3V\FKRORJLFD Asbrock, F and Van Hiel, A (2017). An Insiders’ Outside Perspective on the Flemish-Walloon Conflict: The Role of Identification and ψ %HOJLFD Disidentification for the German-Speaking Minority.Psychologica Belgica, 57(3), pp. 115–131, DOI: https://doi.org/10.5334/pb.347

RESEARCH ARTICLE An Insiders’ Outside Perspective on the Flemish-Walloon Conflict: The Role of Identification and Disidentification for the German-Speaking Minority Frank Asbrock* and Alain Van Hiel†

In this study we analyzed the opinions of citizens of the German-speaking minority in Belgium on the linguistic conflict between the Walloons and the Flemish, as well as their attitudes towards these ­linguistic ­communities. We were especially interested in the effects of identification with the local community and disidentification with Belgium. We distributed a survey questionnaire in Eupen, the capital of the ­German-speaking community, and received replies from 129 inhabitants. Results showed that identification with the German-speaking community was associated with positive attitudes towards the ­German-speaking ­community and with demands for more autonomy of the community within the federal Belgian state. Disidentification with Belgium was not positively correlated with these constructive and positive outcomes, but with negative perceptions of all three Belgian communities, the perception of strong conflicts among these communities, and demands for the separation of the Belgian federal state into independent regions. The results are in line with previous research on these processes and point to unique, positive aspects of a strong local identity.

Keywords: Identification; disidentification; linguistic conflict; German-speaking community; conflict perception

In public perception, especially from an prevalent in the present special issue, which international viewpoint, Belgium is a country includes articles devoted to the linguistic which is divided into two linguistic and cul- conflict between the French-speaking and turally distinct subgroups, the Flemish and the Flemish communities. However, Belgium the Walloon communities. This division is also has a small German-speaking commu- nity with about 76,300 inhabitants, which is located at the Belgian-German border. In * Chemnitz University of Technology, DE the linguistic conflict between the Dutch- † Ghent University, BE speaking and French-speaking communities, Corresponding author: Frank Asbrock the German-speaking community holds an ([email protected]) insider’s outside position – although they 116 Asbrock and Van Hiel: Identification and Disidentification are part of the superordinate group by being minority groups (González & Brown, 2006). Belgian, their ingroup is not involved in the Similar to previous studies of dual identities major societal conflict. Up to this point, the we were interested in identification processes perspective of the German-speaking commu- with subordinate and superordinate groups. nity has hardly received any scholarly atten- In contrast to existing studies, however, we tion, at least among (political) psychologists. focused on the effects of identification with The present study is driven by our interest the minority and disidentification with the in the role of identification processes among superordinate group. Recent research sug- this rather invisible minority in Belgium. gests that identification and disidentification Social identification is an important issue in with a particular group are distinct psycho- social psychological research and although logical states, which differentially relate to it has already been investigated in the con- attitudes and behavior toward this group text of the linguistic division of the Belgian (Becker & Tausch, 2014). We therefore aimed communities, the existing research has at analyzing these differential relationships been exclusively focused on the Flemish for identification with a subordinate group and Walloon communities as the two major (the German-speaking community) and disi- actors in this conflict (e.g., Billiet, Jaspaert, dentification with a superordinate category & Swyngedouw, 2012; Klein, Licata, Van (the Belgian state). Here, our study adds to der Linden, Mercy, & Luminet, 2012). As a the small but growing body of research on ­community in Belgium, which has not been disidentification, with a special emphasis on directly involved in the conflict, but is none- a minority ingroup’s perspective. Secondly, theless affected by reforms negotiated by we focused on the interesting but uninves- the Flemish and Walloon communities, we tigated perspective of the German-speaking consider it important to add a focus on the community on the linguistic conflict and German-speaking community to the existing the other communities in Belgium. Before body of research. Given this unique insider’s elaborating on our theoretical perspective outside perspective, we want to focus our and reviewing previous research on identi- analysis on the relationship between iden- fication and disidentification, we will briefly tification processes and attitudes towards introduce the linguistic conflict and the situ- the linguistic conflict, as well as towards the ation of the German-speaking community in other linguistic communities. Belgium in the following section. The goal of the present paper was ­twofold. Firstly, we were interested in the relation- The historical-political context of ship between the identification with the the German-speaking community ­German-speaking community and the Belgium’s German-speaking territories have disidentification with Belgium and the per- only been part of Belgium since Germany’s ception of the linguistic conflict, the superor- defeat in World War I (1914–1918). The treaty dinate category (i.e., Belgium), and the other of Versailles postulated that the Eupen- two sub-groups within this category (i.e., the Malmédy region and Moresnet had to be Flemish and Walloons). We build upon a dual handed over to Belgium in order to compen- identity perspective (e.g., Gaertner, Dovidio, sate for the losses and damages caused by the Guerra, Hehman, & Saguy, 2016; González & war. Two decades later World War II (1939– Brown, 2003), which indicates that minority 1945) began and Germany again occupied group members can have a positive minor- these territories. Its inhabitants were consid- ity identity, as well as a positive identity as ered to be German, and they often referred a member of the superordinate group. Dual to themselves as Germans (Wenselaers, identities have been shown to have paci- 2008). Young men had to join the German fying effects, as they can reduce ingroup armed forces, and not less than 3,200 of bias (Hornsey & Hogg, 2000), especially for them would never return to their homes. Asbrock and Van Hiel: Identification and Disidentification 117

Following the defeat of Germany in 1945, held by the Walloon Region, such as social the territories again became part of Belgium, policy and (public) transport (Wenselaers, and, as a result of alleged collaboration with 2008). The German-speaking community’s Nazi Germany, the Belgian and Walloon important political parties are associated authorities attempted to de-Germanize the with the respective mother parties in the local population (Dewulf, 2009). Almost half Walloon region, and they constitute local of the population faced juridical procedures lists of Christian Democrat, Green, Liberal and one sixth of them were imprisoned (DG, and Socialist parties. The ProDG as the sole 2016). These procedures were generally regional party emphasizes their independ- seen as exaggerated and unfair and the local ence from the Walloon and Belgian establish- population felt that the Belgian authorities ment as a selling point to its voters (ProDG, showed little understanding for the spe- 2016). cific situation of these territories (see van Istendael, 2012; Wenselaers, 2008). Social Identification and Nowadays, Belgium is a federal state, which disidentification is composed of communities and regions. It Membership in social groups has strong has three communities: the Dutch, French, effects on the individual (Tajfel & Turner, and German-speaking Community. The com- 1979). People tend to systematically value munities have internal autonomy regarding others more when they are perceived as policies related to language and culture in a being members of the same category. broad sense. Belgium also consists of three However, not all group memberships are regions: the Flemish, Walloon, and Brussels equally important: Individuals can identify regions. Regions have a say in economic more or less strongly with various groups, issues. The German-speaking part of Belgium which in turn affects the social consequences thus comprises a community in itself, with of group membership (Ellemers, Spears, & autonomy in language and cultural issues, Doosje., 1999; Leach et al., 2008). One of but at the same time it is part of the Walloon the best studied consequences of identifi- region, which decides on economic­ issues cation with a group is ingroup bias (Tajfel, (see also, Klein et al., 2012). Billig, Bundy, & Flament, 1971), although The first state reform was implemented in other effects have been noted as well. As an 1973. Over the years, however, not less than example, though minority group members six state reforms have been implemented. show an increase in perceived discrimina- These reforms have been predominantly tion with higher ingroup identification (e.g., driven by the desire for autonomy of the Verkuyten & Yildiz, 2007), they also report Dutch and French-speaking populations. Yet, weaker effects of discrimination experi- it should be acknowledged that the desire ences on well-being (Schmitt, Branscombe, for further state reform has more and more Postmes, & Garcia, 2014). Moreover, ingroup become a Flemish demand, while Walloon identification is connected with ingroup politicians have been typically opposing such support, but not necessarily with outgroup reforms lately due to the apprehension that derogation (Brewer, 1999). Research has they might lead to the abolishment of the shown that patriotism, which is a positive Belgian state. As a possible side effect, the attitude toward one’s own country, is either succession of reforms resulted in increased negatively or not at all associated with the autonomy for the small German-speaking derogation of outgroups such as foreigners. community (Dewulf, 2009). In the past few Nationalism, which implies dominance and years, however, there was a broad consensus superiority over other nations, is however among political parties to strive for greater positively correlated with outgroup deroga- autonomy. This would lead to the transfer tion (e.g., Wagner, Becker, Christ, Pettigrew, of some competences that are currently & Schmidt, 2012). 118 Asbrock and Van Hiel: Identification and Disidentification

In the case of Belgium, the relevant groups states. In a recent review, Becker and Tausch can be nested within a hierarchical structure. (2014) described disidentification as a multi- Individuals are simultaneously members of a dimensional construct, comprising detach- subordinate and superordinate group. One ment from the group, dissatisfaction with way to effectively deal with this state of group membership, and perceived dissimilarity to membership is to develop dual identities. The other group members. Various researchers Dual Identity Model is an integration of the have demonstrated that disidentification Common Ingroup Identity Model (Gaertner indeed differs from non-identification, and et al., 2016), which stresses the importance therefore disidentification is not the mere of recategorization processes that result into opposite of identification with a group (e.g., a common superordinate group, and the Becker & Tausch, 2014; Ikegami, 2010; Long Mutual Intergroup Differentiation Model & Spears, 1997; Verkuyten & Yildiz, 2007). (Brown & Hewstone, 2005). The Mutual Unlike non-identification, which is a rather Intergroup Differentiation Model empha- neutral process without emotional involve- sizes the importance of maintaining group ment towards the respective group, disiden- boundaries on a subordinate group level. tification implies a high affective investment Thus, dual identities allow for the accept- and an active separation from the group ance of the minority group’s distinctive- (Becker & Tausch, 2014; Dean, 2008). ness while simultaneously being a part of Ikegami and Ishida (2007) showed that the ­superordinate ­majority. These identi- ­disidentification has stronger relationships ties reduce intergroup bias in ­members of with negative evaluations of the ingroup minority groups (González & Brown, 2006). compared to low identification. Also, Becker Hornsey and Hogg (2000) showed that inter- and Tausch (2014) demonstrated that disi- group bias is lowest when both superordi- dentification correlated stronger with nega- nate and subordinate categories are salient tive behavioral intentions, as well as negative and highest when only the superordinate emotions, towards the ingroup than iden- category is salient. tification. Identification, however, showed While dual identities are effective in stronger correlations with positive emotions reducing intergroup bias, minority group and behavioral intentions. members do not always identify with their In the present study, we aimed at ingroup as well as with the superordinate ­examining to what extent members of the group. In a study including samples of ­German-speaking community identify with Turkish-Dutch Muslims, Verkuyten and Yildiz their ingroup and simultaneously disidentify (2007) showed that ethnic and religious with the superordinate group in the form of identification was negatively and positively the federal Belgian state. Our research focus correlated with identification and disidenti- differs from previous studies, which analyzed fication with the superordinate group (i.e., dual identities (but not disidentification; e.g., the Dutch), respectively. Disidentification González & Brown, 2006; Hornsey & Hogg, describes the psychological phenomenon of 2000) or ethnic/religious identification belonging to a group one does not want to and national disidentification (Verkuyten & belong to and which poses a threat to one’s Yildiz, 2007). We argue that ingroup iden- identity (Becker & Tausch, 2014; Dean, 2008). tification does not necessarily lead to disi- Various studies did not distinguish between dentification with the superordinate group non-identification or low identification and (see Verkuyten & Yildiz, 2007). Instead, as disidentification (e.g., Becker, Tausch, Spears, suggested by Becker and Tausch (2014), we & Christ, 2011; Jasinskaja-Lathi, Liebkind, expected identification and disidentifica- & Solheim, 2009). Nevertheless, a growing tion to be uncorrelated. For our present body of research indicates that non- and research, this means that a high identifica- disidentification are distinct psychological tion with the German-speaking community Asbrock and Van Hiel: Identification and Disidentification 119 does not indicate disidentification with German-speaking community and the two Belgium. Both dimensions should rather be other communities (Hypothesis 4). ­independent and therefore uniquely pre- In sum, we predicted that the concepts dict perceptions of and attitudes toward the of identification with the minority ingroup linguistic conflict in Belgium as well as the and disidentification with the superordinate groups involved. group are psychologically distinct and should therefore show differentiated relations to Hypotheses conflict perceptions and with group evalua- Regarding the identification with the tions. Identification and disidentification can German-speaking minority, we hypothesized play distinct roles in the perception of the that a higher identification is related to linguistic conflict in Belgium and allow for more positive attitudes towards the ingroup a differentiated analysis of the perspective of (Hypothesis 1). In line with previous research members of the German-speaking commu- (Becker & Tausch, 2014), we hypothesized nity in Belgium on the linguistic conflict. that disidentification with Belgium is associ- To our knowledge, there is hardly any ated with a negative perception of the two ­psychological research on the perspective main groups involved in the linguistic conflict of the German-speaking community on the (Hypothesis 2a). As the Walloon and Flemish linguistic conflict and the groups involved. communities are both part of the superor- As outlined above, the German-speaking dinate group, a general negative ­evaluation minority profited from the reforms driven of Belgium should also result in negative by the Dutch and French-speaking com- evaluations of its subgroups. Moreover, we munities (Dewulf, 2009). Recently, political expected disidentification to be positively parties showed a broad consensus regard- associated with the perception of a stronger ing the strive for even greater autonomy. linguistic conflict between the communities Therefore, we were interested in the rela- (Hypothesis 2b). As identification focuses tionships of identification with the German- on the ingroup and does not necessarily speaking minority and disidentification with imply outgroup derogation (Brewer, 1999), Belgium with attitudes towards political we expected the effects of disidentification reforms regarding the future of the three to be stronger than any effects of identifica- communities and the Belgian federal state. tion with the German-speaking­ community A related aspect of interest was the relation- on attitudes towards the two main groups ships of identification and disidentification involved in the conflict (Hypothesis 3a) as with political party preferences. In contrast well as on the perception of the conflict itself to the majority of the important parties in (Hypothesis 3b). the German-speaking community, which are The next hypothesis builds upon the out- related to their mother parties in the Walloon sider perspective of the German-speaking region, the ProDG is a strong regional party, community. As outlined above, the German- which most clearly strives for more auton- speaking community is not directly involved omy of the German-speaking community. in the linguistic conflict. Higher identifica- Therefore, a positive correlation between tion with the subordinate ingroup, as well identification with the German-speaking as stronger disidentification with the super- community and preference for ProDG can ordinate Belgian state, should intensify this certainly be expected. Although we did not outsider perspective. We therefore predicted derive specific hypotheses about these atti- that both disidentification and identifica- tudes and party preferences, we were inter- tion are more strongly correlated with the ested in the associations from an exploratory perception of the linguistic conflict as a con- perspective. flict between the Walloons and the Flemish We aimed at testing our hypotheses regard- community than as a conflict between the ing identification and disidentification in a 120 Asbrock and Van Hiel: Identification and Disidentification survey administered to households in the The German version of the scale (see, Becker German-speaking community. & Tausch, 2014) showed sufficient internal consistency (M = 3.52; SD = .88; α = .94). Methods Sample items include “I feel committed with Participants and procedure the German-speaking community” and “I Data were collected in Eupen, the capital of often think about the fact that I am a mem- the German-speaking region in Belgium. We ber of the German-speaking community.” distributed 1,000 envelopes containing the Disidentification with Belgium. questionnaire in the inhabitants’ mailboxes Disidentification was measured with ten in various streets in and nearby the city items from a scale developed by Becker and center. An accompanying letter introduced Tausch (2014) using a five-point Likert scale the study stating that we were interested (1 = totally disagree; 5 = totally agree). The in the beliefs and attitudes about a num- scale assesses group detachment, dissimi- ber of political issues of people living in the larity, and dissatisfaction, and showed good German-speaking community. The address- internal consistent scores (M = 1.34; SD = .46; ees were invited to complete the question- α = .79). Sample items are “I feel a distance naires, which they could return without between myself and the Belgians” and “I have any costs. In order to increase the response nothing in common with most Belgians.” rate participants could win a gift certificate Party preference. Participants indicated of 100 euro, per 50 returned envelopes. A how much they agreed with the political total of 129 questionnaires were returned. program of the CSP (Christian Democrats; Seven persons with German nationality as M = 3.89; SD = 1.76), Ecolo (Green party; well as one person with British nationality M = 3.77; SD = 1.60), SP (Social Democrats; were excluded from the analysis, leaving a M = 3.21; SD = 1.62), PFF (Liberal party; total sample of 121 participants (113 with M = 3.41; SD = 1.73), ProDG (Regional Belgium nationality, 8 unknown), including party M = 3.78; SD = 1.84), or other parties 76 men and 45 women, with a mean age of (M = 2.54; SD = 1.95) via Likert scales from 1 49 years (SD = 17.4). A total of 117 partici- (not at all) to 7 (fully). pants stated that German was the language Attitudes toward members of they used at home. A majority of the partici- ­different language groups and Germans. pants (N = 73) completed higher education Participants provided thermometer ­ratings and 43 participants completed secondary for various groups (0 = cold, ­negative; education. One person only finished primary 100 = warm, positive). Mean ratings education. Most of the respondents worked were 65.25 (SD = 22.38) for the Wallons, fulltime (N = 74) and a significant number 66.64 (SD = 21.52) for the Flemish, 75.25 of participants were retired (N = 24). Also (SD = 18.15) for the German-speaking com- included were a number of part time workers munity, 60.61 (SD = 22.68) for inhabitants (N = 7), unemployed people (N = 4), students of Brussels, and finally 60.76 (SD = 22.44) for (N = 5), and one housewife. Germans living in Germany. Linguistic conflict perceptions. Measures Participants rated the amount of conflict Identification with the German-speaking between the different language groups on community. We assessed identification with a thermometer ranging from 0 (peaceful, 14 items of Leach et al.’s (2008) instrument, no conflict) and 100 (strong conflict). The which measures identification in terms of means values were 66.27 (SD = 23.20) for group solidarity, satisfaction, centrality, self- the conflict between Flanders and Wallonia, stereotyping, and the perception of in-group 43.28 (SD = 25.38) for the conflict between homogeneity by means of five-point Likert Wallonia and the German-speaking commu- scales (1 = totally disagree; 5 = totally agree). nity, and 30.60 (SD = 22.94) for the conflict Asbrock and Van Hiel: Identification and Disidentification 121 between Flanders and the German-speaking the model including the significant paths community. only. The model shows a good fit (χ2 = 6.612, Political reform. We asked participants to df = 6, p = .36; CFI = .995, RMSEA = .030, react to seven statements regarding possible SRMR = .034). In line with Hypothesis 1, iden- state reforms of Belgium. They indicated the tification with the German-speaking commu- extent to which they desired the implemen- nity was positively associated with attitudes tation of these reforms on a scale from 1 to towards the ingroup (β = .36, p < .001), but 5 (1 = not desirable at all; 5 very desirable). not with attitudes toward the other two The items were: (1) The organization of the groups. As predicted in Hypothesis 2a, disi- federal Belgian state will remain unchanged dentification had a negative effect on atti- (M = 3.27; SD = 1.37); (2) The language groups tudes towards the Walloon (β = –.48, p < .001) will acquire more autonomy, but the Belgian and the Flemish communities (β = –.36, state will remain (M = 3.73; SD = 1.24); (3) p = .003), but also on attitudes toward the Belgium will be divided into independ- ­German-speaking community (β = –.22, ent regions (M = 1.30; SD = .88); (4) The p = .01). These results support our assump- German-speaking community will become tion that disidentification with Belgium has a a distinct region (M = 3.00; SD = 1.49); (5) stronger negative effect on attitudes towards The German-speaking community will the Walloon and Flemish communities than join Germany (M = 1.27; SD = .80); (6) The identification with the German-speaking German-speaking community will join the community (Hypothesis 3a).1 Walloon region (M = 1.41; SD = .89); and (7) Conflict perceptions. We tested our pre- The German-speaking community will join dictions regarding conflict perception in the Flemish region (M = 1.49; SD = 1.02). a path model using the same predictors as for attitudes and included the perception of Results the linguistic conflict between the Walloon We tested our hypotheses in separate path and Flemish communities, Walloons and models for conflict perception and atti- the German-speaking community, and the tudes toward the different language groups. Flemish and German-speaking communi- In addition, we ran separate path analyses ties simultaneously as dependent variables. for party preferences and attitudes toward Figure 2 shows the model with significant ­political reform. We controlled for age, paths only, which indicates a good model ­gender, and education for all analyses and fit (χ2 = 4.470, df = 6, p = .61; CFI = 1.000, used robust maximum likelihood estima- RMSEA = .000, SRMR = .037). As predicted tors. As implemented in Mplus (Muthén in Hypothesis 2b, disidentification with & Muthén, 1998–2015) we used full infor- Belgium was associated with stronger con- mation maximum likelihood to handle flict perceptions between the Walloons and missing data (Schäfer & Graham, 2002). the Flemish community (β = .23, p = .001), Table 1 shows intercorrelations of all varia- the Walloons and the German-speaking bles for hypothesis testing and demonstrates community (β = .33, p < .001), and the that disidentification with Belgium is not sig- Flemish and the German-speaking commu- nificantly correlated with identification with nities (β = .30, p = .02). Identification had an the German-speaking community (r = .08, effect on perception of the conflict between p = .38). the Walloon and the Flemish community Attitudes. We regressed attitudes toward (β = .17, p = .02). Supporting Hypothesis the Walloons, the Flemish community, and the 3b, the relations between disidentifica- German-speaking community simultaneously tion and conflict perception between the on disidentification with Belgium, identifica- German-speaking community and the two tion with the German-speaking community, other regions were stronger than those and the control variables. Figure 1 shows between identification and these conflict 122 Asbrock and Van Hiel: Identification and Disidentification .01 .08 .22* –.10 –.11 –.10 –.05 –.04 –.08 .27** –.22* edu –.32*** .15 .01 .00 .07 .09 .04 .07 .06 .19* –.03 –.22* gender .11 .04 .04 –.13 –.03 –.05 –.05 –.18* –.23* age –.34*** .01 .04 –.14 < .05. .31** .68*** .47*** .58*** p –.29** –.42*** FT Bru < .01, * < .01, p .10 .13 .07 .04 .23* .20* .31** .42*** FT Ger < .001, ** < .001, p .05 .20* –.21* .41*** .35*** .57*** –.28** FT GSC .09 –.09 –.03 .48*** –.37*** –.38*** FT Flem –.04 –.04 –.21* –.26** –.48*** FT Wall .00 .24* .31** .50*** Con F-G .14 .32*** .53*** Con W-G .20* .25** Con W-F .08 Ident Correlations. = Conflict between Con W-G and Flemish community, = Conflict between Walloons Disident = Disidentification, Ident identification, Con W-F 121. Feeling thermometer = Feeling FT Wall between Flemish community and German-speaking community, = Conflict Con F-G community, and German-speaking ­ Walloons thermometer FT Ger = Feeling thermometer German-speaking community, FT GSC = Feeling thermometer Flemish community, FT Flem = Feeling Walloons, thermometer people from Brussel, gender (1 = female, 2 male). *** FT Bru = Feeling Germany, = Disident Ident Con W-F Con W-G Con F-G FT Wall FT Flem FT GSC FT Ger FT Bru age gender Table 1: N Asbrock and Van Hiel: Identification and Disidentification 123

Figure 1: Path model predicting attitudes toward outgroups and ingroup. All coefficients are standardized estimators. All reported effects are significant (p < .05). ap = .14.

Figure 2: Path model predicting perceptions of linguistic conflict. GSC = German-speaking community. All coefficients are standardized estimators. All reported effects are significant (p < .05). ap < .09. perceptions. To test the predicted difference paths of disidentification and identification between the relationships of disidentifica- on conflict perception between Walloons 2 tion and identification with conflict percep- and the Flemish community (Δχ SB = 3.734, tion between the Walloon and the Flemish df = 1, p = .053), indicating that disidenti- community, we compared a model where fication had a marginally stronger relation- the paths in question were constrained to ship with conflict perception between the be equal with our unconstrained model. We Walloon and the Flemish community than used a χ2-difference test to evaluate whether identification. the models differed significantly (Satorra Hypothesis 4 was only partially supported. & Bentler, 2001). The test revealed a mar- As expected, identification showed a stronger ginally significant difference between the relationship with conflict perceptions 124 Asbrock and Van Hiel: Identification and Disidentification between the Walloon and Flemish communi- federal Belgian state (β = –.22, p = .003), ties than with the two other conflict percep- with preference for a division of Belgium tions, however the effect of disidentification into independent regions (β = .34, p < .001), on the perception of the Walloon-Flemish and with preference for a unification of the conflict did not differ from the two other con- German-speaking community with Germany flicts. A χ2 difference test between the uncon- (β = .48, p < .001). Identification, but not strained model and a model with all three disidentification, was associated with a pref- paths from disidentification constrained to erence for more autonomy of the German- be equal did not reveal a significant differ- speaking community (β = .34, p < .001), 2 2 ence (Δχ SB = 1.568, df = 2, p = .46). and with opposition to a unification of the Party preferences. We analyzed the rela- German-speaking community with the tionship of identification and disidentifica- Walloon region (β = –.20, p = .03). Both disi- tion with preferences for political parties by dentification (β = .17, p = .03) and identifica- simultaneously regressing the preferences tion (β = .34, p < .001) were associated with for CSP, Ecolo, SP, PFF, and ProDG on disi- preference for the German-speaking commu- dentification, identification, and the control nity as a distinct region in Belgium. Age was variables. The model with significant paths associated with preference for an unchanged only showed a good model fit (χ2 = 5.184, organization of the federal Belgian state df = 4, p = .27; CFI = .984, RMSEA = .052, (β = .31, p < .001). SRMR = .039). Identification with the German-speaking community was associated Discussion with preference for CSP (β = .22, p = .01), Ecolo In the present paper we aimed at ­analyzing (β = .27, p = .003), SP (β = .22, p = .02) and, the perspective of an often neglected third most strongly, for ProDG (β = .56, p < .001). party in the Belgian linguistic conflict – the Disidentification with Belgium showed nega- perspective of the Germany-speaking minor- tive effects on preference for CSP (β = –.15, ity. Although this small community liv- p = .02), Ecolo (β = –.17, p = .04), and SP ing in the East of Belgium is not directly (β = –.23, p = .003). As the only control involved in the conflict, it is nevertheless variable, age was associated with preference affected by political decisions by the two for CSP (β = .33, p < .001). No other effects strong communities and the federal Belgian reached significance. state. Therefore, we considered the view of Political reform. As stated in the descrip- members of this community on the Belgian tion of the measures, the mean values for linguistic conflict to represent an insiders’ unifications of the German-speaking commu- outside perspective. This very specific set- nity with Germany, the Walloon or Flemish ting allowed us not only to conduct one communities, as well as the preference for of the first studies in political psychology independent regions, were very low and on the specific attitudes and perceptions indicated no strong wish for these options in of the German-speaking minority, but also general. A path analysis with simultaneous provided us with the opportunity to analyze regression of all political reform items on more general processes of identification and the predictor variables showed very differ- disidentification with a minority ingroup ent results for disidentification with Belgium and a superordinate group. Our research and identification with the German-speaking supports the distinction between disidenti- community. The model with significant paths fication and identification as separate con- only showed a good model fit (χ2 = 8.402, structs (Becker & Tausch, 2014; Verkuyten & df = 10, p = .59; CFI = 1.000, RMSEA = .000, Yildiz, 2007). Extending previous research SRMR = .041). Disidentification, but not (Becker & Tausch, 2014), we demonstrated identification, was associated with opposi- that identification with the German- tion to an unchanged organization of the speaking community is correlated with a Asbrock and Van Hiel: Identification and Disidentification 125 multitude of positive attitudes toward this ingroup (like in the case of Turkish-Dutch ingroup, but not with such attitudes toward Muslims; Verkuyten & Yildiz 2007), while a outgroups within the same superordinate positive relationship, namely a dual identity, category. Our results thus indicate that is a likely outcome of a successful re-catego- identification constitutes a positive vari- rization process (González & Brown, 2006). able. Conversely, disidentification with the However, as attested by the present results, superordinate group was related to negative it is also possible that both identity concepts attitudes toward the respective outgroups are almost unrelated. within the entire hierarchy, therefore sub- Indeed, the present pattern of results stantiating the theoretical argument that points to an interesting case of dual identity. disidentification is not the mere opposite Based on the Politicized Collective Identity of identification and therefore differs from Model (Simon & Klandermans, 2001), Simon non-identification. Our results indicate that and Grabow (2010) showed that dual identi- disidentification is rather a negative motiva- ties as members of a minority and a superor- tional state, expressing the active separation dinate majority are positively correlated with from an ingroup. This destructive compo- supporting political demands and actions nent of disidentification is illustrated by the within the limits of normative acceptance. positive relationship between disidentifica- For the present context, this means that the tion and support for abolishing the Belgium identification with the German-speaking federal state. community might go along with a critical, Additionally, disidentification and identifi- but constructive view on the federal state as cation were related to the perception of the well as its constituting regions. Our findings linguistic conflict in different ways: While support this view by showing that identifica- disidentification was related to the percep- tion with the German-speaking community tion of stronger conflicts among the three is positively related to demands for more communities, identification was positively autonomy for the region within a federal, correlated with the perception of a strong Belgian state, but not with derogation of the conflict between Walloons and the Flemish other regions or the perception of strong community only. Again, this result points to conflicts among all regions. This negative different motivational states expressed by perception is only related to disidentification identification and disidentification processes with Belgium. and to the necessity of separating these Positive associations between identifica- concepts. tion with the German-speaking community An important aspect in our study was and party preferences provide further sup- the simultaneous analysis of identification port for this idea, especially with a preference with a minority ingroup and disidentifica- for the ProDG, which aligns well with the tion with a superordinate group. Previous strong relation of identification and prefer- research on dual identities indicates that ence for autonomy of the German-speaking­ identification with both the ingroup and the community. Identification was also related superordinate group can reduce intergroup to increased preference of national par- bias (González & Brown, 2006; Hornsey & ties such as Ecolo and the Socialist Party, Hogg, 2000), but only little is known about although these correlations were of lesser identification and disidentification with dif- magnitude than for ProDG. However, this ferent groups within the same hierarchy. As indicates a preference for political action expected, identification with the German- within the limits of normative acceptance. speaking community was not correlated with Conversely, disidentification was negatively disidentification with Belgium. However, a related to preference for the national parties negative relationship may be likely when the CSP, Ecolo, and SP, and unrelated to prefer- superordinate group poses a threat to the ence for ProDG, which further underlines the 126 Asbrock and Van Hiel: Identification and Disidentification differentiation of identification and disiden- collaboration of German-speaking citizens tification processes. This latter finding sup- (although they were often forced to join the ports the notion that disidentification with German troops; see Dewulf, 2009). Finally, a the Belgian state does not lead to support third reason for the noteworthiness of this for a party that demands more autonomy of the observation is that in less than a century, the German-speaking community within the these territories have shifted from Germany federal state. to Belgium two times from one direction to In a broader theoretical sense, our findings the other and vice versa. Inhabitants of neu- add to the growing body of research on the tral Monseret – a small portion of these ter- important differentiation of identification ritories – have even changed nationality five and disidentification for fully understand- times and some citizens have served in two ing positive and negative aspects of ingroup different armies (see, Van Reybroeck, 2016). perception. Another important related observation is In the remainder of the discussion we first that, although the identification level with focus on the extreme low levels of disiden- the German-speaking community is fairly tification with Belgium, which we consider high (M = 3.52, SD = .88), at the same time noteworthy, given the German-speaking ter- this score is not too far from the scale’s theo- ritories’ short common history as a Belgian retical midpoint of 3, even though the differ- territory. Second, we discuss the policy impli- ence is significant, t(119) = .6.47, p < .001. A cations of increased identification with the possible reason for this is that the German- German-speaking community, putting the speaking territory itself is rather heterogene- present results in the context of possible ous. The northern territories around Eupen effects of identification in the other language have been part of the duchy of Limburg, communities. Finally, we pay attention to the while the southern territories around beneficial, constructive role identification Sankt-Vith have been part of the duchy of may play in Belgian policies. Luxemburg. Moreover, the local languages are different and there has not been much The last Belgians contact between the northern and southern Considering the present circumstances of a parts because of the presence of a desolate fierce debate between the dominant com- area in between them (Fonteyn, 2013). In munities constituting the large majority of an interview, Karl-Heinz Lamberts – the for- the Belgian population and political par- mer minister-president – declared that the ties (especially on the Flemish side) which German-speaking community needs to have have either asked for or directly suggested an identity, or put otherwise, is in search of the abolishment of Belgium, it is notewor- an identity as a community (in, Wenselares, thy that a small minority group expresses 2008). This lack of a strong basis for a com- such a low level of disidentification with mon German-speaking identity may be ben- the Belgian state, close to the scale’s mini- eficial to people’s identification with the mum value of 1 (M = 1.34, SD = .46) and superordinate structure. significantly different from the scale mid- point of 3, t(118) = –.38.91, p < .001. In Identification and attitudes about popular literature, the patriotism of the separation and the other language German-speaking citizens has been coined communities as ‘the last Belgians’ (van Istendael, 1989; We consider the present findings to be politi- Wenselaers, 2008). Another reason designat- cally relevant: The increased desire for more ing this finding as remarkable is the fact that autonomy among citizens of the German- these territories have been subject to rather speaking community does not imply an fierce ‘de-Germanization’ policies, especially increased desirability to abolish Belgium as after the Second World War and the alleged the superordinate state structure (r = –.01, Asbrock and Van Hiel: Identification and Disidentification 127 p = .89). Support for these two policies is and Belgian levels correlate positively (Billiet fueled differently by identification and disi- et al., 2003; Billiet et al., 2012). dentification processes: While identification With respect to their attitudes regarding with the German-speaking community was the other language communities, regional- related to support for more autonomy, but not ists of both the Dutch- and French-speaking to abolishing the superordinate Belgium state, communities tend to have a negative atti- disidentification with Belgium was related to tude towards (citizens of) the other region support for abolishing the superordinate state, (see, Nuttin, 1976) as they perceive them- but not for more autonomy. Along similar lines, selves to be victims of the respective other ProDG aims at strengthening the German- community (Klein et al., 2012). Hence, unlike speaking region, while the abolishment of the the German-speaking citizens, identification Belgian state is not included in their party plat- processes in Dutch- and French-speaking form (ProDG, 2016). It remains open whether citizens might go together with negative per- similar results would be obtained in the other ceptions of the other communities. Belgian language communities with regard to their attitudes about the Belgian state and the Limitations and directions for future other language communities. research With respect to their attitudes about the To our knowledge, the present study is Belgian state, one would be inclined to expect the first to analyze the German-speaking another pattern of results, especially for citi- community’s perspective on the linguis- zen living in the Flemish community. Indeed, tic conflict in Belgium under consideration Flemish regionalists tend to have negative of identification­ and disidentification pro- attitudes towards the Belgian state and they cesses. We were able to show unique rela- even have repeatedly stated that the disin- tions of identification with the ingroup and tegration of Belgium is their final goal (see, disidentification with Belgium, the super- Rihoux, Dumont, De Winter, Deruette, & Bol, ordinate group, with perceptions of the 2009). All of this is of great concern for most Walloon and Flemish communities, of the French-speaking politicians and citizens, as linguistic conflict, as well as with political a Flemish regionalist party currently is the attitudes and demands. Still, we are aware of largest Belgian party in terms of electoral sup- some limitations of our study, which should port, and it is even a leading party in the cur- be addressed in future research. First, we did rent national government. Due to the political not include measures of identification and platform of these regionalist parties and their disidentification for both the minority and large share of support, identification among superordinate groups. The selection of only Flemish citizens at the local level might cor- two measures instead of the full quadrant relate positively with disidentification at the was informed by the research question at Belgian level, and both these identification pro- hand, that is, the explicit aim to explain lin- cesses might relate to the wish to abolish the guistic division rather than to search for ele- Belgian state. Conversely, in the Walloon com- ments that bring unification and harmony munity the movement (i.e., ‘Rassemblement among the language groups. However, as we Wallonie-France’) for the (re-) unification of outlined above, it seems likely that members Wallonia with France attracts only 1% of the of the German-speaking community experi- votes, while no other regional parties advocate ence dual identities, which, in turn, should the abolishment of the Belgian state. In the predict a critical, but constructive relation to Walloon community, identification with the the federal Belgian state as well as the local local level should thus not be expected to go community. Even though our findings sup- along with disidentification with the Belgian port this view, we were not able to directly level. In this respect, previous studies have test it. We consider this a very important revealed that identification with the Walloon and fruitful research avenue, as it is not 128 Asbrock and Van Hiel: Identification and Disidentification only of theoretical interest but might indi- Notes cate a special status of the German-speaking 1 For exploratory reasons, we also tested community in perceptions of Belgium and if the interaction of disidentification commitment to the federal state. This pos- with Belgium and identification with the sible dual identity might differentiate the German-speaking community predicts German-speaking community from the other attitudes toward the different groups. communities, for which a high identification Explained variance did not increase with the local community might go along due to including the interaction term with low commitment to the federal state (ps > .44). and/or a negative attitude towards the other 2 Again, including the interaction of disi- language group. dentification with Belgium and identifi- Second, our study design is correlational cation with the German-speaking com- and does not allow for any conclusions munity did not explain any additional about causality. Even though experimental variance (ps > .24). evidence suggests that identification and disidentification predict attitudes and bias Competing Interests (e.g., Becker & Tausch, 2014), we cannot The authors have no competing interests to ­confirm this causal relationship with our declare. data. Consequently, we need experimental and longitudinal studies to replicate and References extend our findings. Becker, J. C., & Tausch, N. (2014). When Third, our sample is not representative Group Memberships are Negative: The for the German-speaking community in Concept, Measurement, and Behavioral Belgium. All participants were residents of Implications of Psychological Disidentifi- Eupen. However, our sample shows a quite cation. Self and Identity, 13, 294–321. DOI: large distribution of age, education and https://doi.org/10.1080/15298868.201 occupation. 3.819991 Becker, J. C., Tausch, N., Spears, R., Conclusion & Christ, O. (2011). Committed The perspective of the German-speaking dis(s)idents: Participation in radical minority in Belgium has not yet been in the collective action fosters disidentifica- focus of any political psychological research. tion with the broader in-group but Even though the German-speaking minority enhances political identification. Per- has an outsider position in the linguistic con- sonality and Social Psychology Bulletin,­ flict, identification with the community and 37, 1104–1116. DOI: https://doi. disidentification with the federal state have org/10.1177/0146167211407076 strong implications for the perception of Billiet, J., Jaspaert, E., & Swyngedouw, this conflict. We have shown that identifica- M. (2012). Diversity in the Relation- tion with the ingroup has rather constructive ship between Perceived Ethnic Threat, and positive effects, while disidentification Islamophobia, and (sub)national iden- with the Belgium state is correlated with tity in ­Belgium: a combination of two negative attitudes toward the other Belgian approaches. In: Salzborn, S., Davidov, E., communities as well as with support for & Reinecke, J. (Eds.), Methods, Theories, policies that aim to abolish the Belgian state. and Empirical Applications in the Social Identification and disidentification thus have Sciences. Festschrift for Peter Schmidt, differential implications, attesting to the 279–290. Wiesbaden: Springer VS. DOI: necessity to differentiate between these two https://doi.org/10.1007/978-3-531- processes. 18898-0_34 Asbrock and Van Hiel: Identification and Disidentification 129

Billiet, J., Maddens, B., & Beerten, R. (2003). of subgroup and superordinate group National identity and attitude toward for- ­identification in intergroup contact. eigners in a multinational state: a repli- European Journal of Social Psychology, 33, cation. Political Psychology, 24, 241–257. 195–214. DOI: https://doi.org/10.1002/ DOI: https://doi.org/10.1111/0162- ejsp.140 895X.00327 González, R., & Brown, R. (2006). Dual Brewer, M. B. (1999). The psychology of prej- identities in intergroup contact: Group udice: Ingroup love or outgroup hate? status and size moderate the generaliza- Journal of Social Issues, 55, 429–444. tion of positive attitude change. Journal DOI: https://doi.org/10.1111/0022- of Experimental Social Psychology, 42, 4537.00126 753–767. DOI: https://doi.org/10.1016/j. Brown, R., & Hewstone, M. (2005). An jesp.2005.11.008 Integrative Theory of Intergroup Con- Hornsey, M. J., & Hogg, M. A. (2000). tact. In: Zanna, M. P. (Ed.), Advances Subgroup relations: A comparison in ­experimental social psychology, 37, of mutual intergroup differentia- 255–343. San Diego, CA, US: ­Elsevier tion and common ingroup identity Academic Press. DOI: https://doi. models of prejudice reduction. Per- org/10.1016/S0065-2601(05)37005-5 sonality and Social Psychology Bul- Dean, J. (2008). The lady doth protest too letin, 26, 242–256. DOI: https://doi. much: Theorizing disidentification in org/10.1177/0146167200264010 contemporary gender politics. Ideology in Ikegami, T. (2010). Precursors and conse- Discourse Analysis, 24, 1–19. quences of ingroup disidentification: Dewulf, J. (2009). “O liebes Land”, “o Status system beliefs and social ­identity. ­Belgiens Erde”: The development of the Identity: An International Journal of German-speaking community in Belgium Theory and Research, 10, 233–253. DOI: reflected in the light of the Flemish strug- https://doi.org/10.1080/15283488.201 gle for autonomy. German Studies Review, 0.523590 32, 65–81. Ikegami, T., & Ishida, Y. (2007). Status hier- DG. (2016). Deutschsprachige Gemeinschaft archy and the role of disidentification in Belgiens. http://www.dg.be/en/desk- discriminatory perception of outgroups. topdefault.aspx/tabid-2809/5353_read- Japanese Psychological Research, 49, 34682/. 136–147. DOI: https://doi.org/10.1111/ Ellemers, N., Spears, R., & Doosje, B. j.1468-5884.2007.00340.x (Eds.) (1999). Social identity: Context, Jasinskaja-Lathi, I., Liebkind, K., & ­commitment, content. Oxford, England: ­Solheim, E. (2009). To identify or not to Blackwell Publishers. identify? National disidentification as an Fonteyn, G. (2013). De laatste Belgen. Article alternative reaction to perceived ethnic­ ­published in KNACK, January 14th, 2013. discrimination. Applied ­Psychology: Gaertner, S. L., Dovidio, J. F., Guerra, R., An international review, 58, 105–128. Hehman, E., & Saguy, T. (2016). A com- DOI: https://doi.org/10.1111/j.1464- mon ingroup identity: Categorization, 0597.2008.00384.x identity, and intergroup relations. In: Klein, O., Licata, L., Van der Linden, N., Nelson, T. D. (Ed.), Handbook of prejudice, Mercy, A., & Luminet, O. (2012). A stereotyping, and discrimination, 2nd ed., waffle-shaped model for how realistic 433–454. New York, NY, US: Psychology dimensions of the Belgian conflict struc- Press. ture collective memories and stereotypes. González, R., & Brown, R. (2003). Gener- Memory Studies, 5, 16–31. DOI: https:// alization of positive attitude as a function doi.org/10.1177/1750698011424028 130 Asbrock and Van Hiel: Identification and Disidentification

Leach, C. W., van Zomeren, M., Zebel, 921–948. DOI: https://doi.org/10.1037/ S., Vliek, M. L. W., Pennekamp, S. F., a0035754 Doosje, B., Ouwerkerk, J. W., & Spears, Simon, B., & Grabow, O. (2010). The R. (2008). Group-level ­self-definition ­politicization of migrants: Further evi- and self-investment:­ A hierarchical dence that politicized collective identity ­(multi-component) model of in-group is a dual identity. Political Psychology, 31, identification. Journal of ­Personality 717–738. DOI: https://doi.org/10.1111/ and Social Psychology, 95, 144–165. j.1467-9221.2010.00782.x DOI: https://doi.org/10.1037/0022- Simon, B., & Klandermans, B. (2001). 3514.95.1.144 Politicized collective identity: A social Long, K., & Spears, R. (1997). The self-esteem psychological analysis. American hypothesis revisited: Differentiation and Psychologist, 56, 319–331. DOI: https:// the disaffected. In: Spears, R., Oakes, P. J., doi.org/10.1037/0003-066X.56.4.319 Ellemers, N., & Haslam, S. A. (Eds.), The Tajfel, H., Billig, M. G., Bundy, R. P., & social psychology of ­stereotyping and ­Flament, C. (1971). Social ­categorization group life, 296–317. Oxford: Blackwell. and intergroup behavior. ­European Muthén, L. K., & Muthén, B. O. (1998–2015). Journal of Social Psychology, 1, Mplus User’s Guide (7th ed). Los Angeles, 149–178. DOI: https://doi.org/10.1002/ CA: Muthén & Muthén. ejsp.2420010202 Nuttin, J. (1976). Het stereotiep beeld Tajfel, H., & Turner, J. C. (1979). An inte- van walen, vlamingen en brusselaars: grative theory of intergroup conflict. In: hun kijk op zichzelf en op elkaar. Een ­Austin, W. G., & Worchel, S. (Eds.), The empirisch onderzoek bij universitairen social psychology of intergroup relations, [The stereotypical­ image of Walloons, 33–47. Monterey, CA: Brooks-Cole. ­Flemings and inhabitants of Brussels: van Istendael, G. (1989). Het Belgisch their look on themselves and on each ­labyrinth; Wakker warden in een ander other]. Brussels, Belgium: Royal Academy land. Amsterdam/Antwerpen: Uitgeverij of Letters and Science. Arbeiderspers. ProDG. (2016). Über uns. Retrieved from: van Istendael, G. (2012). Het Belgisch http://prodg.be/uber-uns/. ­labyrinth. Amsterdam, the Netherlands: Rihoux, B., Dumont, P., De Winter, L., Atlas. Deruette, S., & Bol, D. (2009). Belgium. Van Reybroeck, D. (2016). Zink. de Bezige European Journal of Political Research, 48, Bij: Amsterdam, the Netherlands. 903–912. DOI: https://doi.org/10.1111/ Verkuyten, M., & Yildiz, A. A. (2007). j.1475-6765.2009.01852.x National (dis)identification, and Satorra, A., & Bentler, P. M. (2001). ­ethnic and religious identity: A study A scaled difference chi-square test among Turkish-Dutch Muslims. Per- ­statistic for moment structure ­analysis. sonality and Social Psychology Bulletin,­ Psychometrika,­ 66, 507–514. DOI: 33, 1448–1462. DOI: https://doi. https://doi.org/10.1007/BF02296192 org/10.1177/0146167207304276 Schäfer, J. L., & Graham, J. W. (2002). Miss- Wagner, U., Becker, J. C., Christ, ing Data: Our View of the State of the O., ­Pettigrew, T. F., & Schmidt, P. Art. Psychological Methods, 7, 147–177. (2012). A Longitudinal test of the DOI: https://doi.org/10.1037/1082- relation between German national- 989X.7.2.147 ism, patriotism, and ­outgroup dero- Schmitt, M. T., Branscombe, N. R., gation. European­ Sociological­ Review, ­Postmes, T., & Garcia, A. (2014). The con- 28, 319–332. DOI: https://doi. sequences of perceived ­discrimination for org/10.1093/esr/jcq066 ­psychological well-being: A meta-analytic­ Wenselaers, S. (2008). De laatste Belgen. review. Psychological ­Bulletin, 140, Antwerpen, Belgium: De Bezige Bij. Asbrock and Van Hiel: Identification and Disidentification 131

How to cite this article: Asbrock, F. and Van Hiel, A. (2017). An Insiders’ Outside Perspective on the Flemish-Walloon Conflict: The Role of Identification and Disidentification for the German-Speaking Minority. Psychologica Belgica, 57 (3), pp. 115–131, DOI: https://doi.org/10.5334/pb.347

Submitted: 10 May 2016 Accepted: 23 January 2017 Published: 21 November 2017

Copyright: © 2017 The Author(s). This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International License (CC-BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. See http://creativecommons.org/licenses/by/4.0/.

Psychologica Belgica is a peer-reviewed open access journal OPEN ACCESS published by Ubiquity Press.