Marriage, Violence, and Choice

The literature on Dalit women largely deals with issues of violence and oppression based on intersections of class, caste, and gender. Women’s bodies, sexuality, and reproductive choices are linked to the ideological hegemony of the caste–gender nexus in India, with marriage and sexual relations playing crucial roles in maintaining caste boundaries. Often, the ways in which women manipulate their multiple, interlinked identities as women, Dalits, workers, and homemakers to resist control over their bodies (labor and sexuality), negotiate conjugal loyalty and love, and construct a sense of selfhood is missed in the analyses. Based on research in rural Tamil Nadu, I analyze Dalit women’s narratives that reflect multiple concerns and dilemmas about marital choice and violence, generating in the process a deeper understanding of agency, voice, and gender relations, as fluid, dynamic, and intersecting in response to changing experiences, positionalities, and subjectivities.

and employers, police and their own men, to the silent and hidden violence of everyday hunger, debt, and bondage, exclusion from schools and hospitals, uncertainty, and sex (Anandhi 2013;Kapadia 1995;Racine and Racine 2000). Despite constitutional guarantees, such violence persists, maintained by state and civil impunity, if not active collusion with the perpetrators in the public domain, alongside its normalization as a marker of authority and control in the domestic domain.
Dalit women have largely been homogenized, their triple burdens of economic marginalization, caste discrimination, and gender subordination almost taken for granted in the social sciences (Deshpande 2011;Shah et al. 2006;Srinivas 1998). Recent attention to women's narratives and representations of their own lives, and the ways they exercise agency and voice in their daily interactions, reveals much more nuanced and often contradictory accounts, speaking of gains and losses, and experiences of violence alongside those of love and desire (Bama 2000(Bama , 2005Geetha 2000;Kleinman 2000;Ravikumar and Azhagarasan 2012;Racine and Racine 2000). Their particular location within the caste hierarchy and their subject-position, including age, life cycle, work, and educational status, their personalities and quality of relationships, mutually intersect to shape their agency. Seeking affirmation in a collective mode, they also articulate and assert concerns of gender within the constructions of Dalit as a community (Rege 2006).
Struggles over status, class, and caste/race are inscribed onto women's bodies, sexuality, and notions of femininity (Skeggs 2009). India is no exception, with control over women's sexuality, especially marriage and fertility decisions, playing a central role in the allocation of economic resources (e.g., dowry and inheritance) and maintenance and reproduction of social inequality and status mobility (Dube 2003;Dyson and Moore 1983). Respectability as a marker of status, however, goes beyond economic differences to moral values attributed to particular cultural categorizations. Women constitute "patriarchy's embodied honour, men lose honour through the behaviour of women from their families" (Rege et al. 2013, 35). Rege (2006) has argued that despite the recognition of their economic contributions in a wider context of poverty and deprivation, and despite articulating concerns of gender, Dalit women remain subject to patriarchal domination within their own communities. Taking off from this argument, I explore the multiple ways in which women's agency and resistance is expressed and changes in response to both the local context and changes in their life course. In trying to distance themselves from the negative values attached to their caste/class, Dalit women may be obliged to uphold middle-class gendered norms of sexuality and marriage, linked to patriarchal notions of domesticity and service, while simultaneously critiquing this morality as pretentious (Ilaiah 1996). Engagement with poorly remunerated and sexually exploitative work is often not a choice. Opportunity and vulnerability go hand in hand; they are neither the passive victims nor the promiscuous egalitarians of stereotype. The key contribution of this article is a fine-grained analysis of women's agency-a complex mix of subjection, conformity, and resistance, in response to the ongoing gender politics within an often homogenized group.
In the next section, I discuss the key conceptual ideas of agency, voice, and intersectionality in relation to the role of marriage and sexuality in reinforcing caste and gender boundaries, then set out the methodology and profile the local context before discussing the main insights from the field. I draw on narratives from Dalit women in rural Tamil Nadu to reflect on their divergent experiences of marriage and sexuality and what it means for both gender relations and their own sense of personhood. I focus particularly on the choice of marriage partner and the maintenance of bodily integrity as reflected in their ability to deal with violence. I conclude with some theoretical and methodological reflections on the construction of Dalit women as a unified category, and its implications for their agency.

MaRRiage and SexualiTY in india: ReinfoRcing caSTe and gendeR BoundaRieS
Studies of marriage and sexuality in India have traditionally emphasized its structural role in constructing and maintaining exclusionary boundaries of caste/ethnicity, primarily through developing and enforcing rules and norms for partner selection, and controlling deviance (uberoi 1993;Yuval-Davis 2006). Social reform debates in colonial and postcolonial India have also centered on contestations over the use of marriage and sexuality to maintain group identity. These concepts are, however, not problematized (uberoi 1996, xvi), nor linked to emerging subjectivities related to wider social changes (Ram 2000). Women's agency then remains tied to their caste/class position and the nature of marriage (crosskin or non-kin, arranged or love).
Despite marginal shifts resulting from economic mobility and higher education (Banerjee et al. 2009), the regulation of female sexuality (including bridal virginity) and labor, both productive and reproductive, especially among the upper castes (Chakravarti 1996;Rege 1998Rege , 2006, remains a key strategy for maintaining caste purity (Dube 2003). A tool to control reproduction and ensure paternity and cultural continuity, marriage is also closely tied to the political economy of communities, especially the landed (Raheja and Gold 1996), with any deviance from caste norms evoking violent reactions, not just against the individuals involved but the entire caste group (Chowdhry 2007;Gorringe 2013). 2 Marriage and sexuality is hence politicized, with numerous strategies, individual and collective, used to weaken women's agency and bargaining power (Palriwala and uberoi 2008). These typically include attempts to reinforce village exogamy (marriage in a distant village, making the new bride a "stranger" in the family of her in-laws), ensure a near total rupture from the natal household (Dube 1997), and, more recently, use of modern technologies of sex selection to ensure the birth of a son (Kaur 2004).
While it is generally assumed that Dalit and landless households present egalitarian gender relations in terms of physical freedom, marital mobility, and control over sexuality, shaped as these are by material poverty and women's engagement with the economy and public sphere, these groups have, however, not remained untouched by the upper caste/classdriven patriarchal framework (Irudayam, Mangubhai, and Lee 2011). Though recognizing the advances made by women in redefining their social agency (Sangari 2002), the state, through its development institutions, has reinforced such patriarchal discourses that give value to virginity, ritualize puberty, and glorify marriage and motherhood (Anandhi 2000).
Despite such dominant representations, diversity in marriage systems has always existed, with early mapping by Karve (1953) pointing to the preference for cross-kin and uncle-niece marriages in south India, as against the North Indian proscription of such relationships. In fact, marriage systems in Tamil Nadu, among Dalits and other castes and religious groups (including Muslims and Christians), have been more flexible since the 1920s, following Periyar's Self-Respect movement for social justice, which focused on improving the status of women through the abolition of prepuberty marriage, emphasis on companionship and equality in marriage, women's education, equal rights to property and knowledge, and support for contraception, birth control, and divorce (Geetha and Rajadurai 1998, 375). 3 Cross-kin marriages remained popular, with prior familiarity of young wives with their in-laws meant to ensure conjugal compatibility, love, and care for members of the joint household and the collective organization of labor and reproduction (Ramamurthy 2014). Postmarital residence patterns remained flexible, and natal family support continued (Gough 1981;Kapadia 1995). This seems on the decline, especially among the upper castes/classes who, in the face of global consumerism and the accumulation of wealth, increasingly emphasize notions of honor and the control over women's bodies, mobility, and sexuality (Srinivasan 2005) to protect their class interests. Joshi, Iyer, and Do (2014) find that cross-kin marriages are increasingly confined to the poor and credit-constrained-those who cannot afford large dowries to make a high-quality, socially distant match.
Cross-kin marriages potentially offer women support by creating pressure to maintain respectable behavior patterns, yet an absence of violence, or a greater mutuality in the conjugal relationship, cannot be assumed (cf. Joshi, Iyer, and Do 2014). This hypothesis gains support from studies such as that of the Jain community in both the united States and India. While nonviolence is a core religious belief among the Jains, patriarchal entitlements tolerate woman abuse, though kept invisible to preserve class status and family privacy (Dasgupta and Jain 2007). This also holds true for self-arranged "love" marriages, which are compromised both by the pressure to fulfill economic and social responsibilities and the lack of parental support (heyer 2014).
The nuances of women's agency in relation to marriage and sexuality are complex, and it is only by unpacking local forms of inequality and friction that one understands both the multiple expressions of resistance and its limits (Ortner 1995). Anandhi (2013), in her study of the Mathammas (women dedicated to the goddess Mariamman) in a northern Tamil Nadu village, describes how the Arunthathiyars, considered the lowest among the Dalits in Tamil Nadu, excluded the better-off and educated Paraiyars, another Dalit group, while inviting the upper castes to their annual Mathamma festival. Arunthathiyar men wanted to glorify this ritual practice as a symbol of status, while many of the Mathammas were struggling against the sexual connotations of this identity and organizing solidarity groups, along with women from other caste groups, to strengthen their entitlements.
Bodily and material practices such as that of the Mathammas, then, reflect the power of public and social discourse in regulating sexuality to maintain both gender and caste boundaries. These practices cannot presuppose a choosing subject; rather, they represent strategies to renegotiate and challenge the upper-caste, male construction of the idea of a Mathamma as "free" (and consequently promiscuous), rather than acknowledge the real constraints they face. What becomes evident is that the intersection of multiple marginal identities, in this case as Arunthathiyars, workers, wives, mothers, and women, serve not just as tools for their subordination, but reclaiming these identities has also helped them reject hierarchical thinking on marriage and sexuality and begin resistance (Collins 1998, 77;Rege 1998).
Changing sexual practices in contexts of economic and social change, such as men's absence, women's engagement with factory work, public discussions on adolescent and alternate sexualities, and widespread violence, have only recently come to be explored (Anandhi 2007;John and Nair 2000;Palriwala and uberoi 2008;Rao 2012Rao , 2014. Building on this recent body of work around marriage and sexuality, I critically examine the difficulties of attributing specific practices or patterns of agency to women in relation to their caste position or type of marriage, given both the internal hierarchies and asymmetries that lead to competition for status and survival within these groups (cf. Adur and Puryakastha 2013) and the wider changes in the local political economy context.

agencY, inTeRSecTionaliTY and Voice: SoMe concepTual conSideRaTionS
"Agency" as the ability to act, decide, and influence, has received considerable attention in development research, particularly in the context of women's voice and empowerment (cf. Kabeer 1999). Sen sees the expansion of substantive freedoms across different domains of life as the very aim of development, involving "both the processes that allow freedom of actions and decisions, and the actual opportunities that people have, given their personal and social circumstances" (1999,17). While such an approach acknowledges both the role of social values and prevailing mores in shaping freedom, and the use of individual freedoms to improve lives and negotiate social relations, it prioritizes the "positive" aspects of action, capability, choice, and independence as central elements of personhood. In this conception, constraints, inabilities, necessities, and contingencies are seen as "deficits" to be overcome. Reader (2007) postulates a more complex view of personhood that recognizes the inseparability and continuity of the active and the passive, and gives dignity to endurance, patience, and compassion as a rational response rather than as signs of victimhood. "Women may endure violence, because it is often not in their power to flee or fight. This is not an easy or self-deluded option, endurance is difficult and courageous" (Reader 2007, 597). Based on her engagement with the women's mosque movement in Cairo, Mahmood (2001) too challenges the conflation of agency with resistance to subordination or the subversion of norms, and points to its application towards continuity and stability. Endurance is not just an expression of a performed self; it may contribute to the very making of that self.
It is, then, important to unpack the agency of Dalit women to better understand the necessities and the specific structures of marginalization they are subject to and "the contingencies that structure their responses to the world" (Reader 2007, 604), rather than present a dichotomous view of constraints and freedoms. Women's subjectivities are not universal and static, but, as noted in the previous section, are "simultaneously class differentiated and subject to the frequent cross-class expansion of patriarchal ideologies" (Sangari 2002, 364). Their agency is often indirect, with particular articulations of consent and resistance developed within their material and cultural contexts. These may derive from a perception of material advantage, ideologies of appropriate behavior, or, sometimes, coercion. Whether expressed as consent or resistance, women's agency rests on multiple, intersecting factors including caste/class status and inequality, economic (in)dependence, social pressures, ideological and normative standards, levels of coercion, and affective relationships (Sangari 2002, 374).
For any meaningful analysis, the coordinates of women's agency need to be established-the ways in which structural constraints interact with the opportunities and possibilities available to women at particular historical moments. This requires a more nuanced understanding of intersectionality, in terms of the multiply determined, simultaneous, and interlinked experiences of power and inequality (Yuval-Davis 2006). In her work on marriage and violence among South Asian women in the united States, Margaret Abraham (1995), for instance, points to the importance of recognizing ethnicity alongside gender in examining women's resistance to violence in transnational contexts, especially as institutions, such as the police and health services, work with particular stereotypes, usually negative, of ethnic minority and immigrant women. Among the Dalits, men are often unable to live up to the myth of the provider-protector, and women's labor is central to household reproduction. In such a context, consent and resistance are nuanced differently and often centered on wider social subordination as well as individual abilities rather than patriarchal controls alone.
Theoretically, as Choo and Ferree (2010, 13) note, intersectionality signifies at least three different dimensions: (1) including the perspectives and knowledge of multiply marginalized people, such as women of color (in this case Dalit women); (2) an analytical shift from the addition of multiple, independent strands of inequality to their interactions; and (3) seeing institutions as overlapping and co-constituting social inequalities (cf. Puryakastha 2010). In this article, I take gender (including individual characteristics such as education, work status, and subject position) and caste (collective status, rights, and opportunities as Dalits) identities as inseparable in both their material and cultural dimensions, but move a step further in asking how these intersections and interactions shape women's agency in the local context. Further, how do the expressions of agency and the possibilities for resistance change over an individual's life course?
"Voice" is often assumed to reflect both consciousness of one's interests and an expression of agency (Ardener 1975). Embedded in the ideas of agency, voice, and freedom are relations of power, the ability to not just articulate or to act/decide, but importantly, to be heard, to be able to influence meaning-making, here in relation to marriage and sexuality, within a politics of intersectionality (Butler 2000, 87;Dube 2003, 223;Racine and Racine 2000). There is hence a strong argument for including the standpoints and self-representations of differently positioned women in order to reveal the complexities and contestations of power (Collins 1998;harding 1987). In the case of the Dalits, Guru (1995) and Rege (1998) argue that Dalit women's need to talk differently is linked to a discourse of dissent against the middle-class women's movement, Dalit men, and the moral economy of the peasant movement, and challenges their political and cultural exclusion. While silence can make invisible and even legitimate oppression and violence, one cannot forget that what women say or do is shaped by the prevalent gender orders, by the task of maintaining, negotiating, and, at times, challenging social relationships (Jackson 2012).

MeThodS
This article is based on empirical evidence from one village in Tiruppur district, Tamil Nadu, here named Pandipatti, 4 collected as part of an ESRC-DFID-funded research project on intra-household allocations in 2009-2010. The larger project included a household survey of 400 rural couples, selected randomly from a cluster of five villages, and a baseline experiment, with these couples playing a standard public good game. Based on the levels of income pooling between spouses reported in the experiments (Munro et al. 2011), 40 of these couples, representing different degrees of separation and pooling, were selected for in-depth interviews. Focus group discussions and key informant interviews helped understand the socioeconomic and normative context in each site. The survey data were analyzed using SPSS; the qualitative interviews were first translated into English and then coded in NVivo.
In-depth interviews, lasting about two hours each, were conducted separately with husbands and wives by four researchers, two male and two female, in or near their homes. Graduate students at the local university were trained in interview techniques and basic ethical procedures during a one-week pilot phase, followed by three weeks of field interviews conducted with me. They worked in pairs so one could focus on the interview-listening and responding-to the woman/man being interviewed, while the other could take notes and record the interview. We all spoke the local language Tamil, so communication was not difficult. The interviews sought to gather information on the subjects' work and marital histories (including violence), process of decision making, and control over household resources. Discussing issues of marriage and violence posed several ethical dilemmas. Apart from maintaining confidentiality and anonymity, it was important for me to carefully interpret the data, given the role discursive representation plays in attributing value to particular lives and experiences and ongoing political contestations, and not just producing knowledge (Fox 2008).
Dalits constitute 19 percent of the population in Tamil Nadu. however, they constituted 35 percent of our survey and 50 percent of the qualitative sample. As already mentioned, Dalits too are not a unified category, with the lowest-ranked Arunthathiyars, locally known as Madharis, constituting the majority group in the research location. In this article I use the term "Madhari," a lived category, rather than the more generic Dalit. The remaining, classified as Other Backward Castes (OBCs), included two distinct groups-the landowning Gounders and the handloom-weaving Devanga Chettiars. The survey found a marked contrast in work and educational status across caste groups. Most Madharis are casual-wage workers, the Gounders work on their own farms or in factories, and the Devanga Chettiars are largely self-employed weavers. Educationally too, 70 percent of the Madharis were not literate, while only 20 percent for the OBCs were not literate, a pattern that holds for Tamil Nadu as a whole (Swaminathan 2002).
The narratives presented here are drawn primarily from the qualitative subsample of 20 Madhari households. They represent a few critical cases that demonstrate how intersecting identities of life cycle position, family status, work, and educational history shape Madhari women's agency in its diverse, everyday forms, seeking to make meaning of their lives, rather than focusing on differences between them and the OBCs. There are three main reasons for this. First, while "arranged marriage" emerged as a unifying discourse among the OBCs, negotiations and contestations were visible among the Madharis, pointing to the persistence of some degree of marital choice and control. Secondly, alcohol consumption was a nearuniversal narrative by women of all castes, yet it was only the Madharis who were more freely able to discuss issues of alcoholism and violence and how they had dealt with them in their lives. Perhaps notions of shame, failure, and disgrace stopped the OBC women from speaking of their experiences (cf. Dasgupta and Jain 2007). Third, as I wanted to highlight nuances of agency in women's lives that were not reducible to their caste affiliation, it made conceptual sense to begin with one often homogenized group and uncover the variation within it.
conTexT Of around 400 households in Pandipatti, more than 60 percent are Madharis, mainly laboring in agriculture and construction, and the remaining consisting of the landowning Gounders. Over the last 50 years, while agriculture has declined, the expansion of the Tiruppur industrial cluster as a global center for hosiery and garments has expanded opportunities, especially for men, in the non-farm sector. Alongside improvements in infrastructure and communication, state welfare provision in the form of cheap food grains and guaranteed employment (harriss, Jeyaranjan, and Nagaraj 2010) has transformed labor relations, with patronage-based, tied labor sharply declining (heyer 2010). Despite this, continued economic marginality and dependence of the Madharis on the Gounders for employment (Chari 2004) has necessitated Dalit mobilization, if only to counter the positions taken by intermediate caste coalitions of the Gounders, Vanniyars, and Thevars (Gorringe 2005) seeking to protect their economic interests both through control over the labor markets and opposition to cross-caste marriages. As Manohar Gounder noted, "There is a strong association of power loom owners locally. We have opposed the demand for higher wages given the already small margins in this sector." These changes are further gendered. While jobs are available, the terms of employment are not easy for women: They involve an extension of the working day and are hard to combine with reproductive work. Women either withdraw temporarily from the labor force, especially when children are young, or engage in low-paid, piece-rated, home-based work (Rao 2014). The uncertainty of finding appropriate work and income enhances dependence on their men. At the same time, the social expectation of women's responsibility for procuring and processing food into three meals a day, and managing the household, persists, even within Dalit groups. The state Public Distribution System (PDS), providing cheap rice, has facilitated access to basic cereals, yet with severe inflation in the prices of all other commodities-pulses, oil, vegetables-and unpredictable male contributions-performing this task can pose a challenge.
Forty-year-old Kala noted: My husband worked for Durai Gounder, in his field, poultry farm, whatever work he gave him, for the past 20 years, but refuses to go to work anymore. I have been unwell so unable to work regularly. Life is difficult, as everything is so expensive.
A Gounder landlord, describing the difficulties they faced in securing labor, blamed the habits and attitudes of the "coolies" for their poverty: Earlier we contributed money and gifts at the time of death or marriage in the family of the laborer; this is no longer the case. They take advances as a pre-condition to work and use it to buy cell phones, vehicles, and alcohol-the money disappears in a flash. They want to relax, not work hard, so it is a big risk for us employers.
While work-related links are maintained between the Gounders and Madharis, marriage and cultural practices remain separate and distinct (cf.

Banerjee et al 2009). Duraisami Gounder noted:
Education does change social interaction. Young people meet and get to know each other at school and at work. Yet this does not destroy the basic character of our community. Love marriages are not stable-parents may accept it, but are not happy, and over the years this lack of affection takes its toll.
In a neoliberal context reflecting the coexistence of economic opportunity with insecurity and uncertainty, marriage and sexuality become doubly important in redefining relations of production and reproduction, and negotiating the social constructions of both identity and social position. Arranged marriages have endured among the Gounders, though cross-kin marriages have reduced; the survey data found 35 percent of the OBCs and 55 percent of Madharis in cross-kin marriages. These differing rates are interesting as they highlight generational shifts in the two groups, with the Gounders moving closer to the North Indian pattern of "natal rupture" with improving levels of education and prospects for employment. The hegemony of age and gender seems to be getting stronger, rather than loosening, with "love" or "self-arranged" marriages confined to the Madharis (Rao 2014).
The survey did not ask questions about violence, yet 60 percent of the Madhari women reported high male alcohol consumption and violencephysical, emotional, and structural-in the qualitative interviews. The type or nature of marriage-arranged or love, cross-kin or non-kinhardly seemed to matter; interestingly, of those reporting violence, a majority were married to kin (9 out of 12). Kala further wondered: A year ago my son ran away with a girl, the previous year my daughter had a love marriage; she did not even tell us whom she was marrying. My husband has changed as a result-he drinks a lot, hardly talks to me, and doesn't help with anything. Earlier he would fetch water; we would go together to the market. he blames me for not bringing up the children well.
The explanations for violence are several. In Kerala, Panda and Agarwal (2005) suggest that violence against women is driven by stress linked to uncertain male work and incomes and the inability to adequately perform their role as "provider." This doesn't seem to be the case in the local context, where work is available, even if not optimal. It could, however, be an attempt to break out of the labeling as "wastrels" and "lazy" (McDowell 2007), as seen from the perspective of the norms and morals of the powerful. They then seek to attribute value to their own identities through investing in particular consumer goods, but also gendered bodily and sexual practices (Skeggs 2009). In the case of Kala's husband and several others in the village, their own inability to control their children, or their work, leads to blaming the woman for failing to perform her (homemaking) role in line with middle-class norms-taking care of the children, serving food on time, obeying the husband-and becomes a reason for both alcohol consumption and violence. State ownership and promotion of liquor sales in Tamil Nadu has further intensified alcoholism. 5 While not an issue of morals or innate character as made out by the upper castes, recent evidence suggests a far stronger habit of regular drinking, followed by violence, now embedded in Dalit lives and culture (Kishor and Gupta 2009). I discuss next women's voice and agency in relation to two issues repeatedly raised by them -the choice of marriage partner, and marital violence. The differences in their choices and strategies highlight the importance of an intersectional perspective in better understanding the processes involved in renegotiating conjugal and wider gender relations.

inSighTS fRoM The field choice of Marriage partner
Thulasi, 35, was married at the age of 13 to her father's sister's son. Of her three children, the elder son works at a power loom, the daughter in a mill, and only the youngest goes to school. her sister's children stay with her. She narrated: We were four daughters. I didn't go to school, my parents were poor. When I was 10, my sister and I started going to the company to collect scrap. I worked for two years; then we were all married to relatives residing in the village. One of my elder sisters quarrelled with her husband and committed suicide. That man was a drunkard, cast doubts on her character, always shouted and beat my sister. My younger sister met the same fate. her husband was seeing another woman and I recently heard that he has married her. he too beat her badly, till finally she took many sleeping tablets and died. No one knew they were of such character, but even if we sisters had refused, my parents would not have listened to us.
Thulasi had little choice in the selection of her marriage partner. As a young girl of 12, she was unable to negotiate with her parents the timing of her marriage or indeed the person she would marry, and ended up conforming to the exigencies posed by poverty and social norms. Despite working and earning, and being married to a relative, she points to the real constraints to agency posed by the normalization of patriarchal entitlements, including male violence and sexual infidelity. Of course, she was very young then, her relationships shifting over the years, as I demonstrate in the next sections.
Twenty-nine-year-old Jaya was also married to her cousin-her mother's brother's son-at 16. unlike Thulasi, her parents had a small, twoacre plot of land, and her father had completed secondary education and wanted to fulfil all her desires as a child: I left school after grade five because I was weak in maths and could not tolerate the teacher's scolding in front of all my friends. It was the tradition in my place that teenage girls would not be allowed out of home, so I stayed back and helped with household chores. My uncle suddenly died, so my mother, who always wished me to marry his son, arranged for us to meet at the temple. I liked him a lot, even though we didn't talk to each other. My mother was keen, so I gave my consent.
In Jaya's case, there is a clear emphasis on parental control of her sexuality following puberty, and marrying her to a man of their choice once she dropped out of school. her emphasis on "tradition" can be understood in the context of her family's aspirations for class mobility (cf. Kapadia 1995), but also in a general context of violence where Dalit girls are likely to be sexually harassed/assaulted postpuberty (Bama 2000). Yet, at least symbolically her parents gave her an opportunity to meet the man they expected her to marry and formally give her consent. Over the years, she has attained many goals through conformity and silent negotiation. For her, deference and silence are conscious strategies of self-representation, routine responses to authority, and even assertion in particular contexts (Jackson 2012).
Despite the emphasis on marriages arranged by their parents, selfarranged marriages are not uncommon. Rathi, 50, nonliterate, with four grown children, three of them married, talked about her marriage, the difficulties she encountered, and her survival strategies: he would come to our village once in three months to cut trees. I was a daily wager on a farm, having started working soon after my father's death. I must have been 10 or 12 then. Apparently he saw and liked me. Then one day he proposed to me: "I like you, will you marry me?" I replied, "Now you say this, but you are a man, you can change your words. how can I be sure?" he assured me of loyalty till our death. I agreed. We got married in a temple. Nobody else was there. he bought me a sari; that is all. I had a nose stud given by my mother a long time ago, but later when we needed money, I had to pledge this. My elder son too had a love marriage. They worked in the power loom and met each other daily. They went to a temple, got married, and then came home. If it was an arranged marriage, we would have spent at least Rs 10000, so a love marriage is good, it saved us money. Rathi's was a self-arranged marriage and so was her son's. There was no problem with acceptance on either side, unlike in the cases of Kala or what Duraisami indicated for the Gounders. In fact, she jokingly said that "love" marriages are preferable because they save money: There are no gifts to be given or guests to be fed. her narrative sharply critiques the wasteful expenditures of the upper castes in conducting marriages, and hints at the link between their negative moral valuation of love marriages as unstable to their inability to control the actions of the young couple, challenging hierarchies of age and patriarchal control. Even though young, she was already a farm worker, with independent wages, and could exercise her agency in choosing her marital partner with little parental resistance.
The critique of upper-caste norms on arranged marriages is further exemplified by Amritha, 20, who, having studied up to the 10th grade, is one of the more educated Madhari women. having earlier worked in tea plantations, she proudly mentioned completing three months of computer lessons and securing the government's "marriage loan scheme for the educated." Aware of the violence and insecurity around her, she wanted to be sure of the character of her husband and test his loyalty before actually marrying. She used the mobile phone for this purpose, a technology unavailable to the older Rathi: Before marriage, I once called him and, pretending to be a friend of mine, said that I liked him very much. he got angry and started scolding me, thinking I was the other girl. I cut the call and laughed. My faith in him increased.
Subtle differences exist in the nature of agency and voice exercised by these women. While Thulasi had little choice in the selection of her husband, this is not the case for the others. For Jaya, while the choice was her mother's, she was asked for her consent. Rathi met her husband while working in the fields; Amritha too self-selected, though introduced by a relative. Their linguistic explicitness varies, shaped not only by their context, but equally by their subjective positionality within a particular timespace matrix (Gal 1991), constituted by their level of education (Amritha), ability to work (Rathi), family status pressures (Jaya), and degree of poverty (Thulasi), among other factors. So do their strategies for dealing with life's uncertainties and building conjugal solidarity, which became apparent while discussing how they confronted violence in their lives.

experiencing Violence and Resisting it
Despite suggestions that arranged marriages (especially cross-kin) are more stable, given both kin support and pressure, making violence less likely, the narratives suggest no clear relationship between marital choice and violence. Violence is pervasive, yet women did resist, using a host of strategies-from personal ones like silence, avoidance, talking back, and, in extreme cases, contemplating suicide, to drawing on informal sources of help from the family, neighbors, and in-laws, to, in the last instance, formal support systems such as the police (Abraham 2000). These once again are shaped by their financial (in)dependence, social pressures that they are subjected to, and personal initiative. Thulasi, at one point in her life, seeing no alternative to the daily violence she experienced, contemplated suicide: My mother-in-law started finding fault with me and quarrelling incessantly. My husband listened only to her and beat me for not obeying her wishes. I was upset and wanted to kill myself, so drank "sanipowder mixture." he scolded me but then rushed me to hospital. There everyone spoke to my father-in-law and said, "At least you should be a responsible man." he supported me and told my husband, "She is very young. Let her not suffer. Your mother will always speak like this. Better find a separate way to run your family." We moved out with just one vessel and some rice. My parents helped us out with other things.
Thulasi has seen two of her sisters commit suicide, unable to tolerate the constant drunkenness, violence, suspicion, and waywardness of their husbands. While promiscuity remains an accepted part of masculinity, female sexuality is to be controlled, especially postmarriage (Geetha 2000). Thulasi too tried to commit suicide on one occasion, an extreme strategy of resistance, but was saved. With support from both her fatherin-law and parents after this incident, and given her own ability to work and earn, she was able to overcome her subordinate role, especially vis-àvis her mother-in-law. She set up a separate household with her husband and gradually developed a better conjugal relationship. In speaking of her siblings, parents, in-laws, and children, her narrative points not just to the complexity of gender relations-between sexes and across generationsbut also the availability of avenues of support at a particular moment in time shaping her strategies of resistance.
unlike Amritha (one of the few women who had not experienced violence so far), Rathi hadn't fully managed to explore the character of her husband prior to marriage. She didn't realize, for instance, that he was a regular drinker until after their marriage: he drinks every day, but only after work. When he returns home, I scold him, "Are you not ashamed? Don't come near me and ask money for drinking." Then he turns affectionate and says, "Dear, don't get angry. I don't want money now. I won't drink hereafter." I threatened to leave him and go to my sister's place if he continued like this. One day he came home drunk. I kept quiet and didn't answer. he thought that I had really left and started crying. Now we are getting old. how can I leave him? he does care about me. I tried to change this habit, but it was already too late, so now I just live with it. he is good at heart. While there might have been instances of violence earlier in their marriage, Rathi wasn't explicit about it. As they have aged, had children and now grandchildren, he has possibly calmed down. While enduring and reconciling to their life together, she has not been a silent victim, but has always challenged his behavior-talking back, retaining financial control, and threatening to leave him. Despite having exercised full marital choice and being an earning member of the household literally from the first day of their marriage, and continuing to manage and control money and expenses in the household, Rathi points not only to the ups and downs in their marriage, as in most marriages, but also to a persistent struggle for gaining dignity, reciprocity, and respect based on relations of closeness, proximity, companionship, and intimacy, rather than those characterized by "violent erasure, substitutability and appropriation" (Butler 1993, 46).
These are common narratives. What they make clear is that gender relations are fluid; women's agency and voice are not just shaped by the interplay of structures that marginalize, and individual attributes and experiences, but also change through their life course. Today, Thulasi is in a much stronger position than at the time of her marriage. Pointing, like Rathi, to the importance of endurance as a core element of her agency, she closed the interview with a statement on status: My husband too started drinking with his friends, but he doesn't beat me. he simply goes to bed. When I am working, he fetches the water, prepares coffee for the children, and sometimes cooks the food. My children are always there to support me. This house is in my name, following a government order that free land for house construction should only be in the woman's name. This gives me strength. Yet a woman can only get respect if she stays with her husband. If she stays at her mother's house, she won't have respect. People will remark, "See how she is going around after leaving her husband." So even if her husband beats her, she should not leave him.
Thulasi has systematically invested in resources, both material (the house) and human (her children), following her attempted suicide. These have given her the strength and confidence to resist, but equally to demand reciprocity from her husband-a huge shift from the past.
While reformulating the meanings attributed to the cultural expectations of marriage and sexuality in terms of partnership, endurance, and reconciliation, she has never considered separation an option. her last statement on respect demonstrates awareness of accepted caste ideologies and signs of respectability, in terms of middle-caste/class norms of lifelong service to the husband, and an attempt to claim this status for herself, as an expression of her agency (cf. Butler 1993). At the same time, she contests upper/ middle-caste notions of conjugality, honor, and purity that stigmatize and stereotype Dalit women. For instance, her husband's help with household chores when he doesn't get work is not something that middle/upper-caste men would do (Rege 2006). her strategies reflect the workings of intersectionality, her appreciation of the social constraints that women such as she confront-Dalit, working-class, middle-aged, with no education and few alternate options-yet shifting with her subject-position from a young woman to a mother. They also point to its deep connections with the external environment (such as distribution of land to Dalit women).

concluSionS
Writings on Dalit women often present a contradiction-they point to the pervasive nature of violence and the sense of perpetual threat and its presence in everyday life, as emerging from the hierarchies and inequalities generated by the interlocking structures of caste, class, and gender. At the same time, they suggest that advantages, in terms of employment and cross-kin and self-arranged marriages, make Dalit women less passive and mute. I have tried to address this debate through examining Dalit women's voice and agency from an intersectional perspective, going beyond structural characteristics of caste, class, or gender to refine an understanding of women's agency and voice within a unifying category, namely, the "Dalits." What emerges is that, in their everyday struggle to survive and lead a life of dignity, agency takes different forms, including compassion for a violent husband or producing middle-class honor through respectable femininity. If agency is understood in this broader way, then the evidence of enhanced patriarchal controls being read off Dalit women's "acceptance" of violence (Irudayam, Mangubhai, and Lee 2011), especially in the wake of economic liberalization, can be reinterpreted. While there is emerging evidence on the marketization of the economy and dowry inflation in Tamil Nadu (Jehan 2011;Srinivasan 2005), further study would be needed to understand the implications of these changes for the renegotiation of gender relations and, indeed, women's agency.
Marriage to a "stranger" and "natal rupture" have been highlighted as two key elements of the North Indian kinship systems that subordinate women, especially young wives (cf. Raheja and Gold 1996). Even though the women in the research context potentially knew their husbands prior to marriage (cross-kin) and relationships with their parents and other natal kin endured postmarriage, this did not necessarily ensure voice in key life decisions or freedom from violence. Rather the presence/absence of children, economic opportunities, and availability of support systems, alongside their own confidence in their abilities and skills, shaped their responses. Embedded in specific social and cultural contexts, and engaged centrally in building and maintaining relationships, women's agency ranges from resistances to expected norms to an elaborate subjection to them, especially in the organization of space and time postpuberty. These women have also resisted violence in different ways, from attempting suicide to pleading, cajoling, threatening, and moving out.
The narratives in this article point to the inseparability of agency and the apparent lack of it, of action and patience, as a continuum of strategies in which women invest to challenge hierarchy and strengthen their own bargaining position in the household while also ensuring bodily integrity (cf. Anandhi 2007). They draw out the ambiguities in their personal lives, the small, everyday actions undertaken to expand the spaces available to them, demonstrating that they have not necessarily assimilated their devaluation and submission to male kinsmen or, indeed, upper-caste employers in their conceptions of selfhood. They also point to the fluidity of gender relations within Dalit households, as among other groups, the continuous negotiating and building of new alliances that shape and reshape the social construction of roles and responsibilities. Their voices destabilize received truths and locate debate in the complexities and contradictions of historical life (Mohanty 2003, 244). noTeS 1. Dalit, or "oppressed," is the term used by those living at the bottom of the caste hierarchy, termed by the Indian Constitution as Scheduled Castes, to represent themselves (Omvedt 1995).
2. Based on recent incidents of caste violence in Tamil Nadu, where entire Dalit settlements were laid waste by the Other Backward Castes (OBCs), Gorringe (2013) notes that while the trigger was an intercaste marriage, the violence was also fueled by improving educational status among the Dalits and their refusal to act as submissive agricultural laborers. While women's bodies served