Busca en el resumen del documento
British Journal of Social Psychology (2011), 50, 747–768 C© 2010 The British Psychological Society
The British Psychological Society
The language barrier? Context, identity, and support for political goals in minority ethnolinguistic groups
Andrew G. Livingstone∗, Antony S. R. Manstead, Russell Spears and Dafydd Bowen School of Psychology, Cardiff University, Cardiff, UK
In two studies, we tested the hypothesis that not having a potentially group-defining attribute (e.g., in-group language) can affect social identification and support for group goals (e.g., national autonomy). Focusing on the Welsh minority in the UK, Study 1 provided evidence that Welsh language fluency predicted Welsh identification and support for national autonomy, and that identification accounted for the language– autonomy association. Study 2 extended this by (1) examining British and English as well as Welsh identification; and (2) quasi-manipulating the surrounding context (Welsh speaking vs. non-Welsh speaking). As predicted, low Welsh language fluency predicted stronger British and English identification, but only where language was criterial (Welsh- speaking regions). British identification, in turn, predicted lower support for national autonomy. Implications and prospects for future research are discussed.
Social identities are an important part of who we are, and we are often motivated to defend them when they come under threat. For members of minority groups, this can involve asserting certain criteria as essential for membership of the group, because it is only by having particular attributes that group members can differentiate themselves meaningfully from members of other groups. However, this can have other, perhaps unwanted, consequences for those who do not possess a criterial attribute. The idea tested in this paper is that ‘essential’ criteria for group membership can undermine the potential of group members to mobilize towards political goals. This is because the identity criteria that define group membership can also impact on group identification among those who do not have the identity-defining attributes.
We test this idea in relation to members of minority ethnolinguistic groups for whom the in-group language can be an important identity-defining attribute, and who potentially hold specific political goals in relation to the in-group’s future (e.g., Francophones in Quebec, Basque, and Catalan people in Spain, and Breton people in France). Focusing on Welsh people in the UK, we first examine the role of in-group language fluency (Welsh)
∗Correspondence should be addressed to Dr Andrew G. Livingstone, School of Psychology, Cardiff University, Tower Building, Park Place, Cardiff CF10 3AT, UK (e-mail: email@example.com).
748 Andrew G. Livingstone et al.
in predicting patterns of national identification, and consider how this in turn predicts orientations to a specific potential political goal (national autonomy for Wales).
Social identity, distinctiveness, and group goals Social identities – dimensions of self that correspond to membership of groups and social categories – form the essential bases of intergroup behaviour (e.g., Sherif, 1966; Tajfel & Turner, 1979). Social comparison with similar out-groups can induce threat by undermining the distinctiveness of the in-group in relation to comparison out-groups (Tajfel & Turner, 1979). As a result, group members may attempt to restore in-group distinctiveness by differentiating between the in-group and the out-group on available dimensions of comparison (Jetten, Spears, & Postmes, 2004; Spears, Jetten, & Scheepers, 2002). In turn, intergroup behaviour is often about more than simply maintaining the integrity of social identities. Specifically, intergroup behaviour is often geared towards achieving group goals (Scheepers, Spears, Doosje, & Manstead, 2002, 2006; Spears et al., 2002). Thus, differentiation or discrimination may be directed towards the establishment and maintenance of a social identity as an end in itself, but may also serve goals for which social identity is a means to an end – for example, as the basis from which members of groups with low status can challenge intergroup status differences (Tajfel & Turner, 1979).
Identity and group goals in minority ethnolinguistic groups This connection between identity needs and group goals is particularly clear in the case of minority ethnolinguistic groups. The erosion of an identity-defining attribute, such as the group’s language, as a result of the hegemonic influence of a powerful out- group, can lead to perceptions that in-group identity faces the possibility of disappearing altogether. In response to such threats, members of minority ethnolinguistic groups may assert the importance of threatened attributes, even defining them as essential to the distinctiveness of the group, thereby promoting their use within the in-group (Giles, Bourhis, & Taylor, 1977). A good example of this is how Welsh speakers who assert that the Welsh language is essential for Welsh identity often also assert that non-Welsh speaking Welsh people should learn the language if they want to be accepted as fully Welsh (Livingstone, Spears, & Manstead, 2009).
In such cases, the extent to which one has a social identity-defining attribute such as in-group language has the potential to influence one’s identification with the in- group. On the one hand, having a particular in-group attribute is likely to buttress strong identification with the in-group, especially when that attribute is seen as vital to the in-group’s distinctiveness. On the other hand, lacking this attribute could undermine in-group identification by showing that one does not live up to particular in-group standards. For example, Bourhis, Giles, and Tajfel (1973) found that non-Welsh speakers felt significantly less Welsh than did Welsh speakers. One reason for this may be the explicit exclusion/rejection of non-Welsh speaking people by Welsh speakers, and the resentment that this arouses (Bowie, 1993; Trosset, 1986). Moreover, both Welsh speakers and non-Welsh speakers evaluated a Welsh-speaking target more positively than an English-speaking target with a Welsh accent. Similarly, Giles, Taylor, and Bourhis (1977) found that, relative to Welsh speakers, non-Welsh speakers felt somewhat removed or inconsistent in identity terms, suggesting that not having the defining attribute of in-group language gave rise to identity ambivalence.
Minority identities and group goals 749
As well as asserting the importance of specific attributes (e.g., language), members of ethnolinguistic minorities often engage in campaigns aimed at achieving specific political goals, such as greater official recognition of or autonomy for the in-group, that would benefit from a support base defined as broadly as possible (van Morgan, 2006). This is the case in Wales, where Plaid Cymru – the main nationalist party – campaigns for greater political autonomy for Wales, with the ultimate aim of Welsh national self-government. Moreover, these goals are clearly identity relevant and identity contingent, particularly in how they draw on the notion of a distinctive Welsh national identity as a reason for pursuing the goal of political autonomy. Yet, while there is evidence of a connection between language fluency and support for group autonomy at a broad, sociological level (e.g., Balsom, 1985; van Morgan, 2006), the identity-related processes underlying it remain under-researched (van Morgan, 2006).
The present research The idea we test in the present research is that although a particular attribute like language may buttress identification among those who possess it, it also has the potential to affect social identification in different ways among those who do not (Jetten, Branscombe, & Spears, 2006; Lewin, 1948). To the extent that it affects patterns of identification, it is likely in turn to affect support for group goals such as political autonomy – a recurring theme in the collective action literature (e.g., Klandermans, 1997; Simon et al., 1998; Van Zomeren, Postmes, & Spears, 2008).
We chose to focus on support for national political autonomy not only because of its topicality in the context of debates surrounding the political future of the UK (e.g., All Wales Convention, 2009), but also because it specifically indicates the desire and potential among minority group members for separation from (an in some senses, parity with) a dominant out-group – a central and long-standing theme in social psychology, as well as in political spheres. This is not to say that other potential political goals are unimportant in this context (e.g., Livingstone, Spears, Manstead, & Bruder, 2009); rather, this particular indicator was selected because of its resonance not only with important issues in intergroup relations research, but also its perennial prominence on the political agenda in Wales, thereby giving it theoretical and practical import.
The role of the Welsh language has historically been a potent issue (Davies, 1994). Although only approximately 20% of the population of Wales can speak it (Welsh Language Board, 2004), the Welsh language is widely accepted as a defining dimension of Welsh identity by Welsh speaking and non-Welsh speaking Welsh people alike. Moreover, although a strong sense of national identity does lead a small number of people to learn Welsh as an adult, Welsh language fluency is in the overwhelming majority of cases something of a ‘given’ around which people attempt to structure a sense of national identity. This is most clearly indicated by research conducted by the Welsh Language Board (2008), which suggests that 93.75% of people who can speak Welsh to any degree of fluency learnt to do so as a child at home or at school, rather than as a conscious choice later in life. Rather than being an active choice born out of national identification (cf. Mann, 2007), Welsh language fluency is therefore more accurately characterized as a structurally determined feature of individuals’ identity repertoire, from which they attempt to shape or negotiate their sense of national identity.
750 Andrew G. Livingstone et al.
Importantly, though, it is not the case that a lack of fluency in Welsh necessarily represents an obstacle to strong Welsh identification (Coupland, Bishop, & Garrett, 2003; Coupland, Bishop, Williams, Evans, & Garrett, 2005; see also Jetten et al., 2006, for a discussion of strategies among ‘peripheral’ group members). Rather, non-Welsh speakers can adopt different strategies that allow them to construct a coherent sense of national identity (Coupland, Bishop, Evans, & Garrett, 2006; Livingstone, Spears, & Manstead, 2009). A key reason for this variation in the consequences of Welsh language fluency is the existence of considerable differences in regional contexts across Wales, particularly in terms of Welsh language fluency and how criterial the language is seen as being for Welsh identity (Bowie, 1993; Osmond, 1985; Thompson & Day, 1999). This provided us with the opportunity to test another hypothesis derived from the above reasoning; specifically, that the consequences of having/not having the group- defining attribute of language for national identification would be more pronounced in regions where it is more widely spoken and is seen as more important for in-group identity.
It is important to emphasize here that while the Welsh language and the pursuit of national autonomy are generally regarded as key features of Welsh national identity, they are of course not necessarily so. As other research shows, the way in which the role of the Welsh language is characterized (e.g., essential vs. not essential) is not fixed, and is often bound up in strategic ways with characterizations of the wider relationship between Wales and England (Livingstone, Spears, & Manstead, 2009). While there is an intimate link between these two aspects of Welsh national identity, their precise character is therefore the subject of contestation and debate by Welsh speakers and non-Welsh speakers. Nevertheless, the context of these debates is one in which the importance of the Welsh language is strongly asserted through media, governmental, and non-governmental institutions, to the extent that its importance (though not necessarily its precise role) is often treated as a ‘given’ in political debates over the future of Welsh governance. While reification is therefore a ‘live’ process in political contexts, studying the consequences of reified definitions of identity (e.g., in terms of support for a particular political project) does not imply reification of identity-defining attributes and group goals from an analytic perspective. Specifically, our focus on a specific attribute (language) and possible group goal (national autonomy) reflects their prominence in political discourse in Wales, rather than an assertion on our part that these necessarily ‘matter’ in fixed or pre-determined ways.
Below, we report two studies examining the relation between in-group language fluency, national identification, and the group goal of national autonomy. Drawing on the reasoning presented above, our main hypothesis was that in-group language fluency would predict national identification, such that those who were less fluent in Welsh would identify less strongly as Welsh. We also predicted that Welsh language fluency would predict support for the specific group goal of greater political autonomy for Wales. Finally, we expected that, consistent with mediation, the association between in- group language fluency and this political group goal would be accounted for by national identification.
Study 1 was conducted in a Welsh-medium high school in a region of Wales where Welsh is not widely spoken. The participants were all sufficiently good Welsh speakers to have Welsh as their language of instruction, although their tendency to speak Welsh in everyday contexts outside of school varied. In Study 2, we recruited participants from across Wales as a whole, including those who could and could not speak Welsh. We hypothesized that the extent to which one has an identity-defining attribute (again
Minority identities and group goals 751
operationalized as self-reported fluency in Welsh) would predict group members’ identification with available alternative identities (‘British’ and ‘English’), as well as identification with their self-selected national in-group (Welsh). Both language fluency and national identification were expected to predict support for political autonomy, with the relation between language fluency and desire for political autonomy being accounted for by national identification, consistent with mediation. We also expected the region in which respondents lived to moderate these relations and will return to these predictions when introducing the second study.
Method Participants At a Welsh-medium school (i.e., in which Welsh was the language of instruction) in South Wales, 116 sixth-form students (63 female, 53 male) were recruited after obtaining informed consent from the head teacher. Participants were 16–18 years old. Not all of them had Welsh as their first language. This meant that the students’ competence in the Welsh language ranged from completely fluent to basic competence, with a range of abilities in between.
Materials In-group language fluency was assessed by the question ‘How would you rate your ability to speak Welsh?’ Responses were made on a seven-point scale with anchors ‘unable to speak Welsh’ (1) and ‘completely fluent’ (7). National identification was measured using a 12-item scale ( = .88) adapted version of Cameron’s (2004) identity questionnaire, sample items being ‘I have a lot in common with other Welsh people’, ‘I often think about the fact that I am Welsh’, and ‘In general, I am glad to be Welsh’. Attitudes to political autonomy for Wales were measured by the question ‘In an ideal world, what sort of relationship would you like Wales to have with the UK?’ with a graded seven- point response scale: ‘complete integration with the UK’ (1), ‘much more integration’ (2), ‘slightly more integration’ (3), ‘as it is now’ (4), ‘slightly more independence’ (5), ‘much more independence’ (6), and ‘complete independence from the UK’ (7).
Procedure Before completing the questionnaires, participants read, and signed informed consent sheets. They were also asked not to discuss the questions or the answers with anyone else, to remain quiet until everyone had finished, and to attempt to answer every question. After everyone had finished, participants were thanked and given a debriefing sheet.
Results Data screening revealed five participants to be multivariate outliers with residual values of greater than or equal to three standard deviations. These participants were excluded from all analyses, leaving a sample of 111 participants.
752 Andrew G. Livingstone et al.
Table 1. Means, standard deviations, and bivariate correlations between all variables in Study 1
Relations between language fluency, identification, and political attitudes Means, standard deviations, and bivariate correlations between the three key variables are reported in Table 1. In order to test our hypothesis that the relation between language fluency and desire for political autonomy would be accounted for by national identification, path models were constructed using AMOS 6.0.
The direct relation between the predictor (language fluency) and the criterion (desire for political autonomy) was significant, = 0.27, p = .004. Next, a partially mediated model was tested by adding the paths from language fluency to national identification and from national identification to desire for political autonomy (Holmbeck, 1997). As a saturated model, this had a 20 value of 0. The direct path coefficient from language fluency to desire for political autonomy was no longer significant, = 0.09, p = .330. A fully mediated model with this path constrained to zero was then tested (Model 2). This model had excellent fit with the data, 21 = 0.94, p = .332; RMSEA = .000; CFI = 1.00; GFI = .994.
To test the significance of the indirect effect in the fully mediated model (illustrated in Figure 1), bootstrapping procedures were carried out (see Shrout & Bolger, 2002). Using 5,000 bootstrap samples, these indicated that the mean (unstandardized) mediation effect was significant, b = 0.17, SE = 0.054 (95% CI [0.069, 0.279]). The bootstrap approximation for the significance level (p) of the effect was .001.
R2 = .204
R2 = .177
Figure 1. Standardized estimates for paths linking Welsh language ability, Welsh identification, and support for in-group political autonomy (Study 1); ∗∗∗p < .001.
Minority identities and group goals 753
Discussion The findings are supportive of our predictions that the association between language fluency and the group goal of greater political autonomy for the in-group would be accounted for by group identification. In this sample, Welsh language fluency predicted Welsh identification, which in turn strongly predicted support for greater national autonomy. This is consistent with the view that achieving a sense of group identity is a basis for propelling the group towards a group goal. However, the fact that having this identity-defining attribute was predictive of Welsh identification, even among a group of Welsh-identifying speakers of the Welsh language, indicates the paradoxical role that can be played by a criterial attribute such as a distinctive language. On the one hand, it is a valued symbol of group distinctiveness; on the other hand, to the extent that in-group members feel that they do not have or have access to it, it has the potential to attenuate strength of in-group identity and thereby to weaken resolve to achieve a group goal.
Although these results offer encouraging initial support for our argument, it is also possible that alternative relations among the variables could be specified. For example, while we can rule out the possibility that having/not having a group-defining attribute mediates the effect of identification on support for group goals, the correlational design of Study 1 does not allow us to statistically rule out the possibility that identification simply predicted both of the other variables. It is important to bear in mind, however, that our specification is more consistent with previous theory and research into identification and group affiliation as a flexible outcome that is sensitive to contextual factors and strategic concerns, and with the nature of Welsh language fluency as a structurally determined attribute (Welsh Language Board, 2008).
This point notwithstanding, our understanding of the relationship between having an in-group attribute and social identification would clearly benefit from a fuller examination. This is a key aim of Study 2. Given the near impossibility of directly manipulating in-group language fluency, Study 2 instead examines the impact of pre- existing regional differences in in-group language fluency. It also examines the role of other (national) identities that are available to people in Wales.
Another limitation of Study 1 is its restricted sample of group members who have the group-defining attribute in question (i.e., Welsh speakers). As noted earlier, Welsh is only spoken by a minority of those who identify themselves as Welsh. A fuller picture therefore requires a broader range of language fluency. This is particularly important given our argument that insisting on the criterial nature of particular in-group attributes, such as language, is likely to backfire when it comes to mobilizing those who do not have this attribute towards group goals such as political autonomy. This issue will also be addressed in Study 2. Specifically, this study involves a broader sample that varied more both in terms of in-group language fluency, and in the degree to which the predominance of language fluency was related to local region.
STUDY 2 As demonstrated in previous research, pre-existing settings usually involve more than one category with which group members can identify (e.g., Crisp & Hewstone, 2006; Hornsey & Hogg, 2000; Mummendey & Wenzel, 1999; Turner, Hogg, Oakes, Reicher & Wetherell, 1987). This is certainly true in the setting of the present research, where other national identities available to people in Wales include ‘British’ and ‘English’ (the latter of which is often characterized as an out-group against which ‘Welshness’ is
754 Andrew G. Livingstone et al.
defined). Specifically, in contexts where being able to speak Welsh is seen as essential to being Welsh, those who cannot speak Welsh but would otherwise define themselves as Welsh can find themselves being positioned as ‘English’ by Welsh speakers (Bowie, 1993; Trosset, 1986). The potential consequences of this for patterns of national identification are intriguing. On the one hand, explicitly resisting their labelling as English might seem an obvious strategy (Branscombe, Ellemers, Spears, & Doosje, 1999). On the other hand, however, such direct resistance is contingent upon one’s power to do so. Other research suggests that being positioned by others in a manner that does not match one’s initial or ideal self-definition can result in new forms of self- definition that attempt to make sense of one’s (new) social position – a dynamic that has been vividly demonstrated in the context of crowd events (e.g., Drury & Reicher, 2000), and also resonates with research on the rejection–identification model (e.g., Schmitt, Spears, & Branscombe, 2003). This latter consequence of being positioned as ‘other’ raises that possibility that low Welsh language fluency in Welsh-speaking contexts (in which non-Welsh speakers are not in a majority and can find themselves positioned as ‘English’) may be associated with greater levels of British and even English identification.
We therefore hypothesized that the consequences of in-group language fluency for national identification would be evident not only in Welsh identification, but also in British and English identification. Moreover, we hypothesized that these effects would be more pronounced in contexts where the perceived importance of the language as an in-group-defining attribute was highest. Finally, we hypothesized that in-group language fluency would also predict group members’ support for the group goals of greater national autonomy, and that this effect would be accounted for by national identification.
We tested these hypotheses by sampling from two regional contexts that differ in terms of how widely spoken Welsh is, and in the perceived importance for Welsh identity (see Balsom, 1985). Data on the geographical location of Welsh speakers from the most recent Census underlines the validity of this distinction (Welsh Language Board, 2004). We refer to these as Welsh-speaking regions (Y Fro Gymraeg, as defined by Balsom) and non-Welsh-speaking regions (incorporating British Wales and Welsh Wales, as defined by Balsom, 1985). In each of these regional contexts, we gauged Welsh language fluency, Welsh, British and English identification, and support for national autonomy. We expected that Welsh language fluency would positively predict Welsh identification and negatively predict British and English identification. In turn, we predicted that these effects would be stronger in the Welsh-speaking regions than in the non-Welsh-speaking regions. Finally, we expected that Welsh identification would positively predict support for national autonomy and that British and English identification would negatively predict support for national autonomy.
Method Participants and design Participants were 646 adults who identified themselves as Welsh. There were 319 male and 327 female participants, with an age range of 18 years to over 80 years; 409 were aged between 18 and 55 years, while 327 were aged over 55 years. Participants were recruited as part of a survey of national identity processes in Wales. A total of 456 were from non-Welsh-speaking regions and 190 were from Welsh-speaking regions.
Minority identities and group goals 755
Survey measures After recording their sex and age group, participants were asked about their nationality, choosing from Welsh, British, English, or Other (please specify). It was on the basis of this item that participants identified themselves as Welsh. The Welsh identification scale ( = .79) consisted of six items, three of which tapped centrality and in-group ties (e.g., ‘I identify with Welsh people’, adapted from Cameron (2004) and Doosje, Ellemers, and Spears (1995)) on a scale of 1 (completely disagree) to 7 (completely agree). The other three items measured degree of felt or experienced group membership (e.g., ‘Overall, how Welsh do you feel?’) on a scale of 1 (not at all Welsh) to 5 (completely Welsh). Scores on these items were standardized before their average was taken. The English and British identification scales each consisted of the three centrality and in-group ties items (’s = .60 and .81, respectively).
Welsh language fluency was measured with a single item. Responses to the question, ‘How would you describe your overall ability to understand and use the Welsh language?’ were recorded on a four-point scale ranging from 1 (excellent ability) to 4 (no ability).
The perceived importance of the Welsh language in Welsh identity was assessed using three items, = .76. Participants were asked, ‘when you think about what it means to be Welsh, how important are the following?’, and rated the importance of the dimension represented in each of the items on a scale of 1 (not at all important) to 5 (extremely important). The items were: involvement in the Eisteddfod (an annual Welsh-language festival of Welsh culture) and other Welsh-language events; being able to speak Welsh yourself; other people in Wales speaking Welsh.
The item measuring support for the goal of autonomy asked participants, ideally, what would be Wales’ relationship with the UK? They responded on a scale where 1 = complete integration into the UK, 2 = greater integration with the UK, 3 = no change to the present situation, 4 = more power but still in the UK, and 5 = completely independent.1
Procedure The survey was administered through telephone interviews. These were conducted using the computer assisted telephone interviewing system, which randomly generates telephone numbers. There were no gender or age exclusion criteria, except that participants were adults.
After giving their informed consent to participate in the study, participants were offered the choice of whether the interview would be conducted in English or in Welsh. All interviewers were Welsh, but not Welsh speaking. If a participant wanted a Welsh- medium interview, a time was agreed at which a Welsh-speaking interviewer could conduct the interview. Otherwise, the interview was conducted immediately. Although exact figures were not recorded, it is estimated that fewer than 10 interviews were conducted in Welsh.
After recording the participant’s sex and age group, the interviewer posed the questions in the order outlined above. After each question, the interviewer read out the response scale before recording the participant’s response. After all questions had been asked, the participant was thanked for his or her contribution.
1Although strictly speaking this is an ordinal measure, it was treated as an interval scale in all analyses on the basis that scores were distributed normally (kurtosis = 0.06, skewness = 0.77); furthermore, the parametric tests employed in these analyses are generally robust to minor departures from distributional assumptions.
756 Andrew G. Livingstone et al.
Results Listwise deletion of cases with missing values meant that the final sample used in the following analyses consisted of 618 participants (433 from non-Welsh-speaking regions and 185 from Welsh-speaking regions). There were no multivariate outliers in this sample.
Perceived importance of the Welsh language in Welsh identity A one-way (in-group region) ANOVA on the perceived importance of the in-group language scale revealed a highly significant effect of in-group region, F(1,616) = 63.48, p< .001, 2p = .093. This confirmed that the perceived importance of the Welsh language was greater in the Welsh-speaking regions (M = 3.92, SD = 0.82) than in the non-Welsh- speaking regions (M = 3.21, SD = 1.08).
In-group language fluency and national identification Means, standard deviations, and bivariate correlations between all variables are reported in Table 2. A 2 (region: Welsh-speaking vs. non-Welsh-speaking) × in-group language fluency (continuous, mean-centred) ANOVA on the Welsh identification scale revealed only a main effect of Welsh language fluency, F(1,614) = 53.12, p < .001, 2p = .080 (all other F’s < 1). Greater Welsh language fluency was associated with stronger Welsh identification.2
A similar ANOVA on the English identification scale revealed reliable main effects of region, F(1,614)= 30.30, p< .001, 2p = .047, and of Welsh language fluency, F(1,614)= 9.37, p = .002, 2p = .015. These main effects were qualified by a significant interaction between the two predictors, F(1,614) = 9.69, p = .002, 2p = .016.
A similar ANOVA on the British identification scale revealed a reliable, negative main effect of Welsh language fluency, F(1,614) = 8.41, p = .004, 2p = .014. The effect of region, F(1,614) = 2.71, p = .100, 2p = .004, and the interaction (F < 1) were not significant.
In order to illustrate the interaction between region and Welsh language fluency on English identification, we conducted simple slopes analyses, illustrated in Figure 2. In contrast to non-Welsh-speaking regions, the association between Welsh language fluency and English identification in Welsh-speaking regions is highly significant. Specifically, lower Welsh language fluency is associated with higher levels of English identification in this region.
Support for national autonomy In order to test how these associations between Welsh language fluency and national identification might translate into support for group goals, multi-group path models
2A similar analysis on an unstandardized three-item scale of Welsh identification, consisting of the items comparable to those used in the English identification scale, revealed a virtually identical pattern. Of particular note were the scores on this scale, which in all cases indicated very high levels of Welsh identification (all predicted values 6 on a seven-point scale). The means and standard deviations on this reduced scale were 6.69 (0.57) in the Welsh-speaking regions, and 6.36 (0.87) in the non-Welsh-speaking regions. On a related point, the Welsh identification scale consisted of more items than the English identification because of the primary focus of the survey on dynamics around Welsh identity, and the considerable cost of including the same number of items for the English identification scale. We retained the full complement of six items for the Welsh identification scale in the analysis on the basis that the marginal benefit of direct comparability with the English identification scale did not warrant the sacrifice of half of the items on the scale.
Minority identities and group goals 757
M ea ns ,s ta nd
ar d de vi at io ns ,a nd
bi va ri at e co rr el at io ns
ee n al lv ar ia bl es
in W el sh -s pe ak in g (b el ow
di ag on
al )a nd
no n- W el sh -s pe ak in g (a bo
ve di ag on
re gi on
s (S tu dy
758 Andrew G. Livingstone et al.
Welsh language fluency
Figure 2. Interaction between region and Welsh language fluency on English identification (Study 2). Slope coefficients are standardized (beta) estimates; ∗∗∗p < .001.
were constructed using AMOS 6.0. This approach was chosen over a series of ANOVA/ANCOVAs or regression analyses for several reasons. In the first instance, it allows for simultaneous estimation of parameters at different stages in the model, reducing experiment-wise error. This was particularly useful given the inclusion of multiple mediators in the present case. Relatedly, the path model takes into account covariance amongst endogenous variables, including multiple mediators, in a manner that would not be possible in conventional regression analyses (Byrne, 2001; Kline, 2005).
The first step here was to test the direct effect of Welsh language fluency, constrained to be equal across region:3 this was significant, = 0.13, p < .001 in both regions, indicating that low Welsh language fluency predicted lower support for national autonomy. The second step was to test a partially mediated model by adding indirect paths from Welsh language fluency to support for national autonomy through Welsh, British, and English identification to. This model constrained all structural paths and covariances to be equal in strength across region, with the exception of that between Welsh language fluency and English identification (reflecting the moderating effect of region revealed by the ANOVAs).4 This model had very good fit with the data, 29 = 5.32, p = .806, 2/df = 0.59, RMSEA = .000, CFI = 1.00, PCFI = .450, AGFI = .989. Moreover, the direct path from Welsh language fluency to support for national autonomy was no longer significant, = 0.06, p = .156. Finally, a fully mediated model was tested, in which this direct path was constrained to be 0. This model also had very good fit with the data, 210 = 7.33, p = .694, 2/df = 0.73, RMSEA = .000, CFI = 1.00, PCFI = .500,
3Comparison of this constrained model to one in which the path was free to vary in strength across region revealed no significant reduction in fit. The more parsimonious fully constrained model was therefore chosen. 4Comparison of this model to a model in which all structural paths were free to vary across region indicated that no paths other than the one between Welsh fluency and English identification differed in strength significantly across regions.
Minority identities and group goals 759
E4 R2= .090
English identification –.04
E1 R2= .000
E4 R2= .075
English identification –.04
Figure 3. Standardized estimates for Welsh-speaking and Non-Welsh-speaking regions (Study 2); ∗∗∗p ≤ .001; ∗∗p < .005; ∗p < .05.
AGFI= .986. The fully mediated model was therefore preferred on the basis of parsimony. This model is illustrated in Figure 3.
In order to test the significance of the indirect effects of Welsh language fluency on support for national autonomy, bootstrapping procedures were carried out on the fully mediated model (Shrout & Bolger, 2002). Using 5,000 bootstrap samples, these indicated that the mean (unstandardized) mediation effect was significant in the non- Welsh-speaking regions, b = 0.085, SE = 0.021 (CI [0.129, 0.045]), p < .001, and in the Welsh-speaking regions, b = 0.095, SE = 0.023 (CI [0.141, 0.052]), p < .001.
Post hoc test of the role of English identification One striking aspect of the path analyses is that the negative bivariate correlation between English identification and support for national autonomy became non-significant in the presence of British identification – a pattern consistent with mediation (Baron & Kenny, 1986). Although we did not specifically predict a causal link from English to British
760 Andrew G. Livingstone et al.
identification, it is plausible to suggest that increasing English identification might in turn lead to increased British identification, as the superordinate category of which ‘Englishness’ is typically seen as prototypical (Cohen, 2000; Langlands, 1999). Such a link would in turn represent an additional indirect link between Welsh language fluency and support for national autonomy. To examine this possibility, we conducted a post hoc path analysis in which the correlation between English and British identification was replaced with a causal path. This model had virtually identical fit indices and equal degrees of freedom (10) to that illustrated in Figure 3.
Finally, we tested the significance of each indirect path. This was done by constraining two of the paths between Welsh language fluency and national identity to be 0, and repeating the bootstrapping analysis above. A significant effect here indicates that the remaining (unconstrained) path is significant. The mean (unstandardized) mediation effect through Welsh identification was highly significant in both regions, b = 0.056, SE = 0.017 (CI [0.091, 0.025]), p < .001. Likewise, the mean (unstandardized) mediation effect through British identification was significant in both regions, b = −0.019, SE = 0.010 (CI [0.041, 0.002]), p = .034. For English identification, the mean (unstandardized) mediation effect was not significant in the non-Welsh-speaking regions, b=−0.004, SE = 0.007 (CI [0.010, −0.018]), p = .555, but was significant in the Welsh-speaking regions, b = 0.030, SE = 0.014 (CI [0.062, 0.006]), p = .005, confirming that the path through English identification was significant, but only in the Welsh-speaking regions.
Discussion We tested our hypotheses by sampling from a broad range of group members with varying degrees of self-reported fluency in the Welsh language. We also employed a quasi-manipulation of context by sampling from regions that differed in terms of how widely spoken the Welsh language is, and how important it is seen as being for Welsh identity. Our first prediction was that feeling less fluent in the Welsh language would be associated with lower Welsh identification and stronger British and English identification. We also predicted that these effects would be stronger in contexts where the Welsh language was most prominent – i.e., in Welsh-speaking regions. Finally, we expected that Welsh identification would positively – and British and English identification negatively – predict support for the group goal of national autonomy.
There was considerable support for our hypotheses. As expected, lower Welsh language fluency predicted lower Welsh identification and greater British and English identification. This supports the hypothesis that possessing an identity-defining attribute can impact on group members’ sense of affiliation with an in-group. Moreover, this can take the form of increased or decreased identification with available alternative categories (including a category typically characterized as an out-group), as well with the self-selected national in-group.
In turn, the association between Welsh language fluency and English identification was moderated by regional context. As illustrated in Figure 2, the association between English identification and Welsh language fluency was not significant in the non-Welsh- speaking regions, but was highly significant in the Welsh-speaking regions. Interestingly, Figure 2 shows that the strong association with Welsh language fluency in the Welsh- speaking region is not driven by especially low levels of English identification among those with good Welsh language fluency. Rather, it is driven by increased levels of English identification amongst those with low or no fluency in the Welsh language.
Minority identities and group goals 761
Indeed, the predicted value when Welsh language fluency is low is above the mid-point (4), indicating some level of agreement with the English identification items. In contrast, all other predicted values are well below the mid-point. Thus, in the region in which the perceived importance of the Welsh language for Welsh identity was highest, not being able to speak the Welsh language was associated with much higher levels of English identification than was the case in other regions.
These moderation findings support our hypothesis that the impact of the extent to which one has an identity-defining attribute on patterns of identification would be greater in contexts where the perceived importance of the attribute is highest. Although the association between Welsh language fluency and Welsh identification did not vary depending on regional context, the negative association between Welsh language fluency and English identification did do so. This suggests that while Welsh identification might remain strong amongst group members who do not have an identity-defining attribute, they may nevertheless display relatively increased levels of identification with categories typically characterized as ‘out-group’.
Identification and group goals We also hypothesized that these patterns of national identification would in turn predict support for national autonomy. This is borne out by the path analyses illustrated in Figure 3. As predicted, Welsh identification positively predicted and British identification negatively predicted desire for the goal of national autonomy in each region. Although English identification did not predict support for autonomy in the context of the path model, the significant bivariate correlation between English identification and support for autonomy suggested that there may well be a link, but that this is accounted for by British identification (which also correlates strongly with English identification). Further path analyses confirmed that if the English–British identification link is assumed to be causal, then there is evidence that (1) the overall indirect effect of Welsh language fluency on support for national autonomy is greater in Welsh-speaking regions, and (2) the indirect effect of Welsh language fluency on support for national autonomy through English identification was significant in Welsh-speaking regions, but not in non-Welsh- speaking regions.
Together with the different cross-regional effects of Welsh language fluency on identification, this means that alongside a path through Welsh identification, there is also evidence for a path in the Welsh-speaking regions from Welsh language fluency to support for national autonomy through British (and potentially English) identification. Specifically, low Welsh language fluency predicts higher English and British identifica- tion, the latter of which in turn predicts decreased support for national autonomy.
Compared to the non-Welsh-speaking regions, there are therefore additional ways in which Welsh language fluency may impact on support for the goal of national autonomy in the Welsh-speaking region. However, this should be interpreted in the light of the patterns illustrated in Figure 2 (particularly regarding English identification) and the low overall levels of Welsh language fluency in the Welsh population. Specifically, it supports the hypothesis that in Welsh-speaking areas there are additional ways in which not having the Welsh language can reduce support for the goal of national autonomy. It does so by increasing the extent to which those unable to speak the Welsh language identify with alternative identities, which in turn reduces support for political separation.
The results of Study 2 also underline the value of seeing identification as an outcome sensitive to contextual and structural factors (Minescu & Funke, 2010), presenting in our
762 Andrew G. Livingstone et al.
view a more plausible specification than an alternative account in which identification is the primary predictor (i.e., of language fluency). Specifically, this alternative would require that, for example, identifying more with an ‘out-group’ (while still identifying strongly with the self-selected in-group) makes group members lose (or at the very least, downplay) their fluency in the Welsh language – but only in certain contexts. Such an account is problematic on two counts. First, it must explain the elevated levels of out-group identification among those with low Welsh language fluency, and why this only occurs in Welsh-speaking regions. We suggest that any plausible explanation of this phenomenon must in turn come back to the role that Welsh language fluency has in these contexts, whereby non-Welsh speakers, despite strong Welsh identification, find themselves positioned as ‘English’ (Bowie, 1993; Trosset, 1986). Second, it must reconcile its position with other research that suggests that the dominant identity positions in relation to a lack of fluency in Welsh do not relate to a simple desire to learn the language (as also suggested by the relatively low numbers of people who learn the language as an adult: Welsh Language Board, 2008). Both available evidence and general plausibility therefore suggest that Welsh language acquisition, which in the overwhelming majority of cases occurs in childhood, precedes the identification effects, rather than driving language acquisition. Thus, these identity positions (e.g., redefining the language as a symbolic resource; downplaying its importance; ambivalence in one’s ‘Welshness’) appear to represent an attempt to structure a sense of national identity around the ‘given’ that is a lack of Welsh fluency (e.g., Bowie, 1993; Coupland et al., 2005; Livingstone, Spears, & Manstead, 2009; Trosset, 1986).
It is also worth considering at this point the potential meaning of English identification among non-Welsh-speakers in Welsh-speaking contexts. As we have already suggested, this most likely reflects a sense of shared social position within Welsh-speaking areas, drawing upon the categorizations that can be imposed upon them by Welsh speakers when they position non-Welsh speakers together with the English – and specifically, English incomers to Welsh-speaking communities (Bowie, 1993; Cloke, Goodwin, & Milbourne, 1998; Trosset, 1986). Social interaction within these communities – particularly between non-Welsh speakers and English incomers – may in turn help to crystallize these patterns of identification. We therefore do not mean to suggest that non- Welsh speakers come to see themselves as somehow interchangeable with, say, English people in Leeds or Leicester. Rather, the patterns of identification evident here most likely reflect the local community dynamics within which non-Welsh speakers attempt to make sense of their social position – which is precisely why patterns of identification among non-Welsh speakers are different in contexts in which they are the majority. Thus, our interpretation of these findings is centred on the point that patterns of identification reflect (and indeed, only make sense in) the social reality faced by people in their local context.
While identification may of course prompt individual group members in some cases to attempt to obtain a group-defining attribute (e.g., to learn the Welsh language), the present specification and results therefore reflect the much broader and more widely applicable dynamic that has been touched upon in previous research in this context (e.g., Bowie, 1993; Trosset, 1986) and beyond (e.g., Jetten et al., 2006; Lewin, 1948). Notwithstanding the propensity to learn (or even to forgo) the in-group language as an active, identity-based choice, the much more applicable and ecologically valid specification is therefore one in which one’s fluency in the Welsh language is a structural ‘given’ which, in interaction with sociolinguistic context, impacts on one’s sense of self and national belonging.
Minority identities and group goals 763
GENERAL DISCUSSION For minority groups, and ethnolinguistic groups in particular, long-term threats to the existence of in-group identity can lead in-group members to assert the importance of particular attributes such as in-group language as a way of protecting in-group identity. While a distinct group identity is necessary for the pursuit of group goals (political and otherwise), we hypothesized that the criterial importance of particular attributes can in some cases undermine the group’s potential to mobilize towards achieving these goals. This is because of its potential to affect patterns of identification among those who do not meet these criteria. In both studies presented here, we found evidence that lacking a group-defining attribute (Welsh language fluency) predicts lower Welsh identification, which in turn predicts reduced support for the national identity project of greater autonomy. Moreover, Study 2 also showed that lacking this group-defining attribute can in certain contexts also predict greater identification with a category typically characterized as an out-group, which in turn predicts reduced support for this autonomy project.
Taken together, these findings support the idea that having or not having a group- defining attribute has implications not only for social identification, but also for subsequent support for potential group goals. Building on the support offered by Study 1, the findings of Study 2 suggest that members of minority groups, and ethnolinguistic groups in particular, may face something of a dilemma when it comes to reacting to the identity threats faced by the in-group. On the one hand, a distinct in-group identity is necessary for the pursuit of in-group goals, particularly when it comes to mobilization towards goals such as political autonomy. On the other hand, establishing particular criteria or dimensions as identity defining may in some circumstances undermine the mobilization potential of in-group identity and actually strengthen the identities which the ‘exclusive’ criteria are meant to undermine (e.g., Englishness in Welsh-speaking regions).
Another implication of the present findings is that group identity is more likely to provide a meaningful platform for political mobilization when it is defined in inclusive terms that still allow for meaningful differentiation between in-group and out-group. There is a link here with research on the ways in which political rhetoric aimed at collective mobilization involves defining not only the relevance of particular social categories, but also the dimensions that define the categories and who is prototypical of them (e.g., Reicher & Hopkins, 1996a, b; see Reicher & Hopkins, 2001, for an overview in relation to national identities). The present findings supplement this research by demonstrating how the mobilization potential of particular social identities may be furthered or undermined by the way in which particular definitions (and the attributes that they emphasize) impact on patterns of identification among group members.
Directions for future research Although the present findings offer support for our hypotheses, we have been careful to avoid suggesting that the structural models tested here provide definitive evidence regarding the causal or mediating role of specific variables. In Study 2 in particular, the key findings relate instead to differences in the strengths of certain associations across regional context. The quasi-manipulation of regional context that made these analyses
764 Andrew G. Livingstone et al.
possible is a key strength, because it provided a degree of control over the role of cross-contextual variation in a very ‘real’ setting.
These strengths notwithstanding, there are clearly some unanswered questions regarding the processes underlying the observed contextual variation. An obvious goal for future research would be to delineate what it is about particular contexts that lead group members to define identity in relatively exclusive terms along particular dimensions. The role of different forms of intergroup threat is likely to play an important role here. On the one hand, status-based threat is likely to require a response that improves the in-group’s status vis-à-vis relevant out-groups (Ellemers, Van-Knippenberg, & Wilke, 1990; Reicher & Haslam, 2006; Tajfel & Turner, 1979). Where this requires collective action or effort, we hypothesize that there will be a tendency towards defining in-group identity in more inclusive terms, so as to include as broad a constituency as possible (Morton, Postmes, & Jetten, 2007). On the other hand, where a status threat is unchallengeable (e.g., Ellemers et al., 1990), or where there is a threat to the distinctiveness or existence of in-group identity itself (e.g., Jetten et al., 2004), there may instead be a tendency towards protecting the integrity of in-group identity. This can involve defining in-group identity in relatively exclusive terms by establishing criteria that maximally differentiate in-group from out-group(s), or by directly downgrading in- group members who deviate from a more narrowly defined prototype (e.g., Marques & Yzerbyt, 1988; Marques, Yzerbyt, & Leyens, 1988; Morton et al., 2007).
As indicated in the introduction, our focus on one particular potential group goal (national autonomy) should not be taken to suggest that this is the only salient group goal within this national group (Livingstone, Spears, Manstead, & Bruder, 2009). Clearly, other national identity projects are relevant within Wales, for example, non-Welsh speakers’ support for autonomy may vary depending on whether they see the devolution project as favouring the Welsh language and thus Welsh speakers. On the other hand, support for autonomy among Welsh speakers in North Wales may be attenuated if they feel that the political and social impacts of devolution are too heavily focused on the South East of Wales.
Finally, while the present analysis has provided some evidence for how patterns of national identification relate to the interplay between language fluency and local context, it is also the case that the categorization processes (both perceptual and strategic) at play here are probably more complex and nuanced than our data can portray. To some degree, this reflects the more general limitations of survey methods for examining identity outcomes in a manner that is sufficiently flexible and does not presuppose the nature or relevance of particular categories. This suggests the value of supplementing the current work with analyses that focus more specifically on characterizations of identity (see Livingstone, Spears, & Manstead, 2009, for an example in a Welsh context), and their relation to the social realities evident in local contexts. By the same token, though, such future work is likely to be limited to examinations of how identity is contested and reified, to the neglect of the consequences of such reification for broad patterns of identification and mobilization – an issue to which the current methods are more suited. We suggest that the respective value of different methodological approaches is therefore contingent upon the recognition that acknowledging and examining the range of differing identity constructions does not preclude analyses of what happens when particular constructions become dominant, while the study of dominant identity constructions should in turn be qualified by the realization that alternatives are always out there (Reicher, 2000).
Minority identities and group goals 765
Conclusion The findings reported here suggest that the vulnerable position of minority groups may present a dilemma when it comes to choosing paths of resistance. On the one hand, attempting to protect or restore the distinctiveness of in-group identity may be a necessary step towards achieving particular political goals. On the other hand, defining in-group identity in particular ways has consequences for the mobilization potential of that identity. This is especially so in cases where the attributes used to define in-group identity (e.g., language) are not readily available to all would-be in-group members. In other words, our ability to get to where we might want to be as a group depends on what defines us as a group – and what this means for who is (or who feels) part of the group.
Acknowledgements This research was made possible by support from the Leverhulme Trust (ECF/2007/0050) and the ESRC (RES-148-25-0014), and forms part of the ESRC Identities and Social Action Programme. We would also like to thank three anonymous reviewers for their comments and suggestions.
References All Wales Convention (2009). All wales convention report. Retrieved from http://wales.gov.uk/
docs/awc/publications/091118thereporten.pdf Balsom, D. (1985). The three-Wales model. In J. Osmond (Ed.), The national question again:
Welsh political identity in the 1980s (pp. 1–17). Llandysul: Gomer Press. Baron, R. M., & Kenny, D. A. (1986). The Moderator-Mediator variable distinction in social psycho-
logical research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. doi:10.1037/0022-35126.96.36.1993
Bourhis, R. Y., Giles, H., & Tajfel, H. (1973). Language as a determinant of Welsh identity. European Journal of Social Psychology, 3, 447–460. doi:10.1002/ejsp.2420030407
Bowie, F. (1993). Wales from within: Conflicting interpretations of Welsh identity. In S. Macdonald (Ed.), Inside European identities (pp. 167–193). Oxford: Berg.
Branscombe, N. R., Ellemers, N., Spears, R., & Doosje, B. (1999). The context and content of social identity threat. In N. Ellemers, R. Spears, & B. Doosje (Eds.), Social identity: Context, commitment, content (pp. 35–58). Oxford, England: Blackwell.
Byrne, B. M. (2001). Structural equation modeling with AMOS: Basic concepts, applications and programming. Mahwah, NJ: Erlbaum.
Cameron, J. (2004). A three-factor model of social identity. Self and Identity, 3, 239–262. doi:10. 1080/13576500444000047
Cloke, P., Goodwin, M., & Milbourne, P. (1998). Cultural change and conflict in rural Wales: Competing constructs of identity. Environment and Planning, 30, 463–480. doi:10.1068/ a300463
Cohen, R. (2000). The incredible vagueness of being British/English. International Affairs, 76 , 575–582. doi:10.1111/1468-2346.00152
Coupland, N., Bishop, H., Evans, B., & Garrett, P. (2006). Imagining Wales and the Welsh language: Ethnolinguistic subjectivities and demographic flow. Journal of Language and Social Psychology, 25, 351–376. doi:10.1177/0261927X06292803
Coupland, N., Bishop, H., & Garrett, P. (2003). Home truths: Globalisation and the iconising of Welsh in a Welsh-American newspaper. Journal of Multilingual and Multicultural Development, 24, 153–177. doi:10.1080/01434630308666495
766 Andrew G. Livingstone et al.
Coupland, N., Bishop, H., Williams, A., Evans, B., & Garrett, P. (2005). Affiliation, engagement, language use and vitality: Secondary school students’ subjective orientations to Welsh and Welshness. International Journal of Bilingual Education and Bilingualism, 8, 1–24.
R. J. Crisp, & M. Hewstone (Eds.), (2006). Multiple social categorization: Processes, models and applications. Hove: Psychology Press.
Davies, J. (1994). A history of Wales. London: Penguin. Doosje, B., Ellemers, N., & Spears, R. (1995). Perceived intragroup variability as a function of
group status and identification. Journal of Experimental Social Psychology, 31, 410–436. doi:10.1006/jesp.1995.1018
Drury, J., & Reicher, S. (2000). Collective action and psychological change: The emergence of new social identities. British Journal of Social Psychology, 39, 579–604. doi:10.1348/ 014466600164642
Ellemers, N., Van-Knippenberg, A., & Wilke, H. (1990). The influence of permeability of group boundaries and stability of group status on strategies of individual mobility and social change. British Journal of Social Psychology, 29, 233–246.
Giles, H., Bourhis, R., & Taylor, D. M. (1977). Towards a theory of language in ethnic group relations. In H. Giles (Ed.), Language, ethnicity and intergroup relations (pp. 307–348). London: Academic Press.
Giles, H., Taylor, D. M., & Bourhis, R. Y. (1977). Dimensions of Welsh identity. European Journal of Social Psychology, 7, 165–174. doi:10.1002/ejsp.2420070205
Holmbeck, G. N. (1997). Toward terminological, conceptual and statistical clarity in the study of mediators and moderators: Examples from the child-clinical and paediatric psychology literatures. Journal of Clinical and Consulting Psychology, 65, 599–610. doi:10.1037/ 0022-006X.65.4.599
Hornsey, M. J., & Hogg, M. A. (2000). Assimilation and diversity: An integrative model of subgroup relations. Personality and Social Psychology Review, 4, 143–156. doi:10.1207/ S15327957PSPR0402 03
Jetten, J., Branscombe, N. R., & Spears, R. (2006). Living on the edge: Dynamics of intragroup and intergroup rejection experiences. In R. Brown & D. Capozza (Eds.), Social identities: Motivational, emotional and cultural influences (pp. 91–107). London: Sage.
Jetten, J., Spears, R., & Postmes, T. (2004). Intergroup distinctiveness and differentiation: A meta- analytic integration. Journal of Personality and Social Psychology, 86 , 862–879. doi:10.1037/ 0022-35188.8.131.522
Klandermans, B. (1997). The social psychology of protest. Oxford: Basil Blackwell. Kline, R. B. (2005). Principles and practice of structural equation modelling (2nd ed.). New
York: Guildford Publications. Langlands, R. (1999). Britishness or Englishness? The historical problem of national identity in
Britain. Nations and Nationalism, 5, 53–69. doi:10.1111/j.1354-5078.1999.00053.x Lewin, K. (1948). Resolving social conflicts. New York: Harper. Livingstone, A. G., Spears, R., & Manstead, A. S. R. (2009). The language of change? Characterisa-
tions of ingroup social position, threat, and the deployment of ‘distinctive’ group attributes. British Journal of Social Psychology, 48, 295–311. doi:10.1348/014466608X329533
Livingstone, A. G., Spears, R., Manstead, A. S. R., & Bruder, M. (2009). Illegitimacy and identity threat in (inter)action: Predicting intergroup orientations among minority group members. British Journal of Social Psychology, 48, 755–775. doi:10.1348/014466608X398591
Mann, R. (2007). Negotiating the politics of language: Language learning and civic identity in Wales. Ethnicities, 7, 208–224. doi:10.1177/1468796807076845
Marques, J. M., & Yzerbyt, V. Y. (1988). The black sheep effect: Judgemental extremity towards ingroup members in inter- and intra-group situations. European Journal of Social Psychology, 18, 287–292. doi:10.1002/ejsp.2420180308
Marques, J. M., Yzerbyt, V. Y., & Leyens, J. (1988). The black sheep effect: Judgmental extremity towards ingroup members as a function of group identification. European Journal of Social Psychology, 18, 1–16. doi:10.1002/ejsp.2420180102
Minority identities and group goals 767
Minescu, A., & Funke, F. (2010). Group size and support for political goals in the Russian federation: The mediating role of social identifications and perceived threat. Manuscript submitted for publication
Morton, T. A., Postmes, T., & Jetten, J. (2007). Playing the game: When group success is more important than downgrading deviants. European Journal of Social Psychology, 37, 599–616. doi:10.1002/ejsp.385
Mummendey, A., & Wenzel, M. (1999). Social discrimination and tolerance in intergroup relations: Reactions to intergroup difference. Personality and Social Psychology Review, 3, 158–174. doi:10.1207/s15327957pspr0302 4
Osmond, J. (1985). The national question again: Welsh political identity in the 1980s. Llandysul: Gomer Press.
Reicher, S. (2000). Against methodolatry: Some comments on Elliot, Fischer, and Rennie. British Journal of Clinical Psychology, 39, 1–6. doi:10.1348/014466500163031
Reicher, S., & Haslam, S. A. (2006). Rethinking the psychology of tyranny: The BBC prison study. British Journal of Social Psychology, 45, 1–40. doi:10.1348/014466605X48998
Reicher, S., & Hopkins, N. (1996a). Seeking influence through characterizing self-categories: An analysis of anti-abortionist rhetoric. British Journal of Social Psychology, 35, 297–311.
Reicher, S., & Hopkins, N. (1996b). Self-category constructions in political rhetoric: An analysis of Thatcher’s and Kinnock’s speeches concerning the British miners’ strike (1984–5). European Journal of Social Psychology, 26 , 353–371. doi:10.1002/(SICI)1099-0992(199605)26:3 < 353::AID-EJSP757>3.0.CO;2-O
Reicher, S., & Hopkins, N. (2001). Self and nation. London: Sage. Scheepers, D., Spears, R., Doosje, B., & Manstead, A. S. R. (2002). Integrating identity and
instrumental approaches to intergroup differentiation: Different contexts, different motives. Personality and Social Psychology Bulletin, 28, 1455–1467. doi:10.1177/014616702237574
Scheepers, D., Spears, R., Doosje, B., & Manstead, A. S. R. (2006). Diversity in in-group bias: Structural factors, situational features, and social functions. Journal of Personality and Social Psychology, 90, 944–960. doi:10.1037/0022-35184.108.40.2064
Schmitt, M. T., Spears, R., & Branscombe, N. R. (2003). Constructing a minority group identity out of shared rejection: The case of international students. European Journal of Social Psychology, 33, 1–12. doi:10.1002/ejsp.131
Sherif, M. (1966). Group conflict and co-operation: Their social psychology. London: Routledge and Kegan Paul.
Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and nonexperimental studies: New procedures and recommendations. Psychological Methods, 7, 422–445. doi:10.1037/ 1082-989X.7.4.422
Simon, B., Loewy, M., Sturmer, S., Weber, U., Freytag, P., Habig, C., . . . Spahlinger, P. (1998). Collective identification and social movement participation. Journal of Personality and Social Psychology, 74, 646–658. doi:10.1037/0022-35220.127.116.116
Spears, R., Jetten, J., & Scheepers, D. (2002). Distinctiveness and the definition of collective self: A tripartite model. In A. Tesser, J. V. Wood, & D. A. Stapel (Eds.), Self and motivation: Emerging psychological perspectives (pp. 147–171). Washington, DC: American Psychological Association.
Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In W. G. Austin & S. Worchel (Eds.), The social psychology of intergroup relations (pp. 33–47). Monterey, CA: Brooks-Cole.
Thompson, A., & Day, G. (1999). Situating Welshness: ‘Local’ experience and national identity. In R. Fevre, A. Thompson (Eds.), Nation, identity and social theory: Perspectives from Wales (pp. 27–47). Cardiff: University of Wales Press.
Trosset, C. S. (1986). The social identity of Welsh learners. Language in Society, 15, 165–192. doi:10.1017/S0047404500000178
Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. New York, NY: Basil Blackwell.
768 Andrew G. Livingstone et al.
van Morgan, S. (2006). Language politics and regional nationalist mobilization in Galicia and Wales. Ethnicities, 6 , 451–475. doi:10.1177/1468796806070718
Van Zomeren, M., Postmes, T., & Spears, R. (2008). Toward an integrative social identity model of collective action: A quantitative research synthesis of three socio-psychological perspectives. Psychological Bulletin, 134, 504–535. doi:10.1037/0033-2909.134.4.504
Welsh Language Board (2004). Percentage able to speak Welsh, 2001: All aged 3 and over. Retrieved from http://www.byig-wlb.org.uk/English/publications/Publications/1339.JPG
Welsh Language Board (2008). The Welsh language use surveys of 2004-06 . Retrieved from http://www.byig-wlb.org.uk/English/publications/Publications/Welsh%20Language%20Use% 20Surveys%202004-06.pdf
Received 1 May 2009; revised version received 26 July 2010
Copyright of British Journal of Social Psychology is the property of Wiley-Blackwell and its content may not
be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.