Abstract
Abstract
As majoritarian electoral politics and religious conservatism are rising in the major multi-ethnic South Asian countries, such as India and Pakistan, the events of mob lynching, ethnic clashes and targeting non-plural and minority communities are becoming more frequent. This article analyses which cleavages of marginalisation make some ethnic groups prone to violent social movements vis-a-vis others. Theoretically, through social constructivism and horizontal inequality, the study argues that socioeconomic condition, religion and language are the three broad cleavages that influence political behaviour of ethnic groups. Explicating the theory about underlying versus facilitating conditions of ethnic–civil conflicts, this article examines the prerequisites of ethnic conflicts. Thereafter, it evaluates which single cleavages and combinations of the aforementioned cleavages increase the probability of conflict occurrence in the Indian subcontinent. The argument is empirically evaluated on a sample of 60 ethnic groups of the Indian subcontinents over the period of 1947–2013. We find that groups affected by reinforcing cleavages of religious and economic marginalisation, and religious, economic and lingual marginalisation have engaged in active violence over the period of our study. Additionally, the reinforcing cleavages of language and economy, and language and religion are associated with sporadic violence. Apart from the combined effects, we find that the ethnic groups facing economic disadvantage alone can also engage in violence.
Introduction
Terrorism and regional conflicts based on ethnic identities have become a source of a persistent problem in various parts of the world. Militant organisations functioning as units belonging to a common ethnic group threaten the lives of an array of individuals in a region, including not only state officials of the particular region but also other rival ethnic groups. More than one-third of terrorist organisations in the world operate to advance the interests of ethnic groups, and ethnic forms of terrorism are considered most prevalent in terms of a number of attacks and casualties (Masters, 2008). Apart from common ethnic interests, religious intolerance is also visible in various parts of the world. While the trend of majoritarian politics is rising in every continent, the communal riots witnessed globally are being triggered based on religious and socioeconomic interests among the various warring parties. The Indian Subcontinent, too, is not an exception.
Due to its vast size and heterogeneous society and polity, India has been the subject of various conflicts between subnational regions and the central government (Malone & Mukherjee, 2010). India has a high rate of political violence, and over 90 per cent of all events are riots and protests. The number of demonstrations is highest in northern India, including the Jammu and Kashmir region. The second most active conflict zone contributing to almost a third of political violence in India is the so-called Red Corridor, which spans over a vast territory in eastern, central and southern India. Political violence in that area is connected to the ongoing Naxalite–Maoist insurgency led by the Communist Party of India (Maoist). The third conflict zone—contributing to about a tenth of the overall political violence in India is in northeast India, where several ethnic groups including different tribes as well as migrants from other parts of India and neighbouring countries, such as Bangladesh, are engaged in a multi-layered conflict between separatist groups and the Indian government (Summary of Political Violence and Protests, 2018a).
Additionally, Pakistan has the second highest rate of political violence and protests in South and Southeast Asia and it accounts for over 20 per cent of all politically violent events in the region. Hundred and fourteen distinct non-state actors operate in Pakistan. From 2010 to mid-2017, rebel ethnic groups were involved in most events of political violence. The Tehreek-i-Taliban Pakistan (TTP) were involved in over 80 per cent of politically violent events, resulting in almost 17,000 fatalities. Baloch separatists were the second most active non-state actors in Pakistan, involved in 336 events. Baloch Separatists and Lashkar-e-Jabbar (LeJ) are active in Baluchistan, while the TTP, Afghani Taliban, Al Qaeda, Lashkar-e-Islam (LeI) and Jamaat-ul-Ahrar are mostly active in Federally Administered Tribal Areas (FATA) and Khyber Pakhtunkhwa. Additionally, a large majority of the activity in Sindh are riots and protests (Summary of Political Violence and Protests, 2018b).
On the other hand, Bangladesh and Sri Lanka too have been besieged with violence and protests over the last couple of decades. Series of political coups and the Chittagong Hill Tracts mobilisation perturbed the democratic stability of Bangladesh in the 1970s, 1980s, as well as 1990s. Various events in 2013, including the February 2013 International Crimes Tribunal (ICT) verdict against the Jamaat-e-Islami leader spurred widespread political instability across Bangladesh, particularly sparking violence by Islamist groups. The instability continued into January 2014, during which a general election was held, and widespread election-related violence occurred (Summary of Political Violence and Protests, 2018c). Sri Lanka too had been besieged with civil war since the 1980s until 2009 when the Tamil Eelam militants were finally defeated by the Sinhalese government. However, the key cleavages of marginalisation against the minority Tamils still loom large in this island state.
According to recent findings, 64 per cent of the civil wars have been fought along ethnic lines in recent decades globally (Denny & Walter, 2014; Themner & Wallensteen, 2012). Nevertheless, not all ethnic and religious groups are prone to violence and secessionist movements. Many of these ethnic and religious groups are integrated and cling to both their state and ethnic identity. Thus, the query remains, why do some ethnic groups adopt violence to fight against the state? This article examines this question and tries to explain which factors make some ethnic groups more prone to violence vis-a-vis others. While doing so, it elaborates the key idea of reinforcing versus cross-cutting cleavages of marginalisation that might lead to adopting violent practices among members of an ethnic group. Following ETH Zurich’s ethnic power relations dataset (Vogt et al., 2015) and EPR Ethnic Dimensions dataset (Bormann, Cederman, & Vogt, 2017), this article thereby tries to categorise the ethnic groups and the causal factors leading to the use of violence by various ethnic groups of the Indian subcontinent.
Although there is debate among scholars about the most important causal factors behind the onset of the civil conflict, the more nuanced task is to determine how these factors exist in combination leading to the onset and severity of the conflict. The ‘Minorities at Risk’ (MAR) survey shows that about 80 per cent of the politically active ethnic groups in the 1990s were disadvantaged because of historical or contemporary discrimination. Forty per cent of these groups (111 out of 275) surveyed face discriminatory policies and practices harmful to their material well-being (Harff & Gurr, 2004), but relatively few promote violent conflict. This study combines prerequisites of ethnic conflict into three broad cleavages—religion, culture and socioeconomic condition, thereby theoretically arguing that these three broad cleavages of distinction and marginalisation influence group behaviour. A combination of these stated cleavages develops a continuum of peaceful to violent politics. The present literature has established that individual cleavages are important by analysing the causality of one cleavage at a time. The contribution of this article, however, is multifaceted. First, the article studies the impact of multiple, simultaneous, cleavages rather than in isolation. Cleavages are studied individually, in combinations of two and finally all together; providing a unique contribution and addition to past work. Second, it provides greater precision on the magnitude of the effect of cleavages, both individually and in combination. Previous work has sometimes found connections between cleavages and conflict but has not necessarily identified their relative and absolute effects.
Conflict and Cleavages: Theoretical Literature and the Proposed Hypotheses
According to recent findings, 14 per cent of the ethnic minorities in the world have been involved in significant violence against the state (Fearon, 2008; Gurr, 1996; Seymour & Cunningham, 2016). When violent conflicts are organised by identity, cleavages must be sufficiently important for the masses to prepare them to fight, kill and even die. Personal motives are important, but without group motives and grievances, they may not be a principal driving force. Explicating the theory about necessary versus sufficient conditions of violent social movements, this article examines the prerequisites of ethnic conflict and evaluates which single cleavages and combinations of cleavages matter the most for conflict occurrence and severity.
Theoretically, the causality of ethnic conflict can be explained by primordialism, instrumentalism or constructivism. According to primordialists, such as Horowitz, group allegiances and comparisons are a fundamental aspect of social life (Horowitz, 1985). In fact, the primordialists argue that dichotomy between backward and advanced groups arose during the colonial period out of the differential distribution of resources, leading to select migration, education and employment for some but not for the others. Thus, the differential modernisation led to the dichotomous characterisation of ethnic groups (Horowitz, 1985). On the contrary, the instrumentalist approach disagrees with the primordialist view of ethnic identity formation as written earlier. Ethnic identity formation, in instrumentalism, is seen as a process created in the dynamics of elite competition within the boundaries determined by economic and political realities (Brass, 1991). However, in doing so, the instrumentalists tend to undermine the congenital identity of the people of a community and the process through which these ethnic boundaries are (re)shaped. For instance, individuals have many ethnic and non-ethnic identities with which they might identify politically. The challenge for politicians, according to instrumentalists, is simply to ensure that the identity that favours their party is the one that is most salient in the minds of a plurality of the voters (Wilkinson, 2006).
Social constructivists also argue that ethnicities are frequently used instrumentally for political purposes, but their emphasis is on ‘making’ and ‘remaking’ of ethnic boundaries that must occur to make such instrumentalism possible (Stewart, 2008). For example, both religion and linguistic symbols are potential bases for differentiating one ethnic group from the other (Brass, 1991), and constructivism successfully explicates the process of promoting multi-symbol congruence or incongruence using these innate cleavages of identities. Nevertheless, both instrumentalists and constructivists recognise that there need to be some differences in behaviour, customs, ideology or religion to make it possible to raise ethnic or other consciousness in an instrumental way (Stewart, 2008). These distinct group level cleavages causing conflict are the focal point of this study.
What Are Cleavages?
‘Cleavages’ include racial, political and religious divisions in society. Stewart (2008) categorised cleavages into four areas: political participation, economic aspects, social aspects and cultural aspects. She argues that these four broad categories are relevant to every society (Stewart, 2008). I incorporate the additional cleavage of religion in this article.
According to Cederman, Weidmann, and Gleditsch (2011) and Stewart (2008), group-level cleavages lead to collective identity formation. A cross-cutting cleavage exists when some characteristics of a group overlap or are in common with another group(s). For example, the Tuareg group in Mali shares a common religion (Islam) with dominant Bambara but speak Tameshaq language, which is different and is not recognised in the country. Another example is the Bengali Muslims of Bangladesh (erstwhile East Pakistan), who shared the common religion with the dominant majority of Pakistani Muslims but spoke a different language than the Urdu speaking Muslims. In contrast, reinforcing cleavage occurs when identities have no overlapping or commonality across groups. For instance, if two groups speak different languages, follow different religions, and belong to different ethnic groups—they have much less in common, and their cleavages of difference are reinforcing. For example, the Sudanese Arabs and South Sudanese are religiously, culturally and economically distinct from each other. The Sudanese Arabs (North Sudanese) are predominantly Muslim, economically better off than South Sudanese who are Christians and are culturally distinct too. The southerners were economically, socially and religiously marginalised for decades.
Role of Cleavages
Cleavages divide people into various groups having similar characteristics and common identity. These work in combination to influence group behaviour. Therefore, one group may often feel distinct from another group and be distrustful or wary of the behavioural pattern of another group. If they share nothing (reinforcing all cleavages of distinction), or few things in common (cross-cutting cleavages of distinction), the intergroup proximity, bilateral exchanges and trust shrink, giving rise to a void that is filled with mistrust and hatred, finally leading to communal or ethnic mobilisation.
Cleavages can also spur violence. According to Huntington and Lerner, the process of modernisation in developing societies causes rapid social and political mobilisation by breaking down the traditional order and expanding the communications and transportation networks (Taras & Ganguly, 2006). While this leads to a sharp increase in political participation in a multi-ethnic society, an equitable distribution of resource and socioeconomic benefits are hardly achieved due to lack of public service and public goods. Thus, some groups develop faster than others leading to a widening horizontal inequality between them. The Tamils and Sinhalese of Sri Lanka, and the dominant upper-caste Hindus and scheduled tribes of India are few of the many examples. Following the logic that group-level cleavages lead to collective identity formation, group-based discrimination and inequality leads to the common grievance and might spur group-based mobilisation and violence (Cederman et al., 2011; Stewart, 2008). For testing the likelihood of conflict, Stewart (2008) combined political, economic and cultural cleavage together, and analysed the effect of lack of cultural recognition and equity on group mobilisation. In this article, I have combined group-specific measures of religious, cultural and economic marginalisation in all possible continuation to study their effect on conflict likelihood. They have been studied individually, then combined in groups of two and finally all together. This is the unique contribution of this article in the field of conflict literature.
Inequality, Cleavages and Mobilisation
Cleavages make a society deeply divisible. Thereafter, inequality often runs along the lines of cleavages as not every group develops equally in such a deeply divided society. This leads to discrimination not merely among individuals, but among groups. In such cases, groups become different from each other divided by cleavages, and members of one group are better off than members of another group. Consequently, grievance among members of the discriminated groups builds up against the members of better-off groups, further leading to ethnic mobilisation and violence. To explain the impact of marginalisation on group mobilisation, Gurr wrote
The more strongly a person identifies with an ethnic group that is subject to marginalisation, the more likely he or she is to be motivated into action. Factors other than shared grievances, including a shared religion, language, history and culture, and place of residence strengthen group identity. The greater the number of traits common to a group, the stronger the group identity and stronger is the motivation to involve in action.(Harff & Gurr, 2004)
The causal argument mentioned earlier can be summarised and visually explained in Figure 1.

Harff and Gurr (2004) also framed a model for understating the facilitating conditions of ethnic conflict. The model explained the causality of conflict with a focus on contexts and facilitating conditions that involve societal norms of the disadvantaged groups, regime type, government response, group cohesion or factions, external support etc. Taking the argument forward, Fearon and Laitin (2003) argue that factors that explain which countries have been at risk for civil war are not their ethnic or religious characteristics but rather the conditions that favour insurgency. These include poverty—marking financially and bureaucratically weak states and also favouring rebel recruitment, political instability, rough terrain and large populations (Fearon & Laitin, 2003). The purpose of this research is, however, not to analyse the facilitating conditions or triggers of conflict as done by Fearon and Laitin (2003) but to focus on assessing the causality of the underlying conditions of ethnic violence. Contrary to Fearon and Laitin (2003), we evaluate the causal relationship of group-level marginalisation on the likelihood of conflict and identify the significant underlying necessary conditions of conflict—the cleavages of marginalisation.
Explaining the Cleavages of Marginalisation or Discrimination
Religion is a common source of division or marginalisation and had been a predominant divisive cleavage in the subcontinent. Temples, churches and mosques have not only been the houses of prayers. Traditionally, social exchange, cultural practices and governance have been linked to religion. Group grievances can also emerge out of inferior treatment to religious practices of groups. Designating a national religion or being biased in favour of a sect of a religion can generate resentment among others in a heterogeneous society. An example is predominantly Muslim Indonesia with episodes of violence between Hindus and Muslims and Muslims and Christians, where minority Hindus and Christians had been mistreated as well. Similarly, India is besieged by conflicts due to religious resentments between groups. After the mid-to-late 1970s, a rising curve of violence can be noticed which reached its peak in 1992 during the demolition of Babri Mosque and the subsequent Mumbai riots (Varshney, 2003). After a brief period of a downward trend, the communal clashes began erupting since the early 2000s. The 2002 Gujrat riots and Fedayeen attacks on Raghunath temple in Kashmir, 2010 Deganga riots in West Bengal, and the 2013 Muzaffarnagar riots and Canning riots are the instances of such events.
H
Economic inequality and associated active or passive repression have been the source of popular uprisings globally. The Maoist rebellion in the central Indian jungles is also primarily based on economic inequality between the sharecropper tribal and the upper caste economically better off populace. Since the end of colonial rule, the communists campaigned about false independence and argued that the transfer of power has only benefited the horizontally better-off section of India. One may rightly defend that such claims are too simplistic, but we have to agree that the mineral-rich Indian inland repeatedly witnesses sporadic instances of violence between the state and the Maoists who are supported by poor tribal and villagers.
Humphreys (2003) showed a striking negative relationship between the wealth of a nation and its chances of having a civil war (Humphreys, 2003). On the other hand, case study work also suggests that it is not inequality between individuals that matter for conflict, but rather inequality between ethnic groups or between regions—sometimes referred to as horizontal or categorical inequality (Tilly, 1999). The concept of horizontal inequality indicates that ethnic groups find themselves in radically different situations for various historical reasons. Whereas some ethnic groups came out on top of the geopolitical game, others were conquered early on and therefore lost out in the competition for wealth and influence (Horowitz, 1985). Thus, the distribution of power and wealth becomes a very crucial component of ethnic conflict. Woodward (1995) argues that alleged ethnic conflicts are really driven by underlying economic inequalities that lead ethnic identities to become politicised. If group-based income inequality combines with individual greed, grievance and other cleavages of conflict it becomes a more significant issue. Thus, I argue that the distribution of power and wealth is a very crucial component of ethnic conflict.
H
Another major cleavage of marginalisation is language as group grievances often emerge out of the inferior treatment of minorities’ cultural practices. Globally, language has been a symbol of identity and culture. The expression of distinct culture is observable in language or the mother tongue. The festivals, music and literature in which people grow up, listen to and read define who they are in communitarian societies. Nationalism grew in Europe in the period 1814–1848, and beyond, and language played a key role in the process. France and Great Britain were strong nation-states and as such stimulated other nations including Germany, Italy and Russia towards unification.
Alike Europe, Asia too has experienced political movements (often violent) under the lingual–cultural influence. Although promoting one single language as the national language can develop a cohesive national identity, but it can also generate resentment among minority lingual groups in a heterogeneous or multi-ethnic society. For instance, the dominance of Urdu language and culture had been a reason for resentment in East Pakistan leading to the evasive language movement and the war of independence in 1971. Similarly, in the 1950s, India faced major language movements by Telugu, Marathi and Punjabi speakers. Telugu speakers had a rich literary history and were associated with such symbols of Andhra glory as the Vijayanagar Empire (Guha, 2008). Post-independence, the Telugu community demanded the Congress to implement its old resolution of linguistic states. And, as methods, petitions, marches, representations and fasts were used (Guha, 2008). Potti Sriramulu began fast unto death on 19 October 1952 and died 58 days later into this fast. The news of his death triggered violent clashes in the Andhra province, where government offices were attacked, trains were defaced and police fired on protesters killing several. Two days later, Prime Minister Nehru declared that the state of Andhra will come into being for Telugu speakers. This incident spiked lingual demands across the country, and finally, a States Reorganisation Commission (SRC) was set up to make recommendations in regard to the broad principles which should govern the solution of the linguistic problem (Guha, 2008).
H3: Lingual–cultural marginalised groups are prone to conflict.
Are Reinforcing Cleavages More Dangerous Than Cross-cutting Ones?
The major story of this study revolves around the multidimensional cleavages, that is, cross-cutting versus reinforcing cleavages. It can be argued that the resentments due to marginalisation on one cleavage (e.g., only cultural marginalisation) are easier to address when two groups have many other things in common (e.g., religion, political power and socioeconomic status). When two groups differ and/or are disadvantaged from each other based on any single cleavage, it means that they share some other common identities too. Having some connections with the other groups can change the attitudes of the people. They might celebrate common festivals, or speak a common language, which reduces the alienation, anxiety and mistrust between them. The members of both groups in such cases have some shared identity and interests that lessens the feeling of discrimination, grievances and alienation (Guelke, 2012).
The scenario changes when the groups are fully alien to each other. When the cleavages of marginalisation overlap with each other, it can further heighten the conflict and be more divisive (Guelke, 2012). When an ethnic group is wholly alienated from another group, shares nothing in common and are disadvantaged in various aspects, it also means that some other group is enjoying reinforcing cycles of privilege (Stewart, 2008). This makes it harder for the disadvantaged group to accumulate assets in the future. The horizontal inequality in such cases persists because of persistent asymmetries in social capital and overt (or implicit) discrimination or favouritism by non-group members (Stewart, 2008).
These conditions primarily lead to a civil war. For example, in Burundi in the 1990s, half of the government investment went to the Bujumbura region and its vicinity, as it is the home of the elite Tutsi group (Gaffney, 2000). There are connections between economic and social elements too as lack of access to education leads to poor economic opportunities, while low incomes tend to result in poor educational access achievements in a vicious cycle of deprivation. The Chinese–Malay inequalities in Malaysia, the Tamils–Sinhalese inequalities of Sri Lanka are few examples of such cycles of deprivation. Hindi speaking animists and Christian scheduled tribes are both religiously and economically marginalised in India and have engaged in limited to severe violence since the 1950s. Our data covers 20 groups including Nagas, Kashmiri Muslims and Baluchis who are marginalised by all three cleavages of religion, economy and language and have engaged in sporadic violence (62.55%) during 1947–2013.
This presence of sharp reinforcing horizontal inequalities and cleavages can either provide an intense motive for political mobilisation or make the marginalised ethnic groups incapable and too weak to even organise an effective political movement. However, if the political mobilisation happens, it often becomes violent with the power of identities binding people together as group leaders find violence is the only way to secure political power. And, in this manner, the groups facing reinforcing cleavages become prone to mobilisation and violence.
Hence, it is important to study the various probable combinations of reinforcement of cleavages. We study the cleavages singly, in groups of two and finally altogether, which is a novel contribution to the literature. For instance, group A might be religiously and economically disadvantaged, while group B might have a religious and cultural disadvantage. The cleavage of religion is common for both; however, the overlapping of cleavages in both cases are different. It is important to study these kinds of combinations, as it helps to understand the reason for group-based violence and analyse which combinations of cleavages have a causal impact on conflict occurrence. This research, therefore, analyses the effect of all the possible combination of reinforcements of religious, economic and lingual–cultural marginalisation on conflict likelihood. This leads to four more hypotheses.
H H H H
Research Design
Spatiotemporal Domain
Ethnic groups have some unique features and a network of connections among them within a geographical region. Although many studies are global, it is a deliberate decision to restrict the geographical focus of this study in order to deal with the sub-national variations and ethnic divisions that are unique to a region. Subnational variation in South Asia matters both in theory and for its empirical properties. Focusing on this region allows us to control for a number of factors including geography, history and political system. Yet, we still have substantial variation in their empirical outcomes. Thus, India and her neighbours became a good test case for the theory. Another associated reason for selecting politically active ethnic groups of India, Pakistan and Bangladesh is because they were one single country before the independence and partition of India in 1947. Naturally, the ethnic groups are spread across the region and on both sides of the modern international border, and it shall be a challenge to justify our choice if all of them are not considered. For a similar reason, Sri Lanka is also included because of the presence of a significant Tamil population there—a notable ethnic group of south India.
Following ETH Zurich’s ethnic power relations dataset (Vogt et al., 2015) and EPR Ethnic Dimensions dataset (Bormann et al., 2017), the unit of analysis is group year. The data consists of 60 ethnic groups of India, Pakistan, Sri Lanka and Bangladesh. The total sample of observation (N) is 3,001 group years. The time frame for Bangladesh is 1972–2013 (as Bangladesh gained independence in 1971 December), and, for others, it is 1947–2013. A comprehensive study since 1947 is essential to include the language movements, the ethnic rebellions, and the class rebellions over time and space. As the data on the groups are available till 2013, hence the timeline of the study is until 2013. The panel logit model was run to study the presence/absence of conflict.
Dependent Variable
The article analyses the occurrence of conflict defined by any battle-related death. The ‘conflict occurrence’ variable in this research is a bivariate index coded from the ‘intensity’ indicator of the Uppsala Conflict Data Programme (UCDP) (Gleditsch, Wallensteen, Eriksson, Sollenberg, & Strand, 2002; Melander, Pettersson, & Themnér, 2016), 1
Also see PRIO’s webpage.
Independent Variables
The independent variables are economic, lingual–cultural and religious marginalisation, respectively. The indicators are coded from MAR data (Minorities at Risk Project, 2009). We created the religious and lingual marginalisation index as the MAR data does not have a comprehensive index for group-level religious and lingual marginalisation before 2004. The primary drivers of India’s mass politics have been religion and caste on the one hand, and prices and poverty on the other (Varshney, 2019). Religion, language and poverty have been the prime determinants of mass politics in Pakistan, Bangladesh and Sri Lanka as well. When politics in the subcontinent is itself a reflection of cross-cutting and reinforcing cleavages of religion, economy and language, we can argue that marginalisation against groups based on these cleavages is, in fact, a combination of political discrimination with their respective cleavages of distinction. For instance, the animist and Christian Bodos are religiously and lingually distinct from the dominant-plural Hindi speaking Hindus of India. Additionally, they have faced political discrimination ranging from passive neglection (1) to social exclusion (3) according to MAR data. Therefore, in this research, the measures of their religious and lingual marginalisation have been created as multiplicative indexes of respective distinctions and the levels of political discrimination.
The religious marginalisation variable is created as a multiplicative index of religious distinction and political discrimination indicators of MAR data. The ‘belief’ index (0–2) in MAR represents religious distinctiveness of ethnic groups, in which 0 determines same religion as the plurality group, 1 defines different sect within same religion as the plurality group and 2 indicates a different religion than the plurality or majority ethnic group (Minorities at Risk Project, 2009). In this analysis, we have created a bivariate variable: groups that score 1 and 2 in MAR are recoded as 1 in this study, and other groups, which are not religiously distinct, are coded as 0. The ‘poldis’ index ranges from 0 to 4 in MAR, and we used the same scale to assess political discrimination of ethnic groups in this study. Thereby, a multiplicative index of political discrimination and religious distinction has been created for each unit of analysis—representing religious marginalisation.
The lingual–cultural marginalisation variable is created as a multiplicative index of lingual distinction and political discrimination indicators of MAR data. The lingual distinction is derived from the ‘language’ index of MAR data. Language, in MAR, determines different lingual–cultural groups of a country, ranging from 0 to 2. The value 1 signifies non-plural groups speaking multiple languages, at least one different from plurality group (southern Sudanese in Sudan), and 2 defines non-plural groups speaking a different language from the plurality group language (Minorities at Risk Project, 2009). Groups scoring both 1 and 2 are considered lingual–cultural distinct, as in the theory section I argued that language is a determinant of culture as well. The variable ‘lingual–cultural’ distinction is bivariate too, in which one indicates lingual–cultural distinction. Groups that score 1 and 2 in ‘language’ index are coded as 1 in our data. The ‘poldis’ index ranges from 0 to 4 in MAR, and we used the same scale to assess political discrimination of ethnic groups in this study. Thereby, a multiplicative index of political discrimination and lingual distinction has been created for each unit of analysis—representing lingual–cultural marginalisation.
The third independent variable—‘economic disadvantage’ is derived from ‘ECDIS’ index of MAR. This index ranges from 0 to 4, in which 0 indicates no discrimination, 1 indicates neglect/remedial policies, 2 indicates neglect/no remedial policies, 3 indicates social exclusion/neutral policy and 4 indicates exclusion/repressive policy (Minorities at Risk Project, 2009). We used this scale to study group level economic discrimination.
Four interaction variables with the three independent variables (religious, lingual–cultural and economic marginalisation, respectively) have been generated to test hypotheses 4–7. These are the interactive terms derived from the combination of the individual variables mentioned earlier. 2
Minorities at risk (MAR) dataset does not have measures for linguacultural and religious marginalisation. As mentioned above in the research design section, this article has created measures for such marginalisation using multiplicative indexes of political discrimination and lingual and religious distinction, respectively. Thereby, the marginalisation indexes have been combined with each other to assess the effect of multiple cleavages of disadvantage on the group behaviour. This complicated and minute recoding of the existing data makes the article unique and establishes a point of departure from the present literature.
LIN–REL (interaction of lingual–cultural marginalisation and religious marginalisation)
LIN–ECO (interaction of lingual–cultural marginalisation and economic disadvantage)
REL–ECO (interaction of religious marginalisation and economic disadvantage)
LIN–REL–ECO (interaction of religious marginalisation, lingual cultural marginalisation and economic disadvantage)
Control Variables
The research focuses on the underlying conditions of conflict, hence, controls the facilitating conditions. Regime type and group population are two of the many facilitating conditions (Harff & Gurr, 2004) and are controlled in the statistical analysis. Polity score determines the regime type and the size index of ETH ethnic power relations dataset determines population size.
It is often seen that larger groups are better able to mobilise, protest and fight due to their numerical strength. Thus, the population ratio of various groups is an important factor of reference in ethnic and civil conflict. Hence, a size (Vogt et al., 2015) variable of ETH is used in this article. If a marginalised group is 10 per cent of the total population, the indicator codes it as 0.1, if a marginalised group is only 2 per cent of the population, then it is coded as 0.02 and so on.
According to Selectorate theory, the size of the winning coalition determines whether policies have a private or public focus (Bueno de Mesquita, 2005). When the size is small, as in autocracies and monarchies, the small number of supporters receive private benefits (Bueno de Mesquita, 2005) keeping the larger mass devoid of such benefits and spurring grievances in them. However, when the coalition is larger, as in democracies, the insufficiency of private goods forces the leaders to switch to policies with a public focus (Bueno de Mesquita, 2005), thus, keeping the mass contented. Thus, regime type becomes an important facilitating condition. We use ‘Polity score’ (Marshall & Gurr, n.d.) to measure the regime type, which is a 20-point scale determining the level of democracy or autocracy in a country. 3
Polity score is rescaled from −10 to +10 to 0 to 20.
Results and Analysis
Descriptive Statistics for Categories of Marginalisation (1947–2013)
**Recoding the intensity indicator of UCDP battle deaths data, level 2 signifies 26–999 deaths (per year) in our study.
***Recoding the intensity indicator of UCDP battle deaths data, level 3 signifies 1,000+ deaths (per year) in our study.
Conflict Occurrence—(0–1) (1947–2013)
4
Effects of country dummy variables were tested in the models but have dropped them in the final table, as they did not show any significant effect. Their inclusion did not affect the results on the key variables.
Effects of country dummy variables were tested in the models but have dropped them in the final table, as they did not show any significant effect. Their inclusion did not affect the results on the key variables.
Individual Effects of the Cleavages
Among the sample of groups analysed in this research, only Shia Muhajirs are affected by religious marginalisation alone. Table 1 indicates that religious marginalisation as a cleavage alone is associated with 78.12 per cent violent conflicts across the timeline of the research. From Table 2 we infer that religious marginalisation has a significant positive effect on the probability of occurrence of conflict, and the effect was consistent across different model specifications. This means as a single cleavage, religious marginalisation can instigate violence or in other words, groups affected by religious marginalisation alone are likely to engage in sporadic instances of violent conflicts. For example, the spewing violence in Karachi by the Muhajirs, who demand their own province is an uprising by a non-plural religious group against the Islamic Republic of Pakistan and has claimed many lives. The Human Rights Commission of Pakistan mentioned that 490 people were killed in Karachi in ethnic and political killings in the first 6 months of 2011.
Although the language is a critical part of a collective identity, none, of all the groups studied, had only linguacultural marginalisation; linguacultural marginalisation was always found with one or the other cleavages. Thus, any effects were only in combination with another cleavage type. Nevertheless, the Gorkhaland movement in India has clamoured against the linguacultural alienation by the culturally and ethnically different Bengalis, because of which the Gorkhas have been demanding for separation from West Bengal since 1903. Similarly, the Assamese instability also represents an apparent clash of two exclusive lingual groups—the Assamese and the Bengalis. However, a closer look at these cases will establish that language is a contributory factor for conflict occurrence in association with others and may not be causal. In fact, there is a considerable economic difference between the Gorkhas and the Bengalis of West Bengal; hence, we cannot deny the causal effect of economic marginalisation as a cleavage in this case.
On the other hand, the Assam movement of 1979–1985 was a popular uprising against all illegal immigrants of Assam. The movement, led by All Assam Students Union (AASU) and the All Assam Gana Sangram Parishad (AAGSP), developed a programme of protests and demonstrations to compel the Indian government to identify and expel illegal immigrants from Assam who fled Bangladesh during the liberation war of 1971. Recently, the citizenship bill of 2016 also declared the Bengali-speaking Muslims without proper residency permits as illegal immigrants, whom the federal government seeks to deport back to Bangladesh. The agitation is not only against the Bengalis (who can be Hindus as well) but against the Bengali Muslims. In terms of this research, the marginalisation in Assam is thus twofold—both religious and linguacultural, instead of being only lingual. In other cases, communities with lingual differences often live happily unless other variables of alienation are present.
While the article argues that linguacultural marginalisation is probably not a causal cleavage for conflict occurrence, it finds that economic disadvantage alone is associated with 83.01 per cent violent conflicts across the spatiotemporal domain of the research (Table 1). Moreover, the economic disadvantage has a significant positive effect on the probability of occurrence of conflict (Table 2). These were consistent across different model specifications. In other words, the result indicates that groups that are economically marginalised, despite sharing the common religion and language with the dominant group(s), are prone to conflict.
Our findings associate well with the Marxist analysis of group conflict or the social conflict theory. In the classic example of historical materialism, Karl Marx and Friedrich Engels argued that the human history is the result of conflict between classes, which evolved over time in accordance with changes in society’s means of meeting its material needs. Therefore, the first chapter of the Communist Manifesto starts with ‘the history of all hitherto existing society is the history of class struggles’ (Marx & Engels, 1848). According to the social conflict theory, contradictions in interests and conflict over scarce resources between groups is the foundation of a social society. Nevertheless, the equitable distribution of resource and socioeconomic benefits is hardly achieved in any society due to a lack of public service and public goods. Thus, some groups develop faster than others leading to a widening horizontal inequality between them. This research empirically validates the social conflict theory and finds that when horizontal inequality is economic in nature, it significantly increases the disadvantaged groups’ proneness towards violence.
Among the groups analysed in this research, nine have been subjected to economic marginalisation alone. For example, the Pashtuns (both Shias and Sunnis) of Pakistan have been economically marginalised for decades despite facing no religious or lingual marginalisation, and they have been involved in sporadic instances of violence with a conflict intensity of 1 (less than 25 deaths per year) and 2 (25–999 casualties per year). The Pashtun ethnic clashes have claimed many lives over the decades and the Khyber Pakhtunkhwa province of Pakistan remains violence-laden even till date. The Maoist rebellion in the central Indian jungles is also primarily based on economic inequality between the sharecropper tribal and the upper caste economically better off populace.
The Maoist rebellion in India began as an armed peasant revolt in 1967 at the Naxalbari block of Siliguri subdivision in Darjeeling district, West Bengal. The movement was organised in the lines of the protracted people’s war (a Marxist military strategy developed by Mao Zedong) to maintain the support of the population and draw the state forces deep into the countryside where the militia aimed to bleed them through a mix of mobile warfare and guerrilla warfare. Although the Naxalite movement in India apparently seems to be an ideologically motivated radical communist uprising by some members of Communist Party of India (Marxists), a closer look at the mobilised participants explicates the ethnic component of it and the effect of economic marginalisation. For organising the sharecroppers and tribes in 1967, the communists created peasants’ cells throughout the Siliguri subdivision in Darjeeling District. In March 1967, the sharecroppers mobilised under the peasants’ cells started seizing lands from landlords (jotedars), who owned relatively extensive tracts of land. In contrast to the property owners or jotedars, who were upper-caste members, the sharecroppers or bargadars belonged to lower castes and tribes and were landless tillers (Iqbal, 2010). Although the Bengal Tenancy Act of 1885 was abolished after the independence of India, the erstwhile landlords continued to enjoy the socioeconomic privileges that sustained the existing horizontal inequality between the groups. The Naxalbari uprising was suppressed in July 1967 when paramilitary was deployed by the government. The Communist Party of India (Maoists) (CPI (M)) expelled many of its members supporting the uprising who organised themselves into the Communist Party of India (Marxist–Leninist). CPI (ML) remained at the centre of the Naxalite movement till 1975, gathering enthusiastic urban youth support as well as expanding to various scheduled tribes of the east, southeast and central Indian jungles.
The conflict in its present form began in 2004 after the formation of the CPI (M). The ongoing conflict has taken place over a vast territory (around half of India’s 29 states) with hundreds of people being killed annually in clashes between the CPI-Maoists and the government every year since 2005 (Gleditsch et al., 2002). 5
Also see PRIO’s webpage.
The key contributions of this article, however, are not only the individual effects of the single cleavages but also their interaction effects. Often the cleavages intersect with each other, and this article explains the interaction effects in the next section.
Interactive Effects of the Cleavages
Among the groups studied in this research, the combination of religious and economic marginalisation affects four groups while the cleavage of language cuts them across. These groups are Bengali Hindus of Bangladesh, Christians of Pakistan, Punjabi speaking Hindus of Pakistan, and Hindi speaking Christian and Animist scheduled tribes of India. Table 1 reports that the groups affected by the combination of religious and economic marginalisation are associated with 74.4 per cent violent conflicts across the period of our research. When these two cleavages intersect, it has a significant positive effect on the occurrence of conflict as well (Table 2). In other words, the groups that are both religiously and economically marginalised, despite speaking the same language as the dominant ethnic group are prone to conflict.
The Hindi speaking non-Hindu scheduled tribes of India (Animists and Christians) have sporadically revolted against the union government in central Indian jungles under the banner of Naxalite movements, despite sharing a common language with the plural Hindi speaking population. These tribes have been religiously and economically discriminated since the 1950s and have engaged in limited sporadic violence during 1950s and early 1960s. The conflict intensity increased during the Maoist rebellion in the late 1960s and since 2005. The Maoist movement is often seen as an ideological and class struggle, but their targeted groups of mobilisations have always been various scheduled tribes of India who face various cleavages of marginalisation. This makes us argue that the ethnic dimension is an important conjecture in the Maoist movements. The findings indicate that the combination of religious and economic marginalisation has the potentiality to cause protests and violence, even if the lingual cleavage cuts across the groups. The statistical findings of this research also show that as the intensity of religious marginalisation moves up from level 1 (neglect) to level 4 (repression), the chances of economic disadvantage positively affecting the probability of conflict gradually go up from 5.5 per cent to 10.5 per cent.
Table 1 reports that the combination of lingual and economic marginalisation affects eight ethnic groups, and the groups affected by this cleavage of marginalisation are associated with 67.07 per cent of violent conflicts across the period of the research. Table 1 also reports that the combination of lingual and religious marginalisation affects two ethnic groups, and the groups affected by this cleavage of marginalisation are associated with 60.52 per cent of violent conflicts across the period of the research. However, the empirical results (model 2 of Table 2) find no statistical significance of the combined effect of lingual and religious marginalisation on the occurrence of conflict when economic commonalities exist across the groups. Similarly, according to model 3 of Table 2, we find no significant evidence of the combined effect of lingual and economic marginalisation on conflict when they share a common religion. The statistical insignificance can be attributed to factors, such as the absence of a greater number of groups experiencing marginalisation. According to our data, only eight groups are affected by the combination of lingual and economic marginalisation, and only two groups have experienced the combination of lingual and religious marginalisation. Additionally, the intensity of conflict caused by them is also low. These further explain the reasons for statistical incongruence or insignificance.
Furthermore, we finally find that the likelihood of conflict is probable when all the cleavages of marginalisation reinforce. Among the groups analysed in this research, the reinforcing marginalisation of economy, religion and language affects 20 ethnic groups. They are Ahmadis, Shia Balochis, Shia Sindhis, parkari and sindhi speaking Hindus in Pakistan, Christian and animist Bodos, indigenous Tripuris, Kashmiri Muslims, Mizos, Nagas, Shia and Sunni Muslims, and non-Hindi speaking animist scheduled tribes in India, Tamils and Moors in Sri Lanka and Tribal-Buddhists in Bangladesh. These groups have been subjected to three ways marginalisation often and have been associated with 62.55 per cent of violent conflicts across the period of our research (Table 1). The empirical results in Table 2 also suggest that the probability of conflict occurrence is high and significant when a group is totally distinct and marginalised (economically, culturally and religiously) from the dominant group.
India and Sri Lanka have both been fighting a protracted war against Kashmiri militants and Tamil militants, respectively. Not astonishingly, the contexts of both the situations are quite similar—multifaceted marginalisation. During the 1987 state elections in Indian administered Kashmir, various Islamic anti-establishment groups including Jamaat-e-Islami Kashmir were organised under a single banner named Muslim United Front (MUF), that is largely the current Hurriyat. Their slogan indicated the demand to introduce Sharia law of Quran in the state assembly (Schofield, 2010), which is against the secular ethos of the Indian constitution. The MUF lost the election, and it is believed that the elections were rigged in favour of Farook Abdullah, who had an alliance with the Congress party.
The rigged election results led to the rise of an armed insurgency movement and Pakistan supplied these groups with logistical support, arms, recruits and training (Jamal, 2009). As the protests and rallies intensified in the valley, largely organised by Jammu Kashmir Liberation Front (JKLF), the government of India deployed Border Security Force (BSF) and paramilitary under Armed Force Special Power Act (AFSPA) in September 1990 to suppress the insurgency by giving armed forces the powers to kill and arrest without warrant to maintain public order (Behera, 2006). Religiously and culturally distinct from the majority Hindus, now politically disadvantaged due to interventions of the federal government and AFSPA, and economically backwards due to lack of opportunities; the Kashmiri Muslims started feeling marginalised and alienated from the rest of the country. As the grievance grew, JKLF used Islamic themes to mobilise crowds and justify their use of violence. The available battle-related deaths and events data indicate that Kashmiri Muslims have engaged in active violence over the period of this study, and the intensity increased since the late 1980s.
On the other hand, the Tamil–Sinhalese friction in the island state dates to the colonial period when British established a centralised government in 1833. Right after independence, conflict over lingual rights took the form of the Swabhasha movement (the people’s own language movement), which pushed for the Sinhalese language to become the official language of Sri Lanka instead of English (Taras & Ganguly, 2006). Thus, in terms of this research, we find that linguacultural marginalisation has been a contributory factor in Tamil militancy. Aside, Ceylon Citizenship Act of 1948, Parliamentary Elections (Amendment) Act of 1949 and Official Language Act of 1956 were passed that made a vast majority Tamils stateless, reduced the parliamentary capacity to defend the minority civil rights and granted no concession to the Tamils to use their language for education and administration, respectively (Taras & Ganguly, 2006). These factors successively made the Tamils both economically and religiously marginalised. Faced with Tamil protests, although the Tamil Language Act was passed in 1958, its implementation was minimal. Hence, by the end of the 1950s, the Tamils were multidimensionally marginalised.
The Tamils protested the lingual and religious supremacy of the Sinhalese, and the economic disadvantages that they were subjected to. In 1978, the Jayewardene government introduced a new constitution that provided some concessions to the Tamils including the status of a national language and a new system of voting to count the minority votes in national politics (Taras & Ganguly, 2006). However, again these concessions were barely implemented transforming the existing Tamil grievance to organised militancy. Several Tamil insurgent organisations cropped up and the Liberation Tigers of Tamil Eelam (LTTE) led by Velupillai Prabhakaran was the largest of them. The Sinhalese government stopped the basic necessities of food and water in the Jaffna region and carried out genocide against Tamils during the anti-Tamil riots of 1983. The situation soon turned into a civil war. In the Sri Lankan Civil war that lasted for 27 years, over 40,000 civilians, 30,000 Tamil militants, 21,000 Sri Lankan security forces and 1,200 Indian soldiers lost their lives.
The Tamils and Kashmiri Muslims are both marginalised multidimensionally. While the former has fought a three-decade-long civil war, the latter is still engaged in active militancy against the state. Hence, we can conclude that reinforcing cleavages of marginalisation are likely to increase the probability of conflicts. The death of Velupillai Prabhakaran, the founder and leader of the LTTE, in 2009, left the Tamils without an able leadership to reorganise an effective political movement. However, the Sri Lankan Tamils, still today, are disadvantaged multidimensionally and face serious hardships in daily life.
Conclusion
The article aimed to analyse the underlying conditions of conflict—the causal effect of religious, economic and lingual marginalisation. One of our argument was that studying these cleavages alone is not enough as many groups are often affected by more than one disadvantages at a time. Additionally, the existence of various reinforcing and cross-cutting cleavages within society even make it harder to manage these conflicts. After a thorough analysis of the politically active ethnic groups of the Indian subcontinent, this research finds that the reinforcing cleavages significantly increase the probability of conflict in this region. The combination of religious and economic marginalisation and the overlapping cleavage of religious, economic and lingual disadvantage increases the probability of conflict, and, although statistically insignificant, the lingual and religious marginalisation and lingual and economic marginalisation have instigated more than 60 per cent of violent conflicts across their respective spatiotemporal domain. Although reinforcing cleavages do lead to the occurrence of conflict, we found that cross-cutting cleavages also increases the probability of conflict occurrence. If any group experiences only economic marginalisation, despite having religious and lingual similarity, it may be conflict prone. This strengthens the argument that economic disadvantage is an ordering principle in certain sections of the world. The interactive effect of religious and economic marginalisation also has a significant impact on conflict occurrence, thus proving that these two cleavages of marginalisation are the ordering principles in the Indian subcontinent. In brief, our empirical findings support the concept of horizontal inequality, which indicates that grievances develop due to group-level marginalisation and unbalanced economic development.
This article analyses the necessary underlying conditions of conflict besieging the Indian subcontinent. However, the presence or absence of facilitating conditions has an effect on behavioural patterns of the ethnic groups, which future research should address. Second, future research may also address the disputes about common resources between two equally marginalised ethnic groups. Discrimination and cleavages are the basic necessary conditions for the occurrence of the conflict. However, some other associated factors, such as societal norms, regime type, government response and the reaction of majority groups, set the context and trigger the conflict, and in turn, often intensify it. My future research would explore those areas for a more cohesive understanding of the ethnic–civil conflict.
South Asia is a hotspot for economic development and global security in the twenty-first century. In recognition of the increasing connectivity of the Indian and Pacific Oceans, the US Pacific Command has also been renamed to US Indo Pacific Command. If not an uninterrupted peace, at least a manageable governance is essential for such a region to thrive politically and economically. However, with every country besieged with the domestic and bilateral political charade, such tranquillity is a utopia. The 14 February 2019 Pulwama attack in Indian Kashmir by Jaish-e-Mohammed brought India and Pakistan, two nuclear powers and sworn enemies on the brink of an all-out war. This incident once again pointed out that the realist approach of perceiving issues of ethnic conflict through a prism of national security is unreal.
In his seminal work on nationalism, ‘Imagined Communities’, Benedict Anderson argued that the nation is ‘an imagined political community’ (Anderson, 2006). In this sense, the nation is a cultural construction because most members of a nation would probably never meet their fellow members but will retain the strongest emotion for their shared national membership (Taras & Ganguly, 2006). In the subcontinent, the idea of ‘state’ has evolved much after the idea of group-specific nationalisms. The feeling of Telugu, Tamil, Punjabi, Bengali, Bodo, Mizo, Balochi or Kashmiri nationalism dates to centuries while the ‘states’ of India, Sri Lanka and Pakistan are barely seven decades old, with Bangladesh being younger. Hence, it is essential to understand the inherent behavioural pattern of these ethnic groups through a constructivist prism in order to manage the conflicts and to attain long-term stability.
This research has a group-specific focus and instead of a realist approach, undertakes a constructivist approach of analysing group level marginalisation. Thus, it adds value and makes it explicit that reinforcing cleavages of disadvantage increase conflict. Additionally, the article suggests that socioeconomic equitability is necessary for long-term stability and states should prevent ethnic groups from becoming religiously and economically alienated.
Footnotes
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The author received no financial support for the research, authorship and/or publication of this article.
