Public Opinion on Peace as a Reflection of Social Differentiation and Politicisation of Identity in Sri Lanka

This article provides a critical analysis of the public opinion on peace in Sri Lanka, with consideration to two determinants: social differentiation and politicisation of identities. Specifically, it aims at developing arguments about the correlations between public opinion, social position, and political mobilisation. Inspired by Bourdieu’s concepts of habitus, social space, and political field, this article develops an empirical analysis of the links between ethnic identity and public opinion on peace, and between social differentiation and opinions within the Sinhalese majority community in Sri Lanka. This article argues that ethnic polarisation and politicisation were the foremost determinants of public opinion during the peace process in 2002-2009. Introduction How did the public opinion on peace in Sri Lanka vary between different social groups and with changing political dynamics during the peace process of 2002-2009? And what explains these variations and changes? Do they reflect class structures in society or are they the outcome of politicisation of ethnic identities? The purpose of this article is to examine the public opinion on peace with regard to two general determinants: social differentiation and politicisation of identities. Empirically, we will use survey data to develop an argument about the links between public opinion, social position, and political mobilisation. Public Opinion on Peace as a Reflection of Social Differentiation and Politicisation of Identity in Sri Lanka 38 The article is organised in three sections. The first section outlines the conceptual framework for the analysis, drawing especially on Bourdieu’s notions of habitus, social space, and political field. Following from this theoretical discussion, the next section provides a brief analysis of postcolonial social cleavages and politicisation of class interests and ethnic identities. This is followed by empirical analyses of the links between ethnic identity and public opinion on peace, and between social differentiation – in terms of occupation and education – and opinions within the Sinhalese majority community. We conclude that ethnic polarisation and politicisation appears to be the foremost determinant of public opinion on peace from 2002-2009. Social space, political field and political position The existing literature on political opinions and allegiances in Sri Lanka revolve around campaigns and voting patterns in parliamentary, presidential, and local elections. Sri Lanka has enjoyed universal suffrage since 1931, more than 60 years of liberal democracy, and has been marked by regime changes mostly through free and fair elections since independence in 1948. Electoral campaigning is normally intense and voter turnout tends to be relatively high, especially in national elections. The existing literature on voter allegiance in Sri Lanka is marked by a divide between structure/society-orientated and actor/polity-orientated approaches. Structure/society-orientated approaches, on the one hand, emphasise how the composition of the electorate determines voter behaviour and election results. These studies reflect the international literature that argues that social cleavages divide people into voting blocs and parties construct their identities and policies to position themselves with regard to class, region, religion, gender, ethnicity, and other cleavages. In Sri Lanka, it is common to emphasise cleavages based on the urban/rural divide, social class, caste, ethnic identities, language, and religion as structural explanations for the character of political parties and voting patterns. There are, on the one hand, studies that emphasise the centrality of ethnic relations in postcolonial politics, in agreement with the logic of Horowitz (1985). Marxian scholars, on the other hand, have emphasised class relations (see, for example, Gunasinghe, 1984, and Jayawardena, 2003), while Jiggins (1979) has argued that caste cleavages are key determinants of party allegiances in Sri Lanka. According to this structural/functional mode of analysis, political parties are seen as reflections of structural cleavages in society. The United National Party (UNP) is commonly presented as a capitalist, pro-west, antiBuddhist party that has received wide support from landlords and urban and rural business interests. The Sri Lanka Freedom Party 39 PCD Journal Vol. II, No. 1 2010 (SLFP) is often portrayed as a nationalist and socialist party that gains support from Sinhala-Buddhists and, in particular, from the intermediate classes within the Sinhalese majority. Leftist parties, such as the Lanka Sama Samaja Party (LSSP), Communist Party (CP), and the Janatha Vimukthi Peramuna (JVP), have their social base in the working and lower-middle classes. Parties such as the All Ceylon Tamil Congress (ACTC), Federal Party (FP), Tamil United Liberation Front (TULF), Tamil National Alliance (TNA), and the Sri Lanka Muslim Congress (SLMC), reflect ethnic minority interests. Actorand polity-orientated approaches, on the other hand, emphasise the political practices of parties and politicians in constructing electorates and mobilising voters. There are, for example, a number of studies that see ethnic polarisation and conflict in Sri Lanka as products of post-colonial politics rather than reflections of primordial ethnic cleavages. Along this line of inquiry, De Votta (2004) argues that the key mechanisms behind post-colonial Sinhalese and Tamil nationalism are to be found in the way liberal democratic institutions and elite rivalry have fostered Sinhalese and Tamil ethnonationalism, resulting in ethnic majoritarianism, minority resistance, and decaying democratic institutions. Focusing more on local political dynamics, Jayanntha (1992) identifies political clientelism as a determining factor in party allegiance and argues that elections are not fought on the ground of party-ideologies and structural cleavages but are instead driven by patrons mobilising their different local networks. Jayanntha’s analysis resonates well with the international literature on the role of political clientelism for voter behaviour and party allegiances in many states in the global South (Kitschelt and Wilkinson, 2005). The general argument here is that symbolic construction and clientelist representation of social groups are key determinants for political opinions and voter allegiances. Habitus and social space While existing studies of electoral allegiances have tended to emphasise either structural cleavages in society or political dynamics in competitive party systems, Chandra (2004) observes that these are complementary rather than mutually exclusive approaches. Social divisions have an undeniable structuring effect on political identities and interests, but the construction and mobilisation of collectivities are also contingent on political agency. What is needed is a conceptual framework that can bridge the divide between structure/societyand agency/polity-orientated approaches. One useful source of inspiration for such a combined approach may be found in the theory of practice developed by Bourdieu. The following paragraphs will highlight elements of his work that are relevant for our analysis. Public Opinion on Peace as a Reflection of Social Differentiation and Politicisation of Identity in Sri Lanka 40 Bourdieu (1990,1998) argues that social practice, including political position taking, can neither be understood as a simple question of individual rationality nor as a direct outcome of social structures, but suggests the notion of habitus as a mediating link between social structures (relations of capital) and practices in a field. Habitus is a system of internalised social norms, understandings, and patterns of behaviour that make it natural to understand the social world and act in certain ways. Social practices are thus formed by the actors’ habitus, but also the capital they possess and the field within which they operate. This can be schematically summarised by the equation: Practice = (habitus * capital) + field (Figure 1). Figure 1. Key concepts in Bourdieu’s theory of practice. Thompson (1991) points out that the dispositions of the habitus are acquired, structured, durable, and transposable. First, they are not given but acquired, particularly through childhood socialisation, and are constituted through un-reflexive and mundane processes of habit formation. Second, dispositions are structured by the social conditions in which they are acquired. This means that individuals from a middle-class background hold dispositions that differ from those produced in a working-class environment. This also means that habitus may be relatively homogenous among individuals from similar backgrounds. Third, dispositions are durable in the sense that they are embodied and operate at the sub-conscious level and are, thus, not easily available for self41 PCD Journal Vol. II, No. 1 2010 reflection and modification. Fourth, dispositions are generative and transposable, meaning that they can generate perceptions and practices also in other fields than those where they were acquired. These observations imply that the habitus is both the product of social conditions and the producer of strategies acting upon those conditions. It is a structuring structure that yields habitual practices without the actors necessarily reflecting on what they are doing. But the habitus is also a structured structure because the embodied dispositions are rooted in social conditions and generated through contextual habituation. It is this dual character of habitus – as a producer of habits and a product of habituation – that makes it a mediating link between social structures and practice (Grenfell, 2008). Figure 2. General forms of capital with possible conversions. The understanding of habitus as a structured structure brings up the question of social differentiation. Bourdieu’s (1986) entry point to socia


Introduction
How did the public opinion on peace in Sri Lanka vary between different social groups and with changing political dynamics during the peace process of 2002-2009?And what explains these variations and changes?Do they reflect class structures in society or are they the outcome of politicisation of ethnic identities?The purpose of this article is to examine the public opinion on peace with regard to two general determinants: social differentiation and politicisation of identities.Empirically, we will use survey data to develop an argument about the links between public opinion, social position, and political mobilisation.
The article is organised in three sections.The first section outlines the conceptual framework for the analysis, drawing especially on Bourdieu's notions of habitus, social space, and political field.Following from this theoretical discussion, the next section provides a brief analysis of postcolonial social cleavages and politicisation of class interests and ethnic identities.This is followed by empirical analyses of the links between ethnic identity and public opinion on peace, and between social differentiationin terms of occupation and education -and opinions within the Sinhalese majority community.We conclude that ethnic polarisation and politicisation appears to be the foremost determinant of public opinion on peace from 2002-2009.

Social space, political field and political position
The existing literature on political opinions and allegiances in Sri Lanka revolve around campaigns and voting patterns in parliamentary, presidential, and local elections.Sri Lanka has enjoyed universal suffrage since 1931, more than 60 years of liberal democracy, and has been marked by regime changes mostly through free and fair elections since independence in 1948.Electoral campaigning is normally intense and voter turnout tends to be relatively high, especially in national elections.
The existing literature on voter allegiance in Sri Lanka is marked by a divide between structure/society-orientated and actor/polity-orientated approaches.Structure/society-orientated approaches, on the one hand, emphasise how the composition of the electorate determines voter behaviour and election results.These studies reflect the international literature that argues that social cleavages divide people into voting blocs and parties construct their identities and policies to position themselves with regard to class, region, religion, gender, ethnicity, and other cleavages.In Sri Lanka, it is common to emphasise cleavages based on the urban/rural divide, social class, caste, ethnic identities, language, and religion as structural explanations for the character of political parties and voting patterns.There are, on the one hand, studies that emphasise the centrality of ethnic relations in postcolonial politics, in agreement with the logic of Horowitz (1985).Marxian scholars, on the other hand, have emphasised class relations (see, for example, Gunasinghe, 1984, andJayawardena, 2003), while Jiggins (1979) has argued that caste cleavages are key determinants of party allegiances in Sri Lanka.According to this structural/functional mode of analysis, political parties are seen as reflections of structural cleavages in society.The United National Party (UNP) is commonly presented as a capitalist, pro-west, anti-Buddhist party that has received wide support from landlords and urban and rural business interests.The Sri Lanka Freedom Party (SLFP) is often portrayed as a nationalist and socialist party that gains support from Sinhala-Buddhists and, in particular, from the intermediate classes within the Sinhalese majority.Leftist parties, such as the Lanka Sama Samaja Party (LSSP), Communist Party (CP), and the Janatha Vimukthi Peramuna (JVP), have their social base in the working and lower-middle classes.Parties such as the All Ceylon Tamil Congress (ACTC), Federal Party (FP), Tamil United Liberation Front (TULF), Tamil National Alliance (TNA), and the Sri Lanka Muslim Congress (SLMC), reflect ethnic minority interests.
Actor-and polity-orientated approaches, on the other hand, emphasise the political practices of parties and politicians in constructing electorates and mobilising voters.There are, for example, a number of studies that see ethnic polarisation and conflict in Sri Lanka as products of post-colonial politics rather than reflections of primordial ethnic cleavages.Along this line of inquiry, De Votta (2004) argues that the key mechanisms behind post-colonial Sinhalese and Tamil nationalism are to be found in the way liberal democratic institutions and elite rivalry have fostered Sinhalese and Tamil ethnonationalism, resulting in ethnic majoritarianism, minority resistance, and decaying democratic institutions.Focusing more on local political dynamics, Jayanntha (1992) identifies political clientelism as a determining factor in party allegiance and argues that elections are not fought on the ground of party-ideologies and structural cleavages but are instead driven by patrons mobilising their different local networks.Jayanntha's analysis resonates well with the international literature on the role of political clientelism for voter behaviour and party allegiances in many states in the global South (Kitschelt and Wilkinson, 2005).The general argument here is that symbolic construction and clientelist representation of social groups are key determinants for political opinions and voter allegiances.

Habitus and social space
While existing studies of electoral allegiances have tended to emphasise either structural cleavages in society or political dynamics in competitive party systems, Chandra (2004) observes that these are complementary rather than mutually exclusive approaches.Social divisions have an undeniable structuring effect on political identities and interests, but the construction and mobilisation of collectivities are also contingent on political agency.What is needed is a conceptual framework that can bridge the divide between structure/society-and agency/polity-orientated approaches.One useful source of inspiration for such a combined approach may be found in the theory of practice developed by Bourdieu.The following paragraphs will highlight elements of his work that are relevant for our analysis.Bourdieu (1990Bourdieu ( ,1998) ) argues that social practice, including political position taking, can neither be understood as a simple question of individual rationality nor as a direct outcome of social structures, but suggests the notion of habitus as a mediating link between social structures (relations of capital) and practices in a field.Habitus is a system of internalised social norms, understandings, and patterns of behaviour that make it natural to understand the social world and act in certain ways.Social practices are thus formed by the actors' habitus, but also the capital they possess and the field within which they operate.This can be schematically summarised by the equation: Practice = (habitus * capital) + field (Figure 1).Thompson (1991) points out that the dispositions of the habitus are acquired, structured, durable, and transposable.First, they are not given but acquired, particularly through childhood socialisation, and are constituted through un-reflexive and mundane processes of habit formation.Second, dispositions are structured by the social conditions in which they are acquired.This means that individuals from a middle-class background hold dispositions that differ from those produced in a working-class environment.This also means that habitus may be relatively homogenous among individuals from similar backgrounds.Third, dispositions are durable in the sense that they are embodied and operate at the sub-conscious level and are, thus, not easily available for self-reflection and modification.Fourth, dispositions are generative and transposable, meaning that they can generate perceptions and practices also in other fields than those where they were acquired.These observations imply that the habitus is both the product of social conditions and the producer of strategies acting upon those conditions.It is a structuring structure that yields habitual practices without the actors necessarily reflecting on what they are doing.But the habitus is also a structured structure because the embodied dispositions are rooted in social conditions and generated through contextual habituation.It is this dual character of habitus -as a producer of habits and a product of habituation -that makes it a mediating link between social structures and practice (Grenfell, 2008).

Figure 2. General forms of capital with possible conversions.
The understanding of habitus as a structured structure brings up the question of social differentiation.Bourdieu's (1986) entry point to social stratification and power is his concept of capital (Figure 2).Individuals and groups are positioned in social space according to their possession of capital, presenting itself in three fundamental forms: (1) economic capital (material wealth in the form of property, money, etc.); (2) social capital (social resources in the form of networks and contacts based on mutual recognition); (3) cultural capital (informational assets in the form of knowledge and skills acquired through socialisation and education, etc.).Bourdieu observes that one form of capital can be converted into another.This convertibility of capital means, for instance, that educational qualifications (cultural capital) or social networks (social capital) can be converted into an attractive job (economic capital).Most importantly, social, economic, and cultural capital can be converted into symbolic capital, meaning that possession of economic, social, or cultural capital is the basis for legitimate authority in a field (Bourdieu, 1990).
This conceptualisation of capital is important for the understanding of social differentiation.Bourdieu (1984Bourdieu ( , 1987) ) argues that people are stratified according to the volume and composition of the capital they possess (Figure 3).The field of power (i.e. the elite) is made up of those who control large volumes of capital, but the elite is divided by the composition of capital.Thus, the field of power does not imply a unified elite but, rather, diverse elite groups that hold different forms of capital and struggle to gain symbolic recognition for their particular kind of capital.Based on this mapping of social space, Bourdieu argues that there is a homology between social positions and dispositions (habitus).This means that the social space is reflected in values and preferences, including political opinions.

The political field and symbolic representation
Habitus and practices are structured by people's positions in social space, but they are also shaped by the context for practice.Bourdieu contextualises practice through the notion of fields, i.e. social systems that are characterised by a competition for accumulation of field-specific forms of symbolic capital.The defining feature of the field is the form of symbolic capital that is at stake.The political field is, for instance, defined by the struggle for accumulation of political capital, meaning that political professionals strive to gain recognition as a legitimate authority with the right to represent non-professions ('the people'): The political field is thus the site of a competition … for the monopoly of the right to speak and act in the name of some or all of the non-professionals.The spokesperson appropriates not only the words of the group of nonprofessionals, that is, most of the time, its silence, but also the very power of that group, which he helps to produce by lending it a voice recognized as legitimate in the political field (Bourdieu 1991: 190).
Positions as political professionals may be based on the personal capital of the spokesperson -for example, factors such as fame and popularity.However, it can also be 'objectified political capital' (recognition and loyalties) that reside in state institutions and political parties and are granted to political professionals.Transfer of political capital between the party and the professional follows the logic of investment in the sense that political professionals, on the one hand, may invest in a party to gain access to its institutionalised political capital.Political parties, on the other hand, may invest in political professionals to reap benefits from their personal capital.Based on observations of French politics at the time of his writing, Bourdieu (1991) argues that the professional depends as much on the party as the electoral constituency for accumulation of political capital.The 'possession' of an electoral constituency depends on the representative's position within a party and breaking away from the party normally means losing the basis for legitimate political authority.This may not hold true, however, for political systems characterised by personalistic and clientelist politics.In such situations, the balance between institutionalised and personal political capital shifts in favour of the latter.The politician's symbolic capital -derived from his/her economic, social, and cultural capital -combined with the capacity to access state resources is what grants political capital.Hence, under Sri Lanka's clientelist political system and elections based on proportional representation and multiple voter preferences, a politician can break away from the party and still retain his seat, thereby exploiting new bases (Peiris, 2010).
To gain legitimacy as a political spokesperson, competing political actors (parties and politicians) engage in symbolic struggles to impose and normalise representations of the world that suit their own interests.The power of the ideas proposed by these political actors is measured by their ability to produce and mobilise a group of people, meaning that they must resonate with the habitus of those they intend to mobilise.Bourdieu (1991) argues that this symbolic relationship between political representatives and those that are being represented is structured by the competitive relations between political professionals.Therefore, political representation is determined both by the relationship between representatives and represented, and between competing representatives (Stokke and Selboe, 2009).The relevance of this for our analysis is that political construction and politicisation of issues and identities, stemming from the struggle for accumulation of political capital, constitute a source of public opinion.Thus, we have two different but interrelated logics that might frame public opinion: one stemming from structural cleavages in society and the other originating from strategies and relations in the political field.This general framework will guide our empirical analysis of public opinion on peace in Sri Lanka.

Post-colonial cleavages and political mobilisation
Post-colonial Sri Lankan politics has been characterised by complexity and political fragmentation, originating both in multi-dimensional cleavages in society and from divisions and competition within the political elite.It can be observed that class and ethnicity have played a pivotal role in post-colonial politics, but this is compounded by identities and allegiances based on religion, language, caste, region, gender, and family.It can also be noted that there has been a shift in the relative importance of class and ethnicity.While there was a clear primacy of class politics in the early post-colonial period, this has been overtaken by a growing politicisation and polarisation of ethnic identities.
Sri Lankan politics at the time of independence was marked by colonial class cleavages and politicisation of class interests rather than ethnic identities.Colonialism had produced a multiethnic dominant class that was subordinated to British capital but also far removed from the domestic popular classes (Jayawardena, 2003).The political project that was pursued by this multi-ethnic elite was essentially a continuation of the colonial accumulation regime and elite domination (Uyangoda, 1992).Colonialism had also institutionalised ethnicity as an administrative category and linked communal identities to political representation, but communal tensions were largely confined to debates about constitutional arrangements and were overshadowed by elite collaboration across ethnic divides (Wickramasinghe, 2006).The multi-ethnic dominant class was challenged by leftist parties and trade unions led by western-educated radical intellectuals, politicising the interests of the working class and advocating social revolution (Jayawardena, 1985).Thus, early post-colonial mass politics was characterised by elite-led party politicisation of class interests, with ideological polarisation between the rightist United National Party (UNP) and All Ceylon Tamil Congress (ACTC), on the one hand, and the leftist Lanka Sama Samaja Party (LSSP) and Communist Party (CP), on the other.This situation is presented as a Bourdieu-inspired map of the political field in Figure 4.The class politics of the early post-colonial period was also characterised by political under-representation, despite numerical dominance of intermediate classes, including peasants, small traders, public sector employees, and monks.These intermediate classes became the subject of ethnonationalist political incorporation from the 1950s, propelled by political competition within the elite as much as by class cleavages in society.A breakaway section from the UNP formed the Sri Lanka Freedom Party (SLFP) and pursued mobilisation of intermediate classes through a combination of religious, linguistic, and livelihood issues among the Sinhalese Buddhist majority, placing themselves in an antagonistic relationship to both the anglicised elite within the UNP and the class politics of the leftist parties (Figure 5).This strategy of merging intermediate class interests and ethnic identity delivered a landslide electoral victory for SLFP in the 1956 elections and has been a hegemonic electoral strategy for all Sinhalese parties since then (De Votta, 2004;Manor, 1989).
The elitist political incorporation within the Sinhalese majority was paralleled by similar developments within Tamil minority politics.A section broke away from the ACTC and formed the Federal Party (FP) in the 1950s, mobilising popular support for a nonviolent campaign for Tamil self-determination within a federal state (Wilson, 1988).Thus, the early post-colonial period was marked by the rise of two parallel ethnonationalist political projects, emerging from colonial identity constructions and formulated in opposition to conservative-and radical-class politics as much as the ethnonational 'other'.This shift from class to ethnic politics prepared the ground for Sinhalese majoritarianism, Tamil minority resistance and the escalation of ethnic conflict in subsequent decades, but there were also important tendencies towards social and political inclusion.While electoral competition between SLFP and UNP gave leftist and minority parties leverage in political negotiations and government coalitions, state-led development and universal welfare programmes supported social development and mobility, especially for the intermediate classes (Moore, 1989;Shastri, 1983).This situation is depicted in Figure 5.
Whereas the late 1950s and the 1960s witnessed the growth of Sinhalese and Tamil ethnonationalist mobilisation in combination with state-led development and welfarism, the 1970s were marked by development crises, social exclusion, and ethnic polarisation.Economic stagnation and voter frustration with unfulfilled campaign promises produced a landslide electoral victory for a SLFP-left coalition government in 1970, but widespread unemployment and soaring costs of living undermined the legitimacy of the government and led to an insurgency by the Janatha Vimukthi Peramuna (JVP), mobilising marginalised youth from the Sinhalese intermediate classes (Obeyesekere, 1984).In this situation, the government responded with a combination of state socialism, ethnonationalist populism, clientelist concessions to key constituencies, and authoritarian repression of JVP, Tamil-elite parties, and emerging militant groups.The strong majority held by the government undermined the political leverage of the minorities, thereby marginalising the Tamil political elite and weakening their legitimacy vis-a-vis radicalised youth groups (Figure 6).In turn, the Tamil political elite joined forces in the Tamil United Liberation Front and radicalised its demands from federalism to a separate Tamil state.TULF advocated non-violent and democratic means, but this was overtaken by the growth of separatist militancy in the face of state repression and anti-Tamil riots in the late 1970s and early 1980s (Hellmann-Rajanayagam, 1994;Swamy, 1994).The Tamil and Sinhalese militant movements that emerged in the 1970s both relied on ethnonationalism to contest and lay claim on political authority.The socialist-ethnonationalist strategy that had been developed for the purpose of elitist political incorporation of intermediate classes in the 1950s and 1960s was, therefore, appropriated and radicalised by non-elite forces in the context of social and political exclusion.Hence, ethnic polarisation and emerging militancy from below were key characteristics of Sri Lankan politics in the 1970s and the early 1980s.This situation is visualised in Figure 6.

Figure 6. Ethnonationalist polarisation and emerging militancy (1970s and early 1980s).
Following the growing ethnic polarisation and militancy in the 1970s, the period from the early 1980s to 2009 was marked by the escalation of ethnic polarisation into armed conflict.In the context of political centralisation, marginalisation of minority representatives, authoritarian rule and anti-Tamil riots, Tamil nationalism was transformed from TULF's democratic campaign for a separate state to the hegemony of militant separatism (Figure 7).
From the anti-Tamil riots in 1983 to the military defeat of the Liberation Tigers of Tamil Eelam (LTTE) in 2009, Sri Lanka was marked by an armed conflict between the Government of Sri Lanka (GOSL) and militant Tamil separatism (Balasingham, 2004;Swamy, 1994).The Tamil separatist movement consisted initially of several militant groups, but organisational, ideological, and personal animosities produced armed clashes within and between the different organisations.The LTTE emerged as the dominant militant organisation in the late 1980s, annihilating or subordinating the other groups.Some of the non-LTTE groups entered into collaboration with the GOSL, exchanging intelligence and paramilitary operations with protection, material rewards, and political power.This situation is schematically represented in Figure 7.The period since the change of government in 2004 and the victory of Mahinda Rajapakse in the presidential election in 2005 has been marked by both continuity and change with regard to the previous periods of post-colonial politics (Figure 8).It can be observed that the political centralisation that was constitutionalised in 1978, with the president as the pivotal political authority, has been furthered in a highly personalistic way.After becoming president, Mahinda Rajapakse has built a political dynasty around a closely knit network of family, kinship relations and political professionals with personal loyalty to himself.This personalistic political network has, to a large extent, replaced the pre-existing SLFP party structure, while political networks and loyalties also cross party political divides, especially between the two main parties SLFP and UNP.The final stage of the war as well as the post-war period has also been marked by militarisation and state coercion against the Tamil insurgency, critical journalists, civil society organisations, and the political opposition.While political clients enjoy patronage in the form of positions of power and material benefits, political parties and individuals that are opposed to the ruling network have been targeted for repressive measures.
The present period is marked by Sinhalese majoritarianism with marginalisation of minority political parties and questions of group rights and devolution of power.Political representatives and parties from the ethnic minorities face the dilemma between entering into patronage politics, that is experienced as an unproductive cooptation, or to remain politically ineffective in the opposition and become potential targets of state repression.Whereas the Tamil National Alliance (TNA), a political alliance that emerged from the old Tamil parties and was close to LTTE during the ceasefire agreement period, has chosen a position in the political opposition, former militant groups such as Eelam People's Democratic Party (EPDP) and Tamil Makkal Viduthalai Pulikal (TMVP) have entered into clientelist networks and have gained positions in national politics and at the province level in the north and east.Within the Sinhalese polity, the ethnonationalist Jathika Hela Urumaya (JHU) has remained loyal to the ruling government while the JVP has broken out of the government coalition.
In this situation, benefits and opportunities from the political field for both elite and non-elite actors are increasingly dependent on skilful handling of political networks and allegiances.Popular mobilisation and political loyalties rely increasingly on ethnonationalist populism and personalistic clientelism, while nonethnicised political programmes play a marginal role for political mobilisation and opinions.This situation is tentatively outlined in Figure 8.

The Sri Lankan peace process
The 26 years of warfare in Sri Lanka was interspersed by attempts at negotiated conflict resolution in 1985 (the Thimpu Talks), in 1987 (the Indo-Lanka Accord), in 1989-1990(the Premadassa-LTTE talks), in 1994-1995 (the Kumaratunga-LTTE talks) and in 2001-2002 (the Wickremasinghe-LTTE talks) (Balasingham, 2004;Gooneratne, 2007;Rupesinghe, 2006;Stokke and Uyangoda, 2011;Uyangoda and Perera, 2003).None of these managed to resolve the conflict, but the 2002-2003 talks produced a ceasefire agreement that formally existed until January, 2009.Thereafter, the conflict ended through a final war and military defeat of the LTTE in May 2009.Before turning to our empirical analysis, we will provide a brief review of the peace process and its different stages in order to make sense of the changing public opinion on peace during the ceasefire period.
The design and dynamics of the peace process were first and foremost shaped by: (1) a military-territorial balance of power that brought GOSL and LTTE to negotiations and kept them from resuming warfare for several years after the talks stalled; (2) entrenched constitutional and political barriers to state reforms for devolution of power to minority areas; (3) severe crises of development in terms of large-scale humanitarian needs in war-affected areas and slow and uneven economic growth throughout the island (Bastian, 2007;Liyanage, 2008;Uyangoda and Perera, 2003;Stokke and Uyangoda, 2011).First, the balance of power between LTTE and GOSL made the peace process resemble international conflict resolution between incompatible state-building projects, i.e. between GOSL's aim of rebuilding the unitary Sri Lankan state and LTTE's goal of achieving self-determination for Tamils in a separate state.The ceasefire agreement froze the military-territorial balance of power, segmented a de facto dual-state structure and institutionalised a degree of parity between the two parties in the peace process.Second, the institutional and political obstacles to power sharing, stemming from constitutionalised political centralisation, the weak majority of the government, and the practices of ethnic outbidding and instrumental opposition to peace within Sinhalese politics, gave the process a pragmatic design where immediate security and humanitarian concerns were prioritised at the expense of political conflict resolution.Third, the crises of development provided a point of convergence between the GOSL and LTTE while also being the basis for a strong internationalisation of peace through the participation of Sri Lanka's international aid donors (Bastian, 2007).Combined with the political/institutional obstacles to substantive conflict resolution, this gave the peace process its distinct character of pursuing 'peace through development' (Shanmugaratnam and Stokke, 2008).
In terms of the different stages in the peace process, we have elsewhere argued that the period from the initiation of the peace process to the military defeat of LTTE in 2009 could be divided into three distinct periods: (1) active-peace negotiations from 2002-2004; (2) negative peace from 2004-2006; (3) war for peace from 2006-2009.This periodisation is also the basis for the empirical analysis in the present article.
The active-peace period (2002-2004) followed after the government's invitation to Norway to facilitate peace negotiations between the LTTE and the GOSL.The marginal electoral victory of the liberal peace and market friendly UNP-led coalition in 2001 paved the way for Norwegian-facilitated talks with the LTTE.A ceasefire agreement was signed in February 2002 between the GOSL and the LTTE and was followed by six rounds of talks in 2002-2003.These were mainly 'talks on talks' in the sense that they focused primarily on non-core issues as a trust-building basis for future talks (Gooneratne, 2007;Liyanage, 2008;Uyangoda and Perera, 2003).The signing and implementation of the ceasefire agreement was the main achievement and created a relatively peaceful situation despite numerous ceasefire violations.The ceasefire agreement also provided a space for internationally funded humanitarian rehabilitation in war-affected areas.In the absence of substantive conflict resolution, development became the prime concern and point of convergence between the protagonists to the conflict.However, development also became highly contentious as the question of power sharing in interim development administration for the north-east impinged on future power-sharing arrangements the negotiations to a stalemate (Shanmugaratnam and Stokke, 2008).Simultaneously, the two parties to the negotiations were weakened by internal changes in their ethnic constituencies.While the government was increasingly challenged by Sinhalese nationalist agitation against the government, the peace process, and the role of the international actors, (especially Norway), the LTTE was weakened by an internal split when their eastern commander defected with large numbers of cadres.Finally, the presidential take-over of three key ministries in 2004 made it increasingly difficult to restart the stalled negotiation process. 1  The negative-peace period (2004)(2005)(2006) covers the period of no war/no peace from the change of government in 2004 to the gradual resumption of hostilities from 2006.During this period, the ceasefire agreement was formally upheld but the two parties did not show interest in pursuing active-peace negotiations.The period was instead marked by numerous ceasefire agreement violations as well as clashes between the LTTE and the breakaway Karuna faction in the Eastern Province.The tsunami disaster in December 2004 delayed the resumption of warfare, but the parties failed to use the disaster as an opportunity to restart the peace negotiations.An attempt to establish a joint mechanism for managing humanitarian aid to the disaster affected areas in the north and east failed due to legal and political obstacles (Rainford and Satkunanathan, 2011).This was considered a victory for Sinhalese nationalism, as was the election of Mahinda Rajapakse from SLFP as president in 2006.At the same time, there was a gradual shift within the LTTE away from political negotiations to militant means, most strikingly manifested in the assassination of Sri Lanka's Foreign Minister in 2005.A final attempt to restart the peace process through two meetings in Geneva in 2006 failed, thus paving the way for a return to armed hostilities.The 2004-2006 period was thus marked by an absence of open warfare and a lack of real progress towards lasting peace.This is a situation that peace researchers describe as 'negative peace' (Uyangoda, 2005).
The war-for-peace period, from 2006-2009, was characterised by the gradual resumption of warfare -first through covert attack by both parties and later through direct military engagementwhich culminated in full-scale war in 2009 and military defeat of the LTTE in May, 2009.As the Sri Lankan armed forces moved into LTTE-controlled areas, civilians were displaced in large numbers and the violations of human rights increased.These human rights violations included extra-judicial killings, extortions, abductions, disappearances, and the continuation of the culture of impunity.The LTTE, too, intensified their attacks on the security forces, but the Sri Lankan armed forces were superior in terms of intelligence, military hardware, man power, and domestic and international backing.This allowed them to pursue an aggressive, coordinated, and non-interrupted military campaign.The LTTE was internationally isolated and confronted by an enemy with political determination, military capacity, and popular support.Thus, the Sri Lankan armed conflict ended through military means, allowing the government to dictate the conditions of peace for the defeated enemy.

Public opinion on peace during the peace process
How did the public opinion vary and change in the context of Sri Lanka's peace process and what explains these variations and changes?The best source for empirical data on this question is the Peace Confidence Index (PCI) compiled by Social Indicator, the survey research unit of the Centre for Policy Alternatives in Colombo. 2 Before we begin to explore the empirical data, it is necessary to introduce the PCI and how we organised the data for this analysis (Centre for Policy Alternatives, 2010aAlternatives, , 2010b)).
The PCI is an island-wide opinion survey designed to capture the trends in public opinion related to the peace process that began in the middle of 2001.Initially the PCI was a bi-monthly survey.However, it was later changed to a quarterly survey.Each survey interviewed approximately 1800 individuals from the Sinhalese, Tamil and Muslim communities across the country, except in areas in the Northern and Eastern provinces which were controlled by the LTTE during the ceasefire agreement period.Therefore, the PCI captured the opinions of the entire Sinhalese community and the opinions of the Tamil and Muslim communities who lived outside the conflict zone.Following the escalation of violence in the Northern and Eastern provinces and the worsening situation of human rights violations in the south during the negative-peace period, some of the PCI surveys could not cover the opinion of the Tamil community.
For the present analysis, we have selected two surveys -the first and third surveys -from each year.Since the analysis was done according to the aforementioned three periods -active peace, negative peace and war for peace -the data sets of the selected surveys in each period were merged together.This meant that the total sample size exceeded 4000 interviews for each period, which allowed us to perform our analysis with a great deal of reliability.Finally, we used the demographic data to construct new education and occupation categories to inquire into the relationship between the public opinion on peace and social differentiation.
The questionnaire used for the PCI carried a set of standard questions throughout the whole period in order to facilitate longitudinal analysis, while current political developments were captured through a set of questions that varied from one survey to another.We have extracted data regarding three key questions about the peace process in Sri Lanka: (1) the preferred mode of conflict resolution; (2) the commitment of GOSL and LTTE to negotiated peace; (3) the support for the role of the international community in the peace process.These will be analysed according to the ethnic identities and social stratification of the respondents.

Ethnicisation of the public opinion on peace
The PCI data shows that the public opinion on peace is highly influenced by the ethnic identity of each individual.Figures 9, 10, and 11 summarise the communal opinions for the selected questions and show distinct variations and changes over time within the three ethnic communities.
The most fundamental question in popular and political discourse throughout the Sri Lankan conflict has been about the mode of conflict resolution: Should the conflict end by political or military means?In the course of the conflict, attempts at negotiating peace between the government and the Tamil insurgency were interspersed by periods dominated by warfare.The ceasefire period also went from a situation dominated by direct negotiations to a phase of negative peace before ending in full-scale warfare.Against this background, Figure 9 shows the public opinion with regard to three general modes of conflict resolution: peace negotiations (liberal peace), military solution (war for peace) or a combination of the previous two (a military weakening of the opponent followed by peace negotiations).The figure shows a divide in the public opinion between the Sinhalese majority and the Tamil and Muslim minorities.The PCI indicates that the Sinhalese community supported a negotiated settlement nearly as much as their minority counterparts during the active-peace period.While almost all Tamils (98.6 per cent) and Muslims (97.4 per cent) preferred a negotiated solution, nine out of 10 Sinhalese (88.9 per cent) did.However, even at the height of the peace process and the pro-peace media campaign by the UNP government, 8.3 per cent of the Sinhalese community preferred a military solution.As the peace process stalled and was replaced by open warfare, support for negotiations decreased and the divide between the majority and minority communities increased.While a clear majority of Tamils (84.4 per cent) and Muslims (84.8 per cent) maintained their support for negotiations during the war-forpeace period, the support for negotiations among the Sinhalese declined to 33 per cent.The largest group of Sinhalese (47.2 per cent) preferred a military solution, whereas very few Tamils (3.4 per cent) and Muslims (3.6 per cent) viewed this as a desirable option.Beyond the basic question of political or military conflict resolution, the PCI also showed that there were communally divided opinions on the meaning of negotiations, both in terms of who should participate in the negotiations and what kind of power sharing arrangements were deemed preferable (Peiris and Stokke, 2011).The preference among the Sinhalese was for negotiations to ensure participation of all stakeholders -from the government, the opposition, LTTE, Tamil and Muslim political parties, etc.The Tamil community, too, preferred inclusive negotiations, but Tamils expressed a stronger support for international facilitation than the Sinhalese.The clear Muslim preference was for a process with maximum inclusivity of all stakeholders.With the examination of data over time, a growing preference for inclusivity could be found in all communities, while the support for exclusive negotiations between the GOSL and LTTE decreased.This means that there existed a widening gap between the public opinion on inclusivity and the actual design of the peace process.The 2002-2003 peace negotiations were designed as narrowly defined talks between LTTE and GOSL, facilitated by Norway.The political opposition, the Muslim minority, non-LTTE Tamil actors, civil society organisations, and the Buddhist Sangha were all excluded from the peace process, and there was no systematic parallel process aimed at building a broader consensus on peace.This design reflected the military balance of power between the GOSL and LTTE and was also based on the assumption that the two parties could negotiate on behalf of their communal constituencies.This narrow negotiation process raised critical questions and debate about the commitment and strategies of the two parties.Figure 10 shows that the popular opinion on the protagonists' commitment to negotiated peace varied between communities and over time.It can be observed that all communities had a fairly high level of trust in the government's commitment to negotiated peace during the active-peace period.Interestingly, Tamil (56.7 per cent) and Muslim (70.9 per cent) minorities showed a higher trust in the government than the Sinhalese (49.0 per cent) majority at the time.The Tamil (73.7 per cent) and Muslim (45.6 per cent) minorities also showed considerable trust in the LTTE's commitment.This was in sharp contrast to the Sinhalese opinion (15.5 per cent) on LTTE's commitment to negotiated peace.Additional data from the PCI survey showed that the prevalent view among Sinhalese was that LTTE was engaging in peace talks for tactical reasons, a sentiment that was also shared by a number of Muslims (Peiris and Stokke, 2011).
Figure 10 also shows that the strong confidence that the Tamil and Muslim minorities had in the GOSL and LTTE's commitment to a negotiated settlement of the conflict eroded in the course of the peace process.A more disaggregated analysis shows that the confidence of the Sinhalese majority, as well as the Tamil and Muslim minorities, declined at the end of the active-peace period before temporarily going up again after the 2004 tsunami disaster (Peiris and Stokke, 2011).After the change of government in 2005/2006, the confidence of the Sinhalese community in the government's commitment to peace remained high despite the hard-line ethnonationalist stance of the Rajapakse government, while the confidence of the minority communities declined rapidly.
Another defining feature of the process was the internationalisation of peace in Sri Lanka.The US-led international community was actively involved in the attempt to craft liberal peace through roles as facilitators of peace negotiations (particularly Norway), monitors of the ceasefire agreement (the Nordic countries), and donors of aid for rehabilitation and reconstruction (especially the co-chairs to Sri Lanka's donor conferences: Japan; European Union; Norway; US).While the internationalisation of peace in Sri Lanka was placed within the context of international security concerns, the international involvement in the peace process revolved primarily around domestic peace-development links rather than the global security-development nexus (Orjuela, 2011;Stokke, 2011).As the peace process broke down and gave way to a war for peace, the emphasis on liberal peace among the international actors was gradually replaced by concerns about state sovereignty, security, and defeating terrorism.Norway played an especially prominent role as facilitator of the negotiations, co-organiser of the Sri Lanka Monitoring Mission (SLMM).and co-chair and donor at Sri Lanka's donor conferences.The obstacles that were encountered in each role and the problems of combining three roles rendered Norway's engagement in Sri Lanka controversial, especially when the negotiations stalled and the peace process became increasingly politicised and contested (Höglund and Svensson, 2011;Stokke, 2011).Figure 11 shows the public support for the international actors in general and the specific role of Norway.The figure shows that public opinion toward the international actors was divided according to ethnicity, with Sinhalese respondents expressing less support than the Tamil and Muslim communities.All three communities, however, displayed relatively stable support for the international community throughout the ceasefire period.When it comes to Norway's role, two main observations can be made.First, there was much higher support for Norway' role in the Tamil (86.8 per cent) and Muslim (60.3 per cent) minorities than in the Sinhalese (31.2 per cent) majority throughout the peace process.Even during the active-peace period, when some positive outcomes of the liberal peace project materialised, only one third of the Sinhala community supported Norwegian facilitation.Second, there was a steep decline in the support for Norway's involvement among all communities when the process entered the negative-peace phase.The support for Norway disappeared almost completely in the Sinhalese (5.2 per cent) community once the negotiation process stalled and the violence between the two protagonists began to rise.It is noteworthy that even the very high degree of support that minority communities extended during the active-peace period drastically dropped among Tamils (37.7 per cent) and Muslims (24.6 per cent) when the activepeace period was replaced by negative peace.
Figure 11 highlights a puzzling question about why there was widespread and relatively stable support for international involvement despite the support for Norway's role being divided between communities and weakened over time.Additional PCI data showed that the Sinhalese majority preferred Indian assistance in the peace process (Peiris and Stokke, 2011).Tamils, in contrast, expressed strong support for Norway, although they supported Indian assistance as well.This indicates that Figure 11 reflects communalised perceptions and politicisation of different international actors.India, on the one hand, was long viewed as a protector of Tamils in Sri Lanka, but had also come down hard on LTTE and proscribed it as a terrorist organisation.Norway, on the other hand, was seen as being supportive of minority grievances.The Norwegians' insistence on impartiality and parity of status between LTTE and GOSL was seen as tilting the balance in favour of the LTTE.Thus, there was relatively high and stable support for internationalisation of peace, but ethnically divided opinions on Norway as a third party to the negotiation process.

Social differentiation and public opinion on peace
Having observed a clear ethnicisation of the public opinion on peace, the question is whether opinion is also divided according to class stratification.In order to answer this question, we have used the demographic information in the PCI survey to examine opinions within the Sinhalese majority according to occupation and education.For the analysis of occupation, the respondents were aggregated into three broad categories: (1) high-level jobs; (2) lowlevel jobs; (3) unemployed.Those who worked as professionals and in middle-level jobs were slotted into the first category.This meant that executives, managers, big and small businessmen, teachers, officers, clerks, and all the middle-level formal employments were considered as high-level jobs.Those who engaged in labourintensive work in the formal or informal sectors were categorised as low-level jobs.The use of 'high' and 'low' does not imply a normative judgement on the occupations, but is simply a descriptive tool to group positions in social space.The third category, 'unemployed' comprised of diversified groups such as those who were unemployed at the time, those waiting to be employed, those who were still students, and housewives who, although technically employed, usually identified themselves as unemployed.Similarly, a more diverse range of educational levels were divided into two broad categories of 'up to ordinary level', or basic education, and 'ordinary level and above', or higher education.Figures 12, 13, and 14 show public opinion on peace within the Sinhalese majority according to occupation. Figure 12 shows a slightly stronger preference for negotiations than a military solution to the conflict among those that are unemployed (in both periods) and among those in low-level jobs in the warfor-peace period.Conversely, there is a stronger preference for a combination of military and political means of conflict resolution among people in high-level jobs in the war-for-peace period.Despite these differences, the most striking finding is that the preferences regarding mode of conflict resolution show relatively little variation across occupational categories.
Figures 13 and 14 show the Sinhalese opinion on GOSL and LTTE commitment to negotiations and on the role of international actors, both analysed according to occupation categories.Regarding the GOSL and LTTE's commitment to peace, the most notable pattern is that the public opinion shows very little variation between different occupation categories (Figure 13).The same observation holds true for the opinions on the role of international actors, in which the foremost observation is that there is little variation between the different occupations (Figure 14).However, it can be noted that those in high-level jobs and those who are unemployed express greater support for the international actors, and that there is a marked decline in the support among high-level jobs during the war-for-peace period.Figures 15,16,and 17 show the same kind of analysis of Sinhalese opinions according to education level (up to or above ordinary level).The most striking finding is again that there are no obvious variations in the opinions among those with basic education and those with higher education.Both groups show the same tendencies as the Sinhalese community as a whole, with very small and no systematic differences between them.It can also be mentioned that an additional analysis of public opinion in urban and rural areas yields no major difference across the urban-rural divide.

Figure 15. Sinhalese opinions on mode of conflict resolution
according to education level.
The occupation-and education-based analyses show that public opinion on peace within the Sinhalese majority is not stratified according to class variables.This is in sharp contrast to the earlier analysis based on ethnicity.The classlessness in the public opinion on the peace process within the Sinhala community does not, however, mean that class is politically insignificant in general or for the questions of peace.Our argument is rather that the ethnicisation and classlessness of public opinion is a product of how people from various class strata have been incorporated into mass politics through ethnonationalist projects, yielding ethnic dispositions for knowing and acting that resonate with the way the conflict and its resolution is construed as an ethnic issue in political and media discourse.This shifts the focus from ethnic identities and class cleavages to political and media discourse on the peace process.

Cleavages, politicisation and public opinion
The analysis points to the prominence of ethnic identities and the relative insignificance of social differentiation within the Sinhalese community in the formation of public opinion toward peace during the ceasefire period.Furthermore, changing popular views over time point to the close links between the dynamics of the peace process and the public opinion.These ethnicised patterns and trends in public opinion can be read in different ways.One interpretation could be that the communal variations in public opinion reflected ethnic cleavages in society, in the sense that fixed ethnic identities constituted a basis for ethnicised assessments of the design, dynamics, and actors in the peace process.Another interpretation could be that the observed trends stemmed from manipulation by political elites, communicated to people through political and media discourse on the peace process.This interpretation gives primacy to the political elite while reducing the public to manipulable subjects and rendering it submissive to ethnicised strategies of governmentality.
There is no doubt that political and media discourse played a key role in shaping the public opinion on peace in Sri Lanka during the ceasefire period.Nadarajah (2005) shows that the coverage of the peace process in the vernacular press became increasingly polarised along communal divides, reinforcing essentialist notions of ethnicity and providing ethnicised interpretations of key issues and actors in the peace process.For example, the Sinhala press was welcoming international participation in the form of Indian intervention, while being highly critical and suspicious of Norway's role.Tamil media, in contrast, was invariably supportive of the peace process and Norwegian facilitation.Such media framing of the peace process resonates with the communal opinions on the international actors that we have identified above.Similar links between media discourse and public opinion can also be made for other key issues such as the questions of power-sharing arrangements, the commitment of GOSL and LTTE to negotiated peace, and interim administration in the north and east.Nadarajah (2005) also highlights that the Sinhala, Tamil and Muslim vernacular press were generally positive in the early stage of the peace process, while gradually increasing the critical attention to the process, thus providing a discursive context for the changing public opinion on peace.
Such observations lead to the conclusion that the observed ethnic divides and changes in public opinion reflected the way the peace process was politicised during the different stages of the ceasefire period.We do not, however, see this as a simple matter of manipulation by calculating political elites, but rather as an articulation between political and media discourses and the dispositions for knowing and acting in the habitus of people.In the context of post-colonial political dynamics, it became commonsense for elite and popular actors to comprehend the conflict and the peace process in an ethnicised manner, and this was expressed and reinforced by political and media discourses.Such discourses are themselves closely related to the competitive struggle for legitimate authority in the political field.Thus, the communal divides and changes in public opinions during the ceasefire period were, in our view, reflective of both the general post-colonial construction of ethnic habitus, the politicisation of ethnic identities that emanated from the dynamics of the political field, and the politicisation of peace in political and media discourses during the different stages of the peace process.
Endnotes 1 The active peace period was marked by a contentious co-habitation between the government led by Prime Minister Ranil Wickremasinghe from the UNP and the directly elected President Chandrika Bandaranaike Kumaratunga from the SLFP. 2 www.cpalanka.org

Figure 1 .
Figure 1.Key concepts in Bourdieu's theory of practice.

Figure 3 .
Figure 3. Social space and orientations towards left and right politics (after Bourdieu 1998).

Figure 9 .
Figure 9. Opinions on mode of conflict resolution according to ethnicity.

Figure 10 .
Figure 10.Opinions on GOSL and LTTE commitment to negotiated peace according to ethnicity.

Figure 11 .
Figure 11.Opinions on the role of international actors according to ethnicity.

Figure 12 .
Figure 12.Sinhalese opinions on mode of conflict resolution according to occupation.

Figure 13 .
Figure 13.Sinhalese opinions on GOSL and LTTE commitment to negotiated peace according to occupation.

Figure 14 .
Figure 14.Sinhalese opinions on the role of international actorsaccording to occupation.

Figure 16 .
Figure 16.Sinhalese opinions on GOSL and LTTE commitment to negotiated peace according to education level.

Figure 17 .
Figure 17.Sinhalese opinions on the role of international actors according to education level.