Linguistic Assimilation and Ethno-religious Conflict

Denial of recognition to the Kurdish language in Turkey is linked to the Turkish nationalist policy of cultural assimilation. In Latvia, despite about 40 % of the population being Russian-speaking, Latvian remains both the sole state language and a requirement for citizenship. In the UK, English language requirements for citizenship tests have been progressively tightened in recent years. Ortega and Tangerås (2008) develop a political-economic analysis of the imposition of mono-lingual education by dominant groups.

The first extends a free trade argument into social policy (e.g., Lazear 1999). It was advanced by colonial administrators and social reformers in colonized countries in the nineteenth and early twentieth centuries, as the justification for westernizing the education system, the legal code, and social behaviour. Civil rights laws and anti-discrimination statutes in the US are motivated at least partly by the belief that social integration promotes economic efficiency (Fredrickson 1999). Contemporary examples of populist articulation of the second and third arguments include political parties such as the French National Front, the Dutch Party for Freedom, the Bharatiya Janata Party of India, Jobbik of Hungary and Golden Dawn of Greece.

Lazear (1999), Kónya (2005), Kuran and Sandholm (2008), Li (2013) and Bowles et al. (2014) develop models of assimilation (or, more generally, social segregation and integration), but do not analyze the implications for ethnic conflict. Akerlof and Kranton (2000) explain forms of dysfunctional individual behaviour in terms of stresses generated by identity norms, but do not model their aggregate consequences for conflict between communities. Conversely, Dasgupta and Kanbur (2005b, 2007) and Esteban and Ray (2008, 2011) examine how exogenous changes in the income distribution affect conflict between communities, and thus do not connect the income distribution to the extent of cultural-linguistic integration prevailing in the society. The connection between cultural integration and social conflict thus remains unexplored in their analysis. Dasgupta (2009) shows how class conflict between workers and employers, and ethnic conflict between different groups of workers, mutually condition one another, but assumes homogeneity within the working class in all employment-relevant aspects except the reservation wage. Bakshi and Dasgupta (2017) explore the dynamic evolution of ethnic conflict when state institutions are susceptible to ethnic capture. Our analysis has a distant family resemblance with that of Bhattacharya et al. (2015), who examine the role of inter-group mobility in the emergence of conflict, but varies greatly from theirs both in its specific structure and modelling of conflict.

In reality, some practices with religious, racial or ethnic identity connotations may also have productivity implications. The broad-brush distinction is porous but nonetheless empirically helpful, and is routinely deployed in the economic analysis of discrimination.

In 2010, France banned the wearing of a face-covering veil in public. The key official justification was productivity-relevant: face-coverings prevent identification, which is both a security risk and a coordination hindrance, in a society which relies on facial recognition and expression in communication. Thus, headscarves were not affected. In contrast, the wearing of all conspicuous religious symbols in public schools was banned in France in 2004 by a different law, which did affect the wearing of both Islamic veils and headscarves. The Turkish government has traditionally banned women who wear headscarves from working in the public sector. In both cases, the ban on headscarves was justified not by any direct negative impact it might have on productivity, but by its symbolic role in keeping the public sphere secular. In a referendum held in Switzerland in 2009, a constitutional amendment banning new mosque minarets was approved by 57.5 %. In Northern Ireland, clashes often break out over rival Catholic and Protestant marches organized annually to commemorate events in the history of past antagonisms, in India Hindus and Muslims contest ownership of medieval structures, while in Europe conflicts rage between rival mobilizations over demands for censorship on grounds of blasphemy. In all these cases, the items of contestation do not appear to have any direct or immanent implications for workplace coordination or economic productivity, but are intrinsically valued by (typically ethno-religious) communities as constitutive symbols of self-expression in the public sphere.

Generalized discrimination against the minority can be modelled as a constant cost component, \( d\le {\rho}_N \), that impacts all assimilating N individuals equally. Thus, an increase in such discrimination simply reduces the returns from assimilation by an identical amount for all minority individuals. Individuals may perceive their own expressive and behavioural habits as norms rather than conventions, in that they may intrinsically value them as ideals to live by. In that case, identity-switching will involve a psychic cost. If such marginal psychic cost increases with the level of workplace effort, individuals may rationally provide less than full effort in an alien work environment. The effort level provided will then vary according to idiosyncratic differences in the marginal psychic cost function. Though evidently compatible with my analysis, I refrain from explicitly modelling this additional source of idiosyncratic differences in productivity on considerations of expositional ease and simplicity.

There exist multi-cultural equilibria involving partial assimilation as well, but these are all unstable.

For a detailed discussion of such movements and their role in the construction of ethno-linguistic nationalism in modern Europe, see Hobsbawm and Ranger (1983) and Hobsbawm (1992).

Furthermore, such an equilibrium necessarily generates lower total output than the one where the minority assimilates, provided M, on average, finds it at least as costly to change its behavioural patterns as the minority (i.e., \( E\left({c}_M\right)\ge E\left({c}_N\right) \)], and may do so even otherwise. Since this is empirically likely, it is therefore of limited normative interest.

An off-shoot of assimilation, at least over time, may conceivably be the weakening of community cohesion within the minority, and therefore of conflicts with the majority. Kuran and Sandholm (2008) offer an evolutionary game theoretic perspective on this view. My static argument is independent of this dynamic argument. Note also that a shift to assimilation may increase inter-group conflicts when N is sufficiently less cohesive, relative to M. Then, income inequality engendered within N by assimilation may actually increase total political spending by N, expressed as a proportion of total societal income.

This follows immediately from condition (22) in the Appendix.

Montalvo and Reynal-Querol (2005, 2008) find that societies which are ethnically more polarized, i.e., where majority and minority communities are close in size, may be more prone to social conflicts (specifically civil wars and genocides). Easterly et al. (2006) present a similar finding in the context of mass killings. The analysis here can be seen as providing a theoretical rationalization of these empirical findings. See also footnote 17.

Conflicts within the African-American community over ‘acting White’ constitute a specific example, of which Austen-Smith and Fryer (2005) provide a formalization.

There is some weak cross-country evidence linking greater external openness with lower internal conflict, and it is well-known that globalization affects domestic conflict in contradictory ways through channels such as income distribution, international prices for contestable mineral resources, revenue base of the government etc. [see Barbieri and Reuveny (2005) and Magee and Massoud (2011) for recent discussions]. I thus add to this literature by highlighting an additional mechanism. Collapse of the Soviet Union and the consequent economic disruption arguably played an important role in the revival of ethno-linguistic tensions in parts of Eastern and Central Europe, as well as in many former Soviet republics. Integration into some third linguistic-cultural tradition shifts the normative issue of a just distribution of gains to a global level (see Van Parijs (2011).

In recent decades, opening up of job opportunities in Saudi Arabia has led to important income gains for some sections of Indian Muslims, but has also incentivized greater adoption of Saudi Wahhabism-inspired behavioural and religious norms and expanded the influence of Arabic in expressive practices. Remittances have funded ethno-religious assertion (e.g., the building and refurbishment of denominational mosques and religious schools, lavish spending on ceremonies, withdrawal of women from the labour market, campaigns for strict observance of dress and dietary codes, etc.), and on mobilizations to organize, defend or enforce such assertion. This in turn has generated conflict and counter-mobilization. The Saudi influence is noticeable in conflicts over organized attempts to impose Wahhabism-inspired linguistic, behavioural and religious norms in Bangladesh and Pakistan as well. See, for example, Boone (2014) for a discussion in the context of Pakistan.

Note that such societies are ethno-linguistically less polarized, and recall footnote 13. Desmet et al. (2009) and Desmet et al. (2012) develop an empirical operationalization of the idea of linguistic distance between communities. Consistent with my analysis, the latter contribution finds deep linguistic cleavages (which imply large assimilation costs) to be empirically better predictors of civil conflict.

On affirmative action, see, for example, Holzer and Neumark (2000). Transfers conditional on assimilation efforts (e.g., participation in language classes) may also generate socially excessive adjustment by minorities: Bougheas et al. (2007) show how conditional anti-poverty transfers may be inefficient, yet persist indefinitely.