degrees of intentional control individuals can exert on linguistic abstraction or on the more overt self-report measures (Maass et al., 2000; Rubini et al., 2014). However, in principle, different theoretical accounts would lead to different predictions with respect to the relationship between identification and minorities’ outgroup derogation. From a social identity perspective (Ellemers et al., 1992; Tajfel & Turner, 1979) one should expect a positive correlation between the two measures—that is, the more members identify with the ingroup, the more they should try to obtain a positive social identity, even by downgrading the outgroup. A similar finding could be expected based on the cognitive account, as minorities’ higher focus on the ingroup should result in outgroup derogation (Lücken & Simon, 2005; Otten et al., 1996). However, as mentioned, optimal distinctiveness theory claims that minorities’ greater satisfaction might prevent them from derogating the outgroup (Leonardelli et al., 2010). Accordingly, one should expect no correlation—or even a negative correlation—between the strength of minority ingroup identification and outgroup derogation. It is therefore important that further research assesses the relationship between ingroup identification and outgroup derogation by employing a different, less implicit measure of minorities’ motivation to hurt the outgroup (e.g., de Dreu, 2010; Halevy, Chou, Cohen, & Bornstein, 2010; Weisel & Böhm, 2015).

Future studies could also test the role of other variables as possible mediators of minorities’ outgroup derogation. For instance, they could test whether minorities feel threatened due to their numerical inferiority. In this respect, one might wonder whether in this study minorities reacted with outgroup derogation as they felt more threatened than majorities (e.g., Islam & Hewstone, 1993) by the specific stimulus behavior we employed, that is, ingroup favoritism shown by the outgroup. Although the procedure we used to obtain linguistic descriptions was similar to that of previous studies (e.g., Moscatelli et al., 2014), further studies should therefore consider less extreme choices as stimuli for descriptions. In line with previous evidence that minimal group members favor the ingroup if they expect the ingroup to favor them, and discriminate against the outgroup if they expect to receive less favorable treatment by the outgroup (e.g., Locksley, Ortiz, & Hepburn, 1980), one may also speculate that in this study linguistic patterns of ingroup favoritism and outgroup derogation might represent an attempt at “reciprocating” the allocation choices allegedly made by ingroup and outgroup members. In addition, although in this study participants’ own allocations were not associated with linguistic abstraction scores, it is possible that participants—as their own choices were closer to the parity end than to the ingroup favoritism end of the matrix—found the allocation choice they had to explain quite surprising. Thus, future studies might test the impact of relative group size on linguistic ingroup favoritism and outgroup derogation by using different stimulus-behaviors.

Finally, in this study relative group size did not affect participants’ allocation choices. This could be due to the fact that we employed a single allocation matrix as a means to create stimuli for open-ended descriptions rather than to measure intergroup bias. In fact, the measurement of intergroup bias in allocation behavior would instead require the use of several matrices in order to calculate the extent to which a specific allocation strategy (e.g., ingroup favoritism) dominates over another (e.g., parity; see for instance Bourhis et al., 1994).

To conclude, these findings add to the literature on linguistic discrimination as they show for the first time that besides group status and group power (Moscatelli et al., 2008; Rubini et al., 2014; Rubini, Moscatelli, Albarello, et al., 2007), the relative size of groups is also a powerful antecedent of a biased use of language. In this respect, Moscatelli et al. (2008) reported that differential group status mainly triggered linguistic outgroup derogation, although higher status groups also showed linguistic ingroup favoritism. Asymmetries in group power affected linguistic favoritism only, although this was possibly due to the fact that in that study group members were able to discriminate at a behavioral level before providing linguistic descriptions (Rubini et al., 2014; Rubini, Moscatelli, Albarello, et al., 2007). Thus, findings of the present studies support the importance of considering the effects of group size per se (Hewstone et al., 2002; Lücken & Simon, 2005; Sachdev & Bourhis, 1984, 1991) in that they reveal that majority and minority membership have distinct and predictable consequences on these two facets of discrimination. At the same time, these results contribute to the literature on group size effects by showing that membership in majority and minority groups has specific effects on ingroup favoritism towards and derogation of the majority outgroup.

More generally, these findings go beyond past research by suggesting that minority membership can lead individuals to cross the fine boundary between expressing one’s attachment to the ingroup and harming the outgroup—at least at the symbolic yet pervasive level of linguistic discrimination. If a certain degree of ingroup favoritism can be considered a direct consequence of group membership, outgroup derogation can actually have detrimental effects on intergroup relations by hindering communication and future interaction with the outgroup. Future studies should address this issue by examining the consequences of portraying the outgroup with negative abstract language.