Political and Linguistic Identities in an Ethnic Conflict

The number of active ethnic conflicts has been steadily growing over the past half-century (Esteban, Mayoral, & Ray, 2012), prompting social scientists to redouble their efforts to understand the factors that influence them. A central problem in this quest is to distinguish between the role of natural group differences in language, religion, race, and customs from the role that political elites play in capitalizing on these differences (Green & Seher, 2003). In this article, we focus on ethnic conflicts where intergroup language differences appear to play a role, typically referred to as ethnolinguistic conflicts, and study the psychological role of language as a marker for social discrimination. It is plausible to expect that when groups polarize along an ethnolinguistic divide, the language that one speaks will influence how she is judged by the others. However, the mechanism through which social judgments polarize along such ethnolinguistic divides is not known yet. We empirically test two separate paths through which this might happen, one based on linguistic identity and one based on political identity.

Language is a powerful tool for social discrimination (Garrett, 2010; Giles & Billings, 2004; Giles & Watson, 2013). While social psychologists have noticed that gender, age, and race provide the basic categories through which we cut the social world (Fiske, 1998), more recently, researchers have added language to the list (Cohen, 2012; Kinzler, Dupoux, & Spelke, 2007; Kinzler, Shutts, DeJesus, & Spelke, 2009; Pietraszewski & Schwartz, 2014). We automatically notice if a person speaks with a particular accent, or if she uses a different dialect or language. In the context of ethnic conflict, as the proverbial biblical Shibboleth story tells us, language can easily become a key in-group/out-group marker, triggering hostility toward out-groups and/or solidarity with in-groups. Determining the precise role of language in intergroup conflict, however, has proven to be a challenging task. In political science, for example, the effect of ethnolinguistic divisions on intergroup violence has been a widely debated problem. Cross-national studies revealed rather complex patterns, where scientists using some measures of ethnolinguistic heterogeneity have found no link between language and intergroup violence (Collier & Hoeffler, 2004; Fearon & Laitin, 2003), while others, using different measures, found a reliable relationship (Montalvo & Reynal-Querol, 2005).

An important caveat of studies on ethnic violence is that they have largely ignored the psychological aspects of conflicts (Green & Seher, 2003). Even when a conflict appears to be well aligned with an ethnolinguistic division, hostility toward the out-group might be caused by different psychological processes. The most straightforward hypothesis is that during an ethnolinguistic conflict language becomes a major social divider, where those who speak the same language are considered an in-group and as a consequence are liked more, while those who speak a different language are considered an out-group and are liked less. Such a hypothesis agrees with various findings from social and cognitive psychology. Social psychologists, for example, have found that people easily form groups based even on arbitrary commonalities, such as color of clothing or shared aesthetic preferences and that group membership affects how we treat others, typically favoring the in-groups at the expense of the out-groups (Tajfel, 1970). Shared language can be seen as a strong indicator for group membership, resulting in the same type of in-group favoritism (Giles & Johnson, 1987). A plethora of empirical findings suggests that even in the absence of a direct intergroup conflict differences in language, dialect, or accent might in fact lead to social discrimination. For example, people speaking with a foreign accent tend to be discriminated against in housing, school, employment and in the courts (Gluszek & Dovidio, 2010). In children cartoons, villains are more likely to speak with a foreign accent independently of the particular historical or cultural reference (Lippi-Green, 1997). Recent work in developmental psychology have found that even 5-month-old infants prefer to look at those who speak the infant’s native language (Kinzler et al., 2007).

While greater preference for people who use one’s own language seems to be a good candidate for explaining polarization in ethnolinguistic conflicts, studies on language attitudes have shown that this is not a particularly robust pattern. Language attitudes are typically based on social evaluations of speakers’ qualities such as intelligence, trustworthiness, and friendliness, and these evaluations are then compared for speakers using different languages, dialects, or accents. In a seminal article on language attitudes (Lambert, Hodgson, Gardner, & Fillenbaum, 1960), Canadians were asked to evaluate English and French speakers. As predicted, Anglophone evaluators rated English speakers more positively on various desirable qualities than French speakers. Unexpectedly, however, Francophone evaluators showed the same pattern and also rated English speakers more positively. Subsequent work has shown that while shared language or accent can lead to more positive social evaluations (Lambert, Anisfeld, & Yeni-Komshian, 1965), there are multiple other factors involved, such as the socioeconomic status of the ethnolinguistic groups (Ryan & Sebastian, 1980), the particular evaluative dimension (Fiske, Cuddy, Glick, & Xu, 2002), gender (Bayard, Weatherall, Gallois, & Pittam, 2001), the surrounding linguistic environment (Dailey, Giles, & Jansma, 2005), observable behavior (Kinzler & DeJesus, 2013), or the degree of ethnocentrism of the evaluators (Bresnahan, Ohashi, Nebashi, Liu, & Shearman, 2002).

One factor that is particularly relevant for ethnic conflicts but received little empirical attention in studies on language attitudes is the politicization of language, where political views become closely linked to language ideology. Early proponents of nationalism have essentialized language as the key element of nation building, associating it with purity, authenticity, unity, and historical continuation of a cultural group (Fishman, 1972). A historical quote by Herder illustrates this point: “a so-called education in French must by necessity deform and misguide German minds. In my opinion this sentence is as clear as the sun at noon” (cited in Fishman, 1972, p. 53). Since the proliferation of the nation-states as a form of governance in the 19th and 20th centuries, in many countries language has become closely related to nationality (Anderson, 1991). In multiethnic societies, official recognition of a minority language often becomes a divisive issue, seen by some as a basic human right, but seen by others as an existential threat to the state. Nationalist politicians, accordingly, recognize the role that language plays in building a group identity and mobilizing supporters, and often make it a central symbol of the political debate (Fishman, 1972).

From the brief review above, it appears that we could expect at least two paths through which language might lead to social discrimination. In this article, we test these two alternative paths as two independent hypotheses. In what follows, we define the terms we are using in the rest of the article. We are interested in how evaluators judge the character of a target speaker, based on the language that the speaker uses. These social evaluations could be influenced by the linguistic identity of the evaluator, where we broadly define the linguistic identity as the language they usually use. According to the linguistic identity hypothesis, an evaluator will perceive more positively those target speakers who use the same language as the evaluator. In addition, we also consider the political identity of the evaluator, by which we mean the attitudes toward relevant political groups. During ethnolinguistic conflicts, the political struggle is typically intertwined with language ideology, where one pole of the political spectrum is associated with one language, and the other pole with another one. ¹ For example, the Kurdish leader Ocalan referred to the political status of Kurdish language as the main cause for the Kurdish rebellion in Turkey (Kinzer, 1999). In such a context, politically pro-Kurdish position also favors elevated status of Kurdish language. Therefore, according to the political identity hypothesis, evaluators who prefer a particular political group will also judge more favorably target speakers who use the language associated with this group. A graphical representation of the relationship between linguistic identity, political identity, and social evaluations of a target speaker is shown in Figure 1. Notice that political and linguistic identity do not necessarily coincide; instead they could overlap to various degrees. For example, 60% of Catalan speakers in Barcelona support a Catalan secession, and 17% are against it, while 62% of the Castilian speakers in Barcelona are against the secession, and 9% support it (Guidi & Karagiannis, 2014).

OPEN IN VIEWER

Figure 1.

Hypothetical role of language in social discrimination in the context of an ethnolinguistic conflict.

We test the two hypotheses in the context of the current conflict in Ukraine, where pro-government forces have been battling with pro-Russian separatists for more than a year. Recent official figures suggest that the clashes have left over 6,000 dead, 15,000 wounded, and 1 million people displaced (United Nations Office for the Coordination of Humanitarian Affairs, 2015). Is the Ukrainian conflict an appropriate context to test our hypotheses? According to Figure 1, two components are particularly important. First, the political poles in the conflict should be at least partially linked to language. It seems that the Ukrainian conflict meets this criteria. Voting patterns, for example, show that politics and language had been intertwined long before the current conflict started in 2014. As Figure 2 shows, regions speaking different languages tended to vote for opposite presidential candidates, and this trend has been particularly strong over the past decade. Furthermore, researchers have documented that the status of Russian language in Ukraine has been a divisive political topic since independence (Arel, 1995; Fournier, 2002). Subsequent governments have largely resisted the demand of Russian speakers for having a second official language. The presence of language tension is also evident from one of the first bills passed by the parliament in the aftermath of the protesters’ victory. The day after the president fled the country, the parliament voted for taking away the regional official status of all minority languages, including Russian. The legislature was vetoed by the new acting president, yet the pro-Russian leaders quickly capitalized on it and depicted the victorious protesters not as prodemocracy but as anti-Russian.

OPEN IN VIEWER

Figure 2.

Language and politics in Ukraine since independence: (A) Percentage of Ukrainian citizens choosing Russian as their mother tongue in 2001. (B) Support for the elected presidents in 1991-2014 elections and Pearson product–moment correlation between voting pattern and the percentage of Russian speakers (based on results from 27 districts). Positive r coefficients mean that regions with higher proportion of Russian speakers showed greater support for the elected president, and negative coefficients mean a weaker support. (Data from http://2001.ukrcensus.gov.ua/eng/ and www.electoralgeography.com.) The strong correlations reveal a consistent relationship between mother tongue and political divides, accompanied by clear geographic clustering along the Northwest–Southeast axis.

The relationship between the Russian and Ukrainian languages can be better understood if we examine the history of language policy in the region and how the status of these languages changed over time. In the 19th century, Ukrainian was affected by Russification policies of Alexander II. The beginning of the 20th century saw a more liberal approach to minority languages within the Russian empire, which was evident in the increase of schools that teach minority languages, including Ukrainian, and in the number of periodicals in local languages (Pavlenko, 2008). Communists, brought to power by the October revolution of 1917, at first were supportive of regional languages, but these policies were reversed by 1930. Despite the fact that the minority languages were not banned in the Soviet Union, there was an obvious difference between Russian and local languages in terms of their social values. Access to good education and jobs was conditioned on the knowledge of Russian (Bilaniuk, 2003). Russian was “the de facto official language of the country and a necessary prerequisite of a true Soviet citizen” (Pavlenko, 2008, p. 281). The breakdown of the Soviet Union in 1991 reversed this trend. Ukraine successfully implemented language policies that sought to replace Russian as the official language of the state and promoted the usage of Ukrainian in all public domains. Thus, according to Pavlenko (2008), in the early 2000s, 78% of students received secondary education in Ukrainian, and only 21% in Russian. Moreover, the political and social milieu in post-Soviet Ukraine elevated local Ukrainian-speaking elites to more prominent positions within the state. If we adopt May’s (2006) approach to minority and majority languages and formulate the difference between the two in terms of “clearly observable differences among groups in relation to power, status, and entitlement” (p. 255), we can consider Russian speakers to be a minority group in post-Soviet Ukraine (Pavlenko, 2008). This discussion shows that language policies in the region of what is now contemporary Ukraine have been driven by the ideal of the nation-state and the consistent effort on the part of political elites to establish one dominant language—Russian during the Soviet times and Ukrainian during the post-Soviet years—at the expense of the minority language(s). Such policies have been known to contribute to ethnic conflicts (cf. May, 2011).

The second critical component from Figure 1 that should be met is that the link between linguistic and political identities should not be too strong. Otherwise we will not be able to distinguish between the two effects. There is evidence that this criterion is satisfied, too. While language is heavily politicized, it is not the sole determinant of identity (Arel, 2002). For example, while 17% of the Ukrainian citizens declared themselves to be ethnic Russians, 30% chose Russian as their mother tongue. ² In addition, the vast majority of Ukrainians are at least passively bilingual, and it is common to overhear a conversation in two different languages or to hear Surzhyk, a mixed Ukrainian–Russian language (Bilaniuk, 2004). In summary, the Ukrainian conflict seems to meet both criteria, which enables the independent tests of our hypotheses.

To test the relative roles of political and linguistic identities in ethnic conflict, we conducted two empirical studies which we present next.

Study 1

Method

Participants

We recruited the participants online, using Google AdWords. Only those who indicated that they understand both Russian and Ukrainian languages were allowed to participate. After excluding incomplete responses, we were able to use the data from 130 participants. The mean age of the participants was 44.5, and 57% were males. ³ Data were collected between February and April 2015.

Stimuli

Stories

We chose three news stories on neutral topics and translated them into three languages—Russian, Ukrainian, and Surzhyk, the mixed language. While it has been known that in general Surzhyk is widespread but disliked (Bilaniuk, 2004), we included it to test if the politically neutral participants would be less disapproving of it. ⁴ The stories were approximately 100 words long.

Linguistic identity scale

Participants indicated what language they prefer to speak in everyday life on a 5-point Likert-type scale: (1) only Ukrainian, (2) mainly Ukrainian, (3) both Ukrainian and Russian languages, (4) mainly Russian, and (5) only Russian.

Political identity scale

The scale was based on four questions asking about the annexation of Crimea, the activities of the separatists in Donetsk and Luhansk, the activities of the Ukrainian army, the activities of the Russian army. Mean answers of 1 indicated extreme pro-Ukrainian attitude, and mean answers of 5 indicated extreme pro-Russian attitudes.

Social evaluations

The scale consisted of three questions about perceived intelligence, trustworthiness, and likability of the writer. High numbers indicated more positive attitudes toward the target person.

Design and Procedure

Each participant read all three stories, and each story was in different language. The order of the stories was counterbalanced across participants. After each story, we measured the social evaluations of the hypothetical journalist who had written the text. After the evaluation questions, we asked demographic questions and questions about linguistic and political identities. For more details of the measures, see the Supplementary Information (SI). ⁵

Results

The analysis of the political identity scale (Cronbach α = .92) revealed that overall our sample was strongly pro-Ukrainian, with 41% of the participants choosing the extreme pro-Ukrainian pole for all four questions (see SI Figure 1b). In addition to being positively skewed, the scale was also highly polarized, with 7% of the participants choosing the extreme pro-Russian pole for all four questions. For statistical purposes, we dichotomized the scale based on the mean (M = 1.98, SD = 1.40), dividing the subjects into politically pro-Russian (31%) and politically pro-Ukrainian (69%). While dichotomization of continuous or ordinal variables has been generally disapproved in modern statistics, the high polarization of the variable suggests the presence of discrete classes rather than of a continuum. Not surprisingly, political identity was significantly related to linguistic identity—Ukrainian speakers were more likely to be politically pro-Ukrainian than Russian speakers were. Nevertheless, the relationship was moderate in magnitude, r_pb(115) = .37, p < .01, see SI Figure 1c. Next, we developed a measure of language attitudes based on the difference between the social evaluation scores given to Russian target writers and the same scores for Ukrainian target writers. We collapsed across the three different contents and analyzed only the language manipulation.

Language attitudes were significantly correlated with the political identity of the evaluators, r_pb(114) = .22, p < .05. Compared with pro-Russian evaluators, pro-Ukrainian evaluators liked the target journalist more when the text was written in Ukrainian rather than in Russian. Linguistic identity was marginally correlated with language attitudes in the expected direction, r(120) = .14, p = .13, and the relationship fully disappeared when controlling for political identity (see Table 1). Compared with those who preferred to speak Ukrainian, those who preferred to speak Russian did not favor Russian target writers more. The observed pattern was consistent with the political identity hypothesis only, providing preliminary evidence that political preferences might lead to stronger language-based social discrimination than one’s own linguistic preferences.

OPEN IN VIEWER

Table 1.

Standardized Beta Coefficients From OLS Regression for Studies 1 and 2.

	Predictors	Study 1	Study 2
Model 1	Political identity	0.22*	0.24*
Model 2	Linguistic identity	0.14	0.03
Model 3	Political identity	0.20*	0.27**
	Linguistic identity	0.07	−0.07
Model 4	Political identity	0.19 †	0.26**
	Linguistic identity	0.07	−0.07
	Age	0.11	0.04
	Gender	0.06	0.03
	Education	0.08	−0.02

Note. The dependent variable is the difference between social evaluations of Russian and Ukrainian target writers/speakers. Political identity reliably predicts the difference, but linguistic identity does not.

†

p < .1. *p < .05. **p < .01.