People’s Relational Agency in Confronting Exclusion in Rural South India

.................................................................................................................................................. ii Acknowledgements ................................................................................................................................ iii


Introduction
Individual and collective struggles against social exclusion in India are inseparable from longstanding attempts to dismantle caste (Ambedkar 1936). Such struggles have been waged for millennia against casteism that interacts with gender, ethnic and religious exclusions. Together, these exclusions are recognised as exacerbating rural poverty and inequality Coleman et al. 2018 andMosse 2018).
In postcolonial India, subjugated people have resisted the denial of access to benefits from development interventions, to resources such as land and to spaces of political power (Bardhan 1985;Agarwal 1994;Tiwari 2014 andElias, Grosse andCampbell 2020). They have thus challenged being 'kept out' or 'left out' of economic benefits and political participation (Sen 2000: 29), while confronting 'unfavourable inclusion' in spaces of development: people may be included in projects and schemes designed by the powerful, only to be controlled and exploited, marginalised and oppressed (Fischer 2011).
Social exclusion studies show how the dominant in society attempt 'opportunity hoarding' in markets and networks (Agarwal and Levien 2019;Acharya 2018;Bosher, Penning-Rowsell and Tapsell 2007); how modern technologies are deployed to undermine the knowledge and skills of rural people (Birkenholtz 2008;Parthasarathy 2002); and how attempts are made to control and exploit nature to harm people belonging to the 'oppressed' castes and Adivasi ethnicities (Bhattacharjee 2014;Williams and Mawdsley 2006). Such forms of exclusion are often produced through the exercise of social power that tries to control the flow of technologies and resources extracted from nature. However, control is never fully realised in practice (Stirling 2019;Arora et al. 2020), because 'excluded' people exercise agency in struggling for dignity and livelihoods (e.g. Agarwal and Levien 2019;Tiwari 2014). At the same time, technologies and nature resist being moulded and deployed to achieve the ends that exclusionary power requires of them (Latour 2000). This sociomateriality means that power is enacted in highly uncertain ways. However, sociomateriality and its uncertainties are very rarely conceptualised in the literature on social exclusion (Mosse 2007;Kabeer 2006).
While the agency of 'excluded' people is recognised in the literature on social exclusion in India and other parts of the world, studies neglect how this agency is constituted by people's diverse relations with nature and technologies. This neglect means technologies and nature are approached typically in one way: as objects that are amenable to control by social power . Also overlooked in this process, are alternative relations of care -with things such as trees and soils, tools and techniques, as well as with people in social movements and families (Puig de la Bellacasa 2017). Such relations of care are central in constituting 'excluded' people's agency to struggle for freedom from exclusion and marginalisation.
We attempt to address this neglect by relying on the life histories of three people who are farmworkers and smallholders in two villages of Tamil Nadu. We propose a new conceptual approach to agency (of the 'excluded') in which relations of care intersect with contrasting relations of control. While the latter relations are characterised by hierarchy and extraction of resources, the former involve qualities such as nurturing attentiveness and mutualism . While relations of control play a central role in constituting exclusionary forms of agency, caring relations lie at the heart 2 of agency to struggle for freedom by the 'excluded' (cf. Sen 1999). Following our interlocutors, who fluently interweave the social with the material in their life history narratives, all relations in our approach are sociomaterial. The two ways of relating as considered here -control and care -span divides between nature and culture, humans and nonhumans (Latour 1993).
We begin with a review of understandings of power and agency in studies on social exclusion in rural India. Building on this, we develop and illustrate our conceptual approach with life histories. In conclusion, we reflect on how our approach can yield insights to help challenge power that constrains the agency of 'excluded' people to realise livelihoods and freedom from oppression. This can in turn help rethink policies and politics to address these constraints on agency and apprehend social exclusion to achieve sustainable development in rural India and further afield.

Social exclusion in rural India: power and agency
Power is central to understanding social exclusion. Studies approach social power through hierarchical structures and relations cutting across caste, gender, religion and ethnicity (e.g. Mosse 2018;Sarap 2017;Jha 2015;Kannan 2014;de Haan 2011;Mosse 2007;Kabeer 2006). A distinction is generally made between 'power over' others to constrain their agency and 'power to' as the agency of asserting one's own will and set agendas. These two ways of approaching power -as an agency that is repressive or productive -are not restricted to social exclusion literature (see e.g. Guérin, Kumar and Agier 2013 for debates among scholars of 'empowerment').
If agency is approached as unfolding relationally, 'power to' act may be aimed at achieving 'power over' others (Foucault 2001). Building on this, for the purposes of the present paper, power is conceptualised simultaneously as a set of actions and as a way of relating characterised by hierarchy and control. Therefore, in our conceptual approach, developed in the next section, we use the term control to describe the space of overlap between 'power to' and 'power over'. Power thus points to actions geared towards controlling others' possible or present actions (Foucault 2001: 341). The 'others' in this definition can include nonhuman forces in nature (Latour 2000). The possibility or presence of others' actions is thus a necessary condition for control to be attempted and for a power relation to exist. Power acts on the (possible) actions of other humans and nonhumans that can never be fully controlled (Latour 2000). Clearly, the exercise of power becomes unnecessary if complete control is actually achieved.
In rural India, driven by hierarchical relations of gender, caste, religion and ethnicity, power is geared towards controlling access to landownership, ecological commons, temples and educational institutions (Agarwal 1994;Mines 2005;Kabeer 2006;Still 2009). The resulting exclusion from access shapes 'economic' relations and outcomes. For example, caste-based and gender-based lack of access to landownership can lead to exploitative labouring relations in agriculture (Bhaduri 1973;Harriss 1982;Sharma et al. 2020). Often, caste, gender, religious and ethnic hierarchies can intersect to enable the powerful to control markets and state interventions (Kapadia 1997;Bosher et al. 2007;Buckingham 2011;Deshpande 2011;Shah et al. 2018). For example, the combination of casteist and gendered oppression shapes Dalit women's lack of access to assets, education and decision making autonomy over crucial issues such as healthcare (Deshpande 2011).
Across markets for jobs, higher education, credit, healthcare, energy, food and groundwater (Acharya 2018;Naz 2015;Bosher et al. 2007;Chavan 2007), dominant castes 'hoard' opportunities by aiming to control access to information and resources. For example, opportunity hoarding by the dominant castes across markets for higher education and credit can push Dalit people onto 'lower' rungs of the job market offering insecure and unremunerative employment (Jodhka 2016). This lack of access to secure employment also points to Dalit people's 'unfavourable inclusion' in the job market (cf. Sen 2000).
Similarly, attempted control of global value chains, local contracts and commodity markets by powerful actors (i.e. large firms) leads to the 'adverse incorporation' of workers and smallholders (Jakobsen and Nielsen 2020;Mitra and Rao 2019;Vicol 2017;Carswell and de Neve 2013). Such forms of exclusion driven from outside villages can exacerbate caste-and gender-based oppression and marginalisation inside villages.
The power to exclude can also impair public policy and civil society's efforts to promote participatory development and decentralised democracy. For example, the Panchayati Raj reform of the 73rd Constitutional Amendment led to the opening up of leadership positions in local government institutions for women and Dalit people (Bonu et al. 2011). In practice, however, Dalit people are often barred from using their elected positions to advance the wellbeing of their communities due to control, coercion and threats of violence by village-level dominant castes (Inbanathan and Sivanna 2010). Similarly, other decentralisation reforms -such as those in Kerala -aimed at active citizenship at the grassroots are argued to be constrained by the social marginalisation of Adivasis, Dalits and Muslims (Williams and Thampi 2013).
Places reserved for women on joint forest management executive committees in Karnataka and Madhya Pradesh are observed as failing to tackle the social exclusion of women (Elias et al. 2020). In joint forest management meetings, women are silenced by men deploying claims of masculine superiority. Women's participation is also impeded by their caste and Adivasi identities, as well as by age and economic status (Elias et al. 2020). Different intersecting relations of control associated with hierarchies of gender, caste and ethnicity can thus shape diverse social exclusion experiences within the same community, or even within a single household. The workings of power to exclude are far from homogeneous for members of the same community or household.
Power to exclude can be tied to technologies and ecologies (Parthasarathy 2002;Naz 2015). For example, the modern techno-scientific interventions of the Green Revolution have been observed to reduce farmers' cultivation options, to pollute agrarian ecologies and to marginalise farmers' knowledge (Parthasarathy 2002;Kumar 2016;Shiva 1988). Similarly, environmental harms, such as depletion of groundwater and land dispossessions, are viewed as disproportionately affecting people marginalised by class, caste and ethnicity (Bhattacharjee 2014;Naz 2015;Williams and Mawdsley 2006).
Conceptualisations of the role of technology and ecologies in constituting power, however, remain neglected in the social exclusion literature. A fuller consideration may direct attention towards how the effects of technologies and ecologies do not play out in the ways expected or intended by power (Latour 2000). Thus, technologies and ecologies may resist attempts by power to control their effects. For instance, as widely reported in India's agrarian crisis, modern technologies deployed and resources extracted from nature can stop working as intended by power (e.g. Arora 2012; Vasavi 2012). 'Excluded' people can also attempt to adapt technologies and resources for their benefit (e.g. Shah 2003; Patnaik, Ruivenkamp and Jongerden, 2017).

Agency of the 'excluded'
It is crucial not to approach 'people as mere objects of exclusionary processes' (Skoda and Nielsen 2013: 6). People confront exclusion by exercising individual and collective agency (e.g. Harriss-White 2005;Carswell and de Neve 2013;Tiwari 2014). Collective agency includes organised protests and social movements for justice and equality (Agarwal and Levien 2019;Williams and Mawdsley 2006). Individual agency of the 'excluded' can also take the form of building a livelihood, which unfolds in spaces where power is geared towards the control of individuals' actions and access to resources.
'Excluded' and marginalised people's individual and collective agency can be oriented towards 'seeing the state' (Corbridge et al. 2005). In the context of government reforms and laws such as the Right to Information Act, Corbridge et al. (2005) highlight how Adivasi smallholders and Dalit labourers engage 'with the state as citizens, or as members of populations with legally defined rights or politically inspired expectations' (Corbridge et al. 2005: 18). These forms of agency are mapped as contingent on a) mobilisation by (left-wing) political parties, social movements and non-governmental organisations; b) previous participation in direct encounters with the state; and c) the circulation of information about welfare provision by the state across boundaries of class, gender, ethnicity and locality (Kruks-Wisner 2018). Harriss-White (2005) observes 'destitute' people as being unable to collectively stake claims for welfare or other state-driven development benefits. Yet they perform agency individually, by altering gendered divisions of labour within households (e.g. men collect water and cook food). Expelled from many spheres of social life, 'destitute' people seek to form supportive networks among themselves.
The role of social networks in constituting agency is also highlighted by Bosher et al. (2007). Dalits and 'backward castes' tend to form village-level networks in cyclone-prone and flood-prone areas of Andhra Pradesh. In contrast, dominating castes form trans-local networks with political actors wielding influence in the state government, which are more effective in securing access to facilities such as financial assistance, healthcare, sanitation, community shelters and cyclone-resistant housing (Bosher et al. 2007).
Breman (2007) foregrounds labourers' agency to resist by fleeing from bondage. Unfortunately, such resistance rarely leads to political and economic transformation in favour of the marginalised. Even more collective forms of resistance, such as Dalit protests against the dispossession of land for Special Economic Zones, may not yield substantial transformation (Agarwal and Levien 2019). In studying protests initiated by Dalit women in one village, Agarwal and Levien observed how other inhabitants of the village joined the protests, while the villagers linked up with wider social movements against land grabs. However, power gradients were not significantly transformed: Dalit people received poor compensation for dispossessed lands and were denied significant participation in the labour market around the Special Economic Zone (Agarwal and Levien 2019).
Building on Sen's (2000) capability approach, Tiwari (2014) distinguishes between individual, collective and relational agency. Individual agency is a person's ability to act. It translates into effective capability through embeddedness in societal structures of opportunity and resources. For example, an individual's agency, as the ability to educate herself, becomes effective capability if she can access schools, teachers and learning resources. According to Tiwari (2014), such access points to relational agency. Collective agency points to the ability of social groups to act. It is enabled through the alignment of individual and collective interests. By mobilising to develop and access resource and opportunity structures, collective agency helps strengthen individual capabilities through relational agency (Tiwari 2014). Building on this, our own conceptualisation of agency, presented below, is geared towards appreciating how all individual and collective agency is relationally constituted. These relations are sociomaterial: they link actors with other humans and nonhumans.
The literature on social exclusion in rural India recognises, to some extent, the agency of 'excluded' people to resist and to build livelihoods. However, these forms of agency are viewed largely as constituted by people's supportive relations with each other, and not with ecologies and technologies. These nonhumans are neglected in the literature on social exclusion in rural India, and the same is generally true for studies on rural exclusion in other parts of the Global South (for exceptions from 'indigenous' South America, see Gómez-Barris 2017 and Guzmán-Gallegos 2019).

Assembling agency beyond social exclusion
There is a vast literature beyond social exclusion that explores the technological and ecological constitution of rural agency in the Global South. Two strands are salient. The first looks at how materials such as solar panels and water meters (re)shape the agency of technology users (Akrich 1992;Fatimah, Raven and Arora 2015;Brown and Pena 2016). To users, the meaning of such technologies is uncertain rather than self-evident. Thus, Brown and Pena (2016) call for the need to educate people on how to use new technologies in desired ways. In contrast, Lau et al. (2020) and Wilshusen (2019) observe how 'education' and 'training' can help enrol users alongside technologies as co-producers of dominant frameworks in rural development, such as ecosystem services and natural capital accounting (Lau et al. 2020;Wilshusen 2019). Thus, technologies are deployed and users' agency is trained for attempted control to co-produce dominant frameworks (Beck 2016;Birkenholtz 2009).
Within the second strand of literature, some scholars, following Latour (1993;, disrupt the modernist separation between nature and culture and illustrate how agency is performed, not just by humans embedded in their cultural contexts, but also by nonhumans in 'nature' (Watts and Scale 2015;Fatimah and Arora 2016). Agency is thus considered as performed by interrelated humans and nonhumans distributed in heterogeneous assemblages (Donovan 2014). Such a view implies that, as actors, nonhumans -like humans -attempt to resist control and therefore any course of action is underdetermined and unpredictable, replete with uncertainties (Arora 2019). One implication of this recognition of nonhuman agency and associated uncertainties is that the heterogeneous assemblages of impoverished people might be approached as networks of resilience, carrying within them the possibilities of sustainable development that is 'locally-sensible and just' (Gareau 2012(Gareau : 1614. Here, emphasis is placed on people's relations of care developed on the ground (Hanrahan 2015), in contexts shaped by local and global gradients of power.
Building on these insights on assemblages, uncertainty and relations of care that constitute people's agency, as well as on the life histories narrated below, our conceptualisation of sociomaterial agency offers the following novel contribution. We propose to focus on care and control as two contrasting, yet often intersecting, ways of relating that constitute agency. While relations of control play a central role in constituting multiple forms of power enacted as exclusionary agency, relations of care are crucial in constituting the agency to struggle for freedom through resistance and the building of livelihoods by the 'excluded'.

Conceptualising relational agency
Taking inspiration from process philosophy, we recognise that 'all things flow' (Whitehead 1978: 234). Things change as they move through time and space, while relating to other nonhumans and humans. Far from being passive objects that lack agency, things are vibrant and alive (Bennett 2010). They decay, slow down, adapt, gather force or create processes. This means agency is distributed in assemblages of interrelating nonhumans and humans (Latour 2005). In these assemblages, at the same time as nonhuman materials contribute to action, people relate to each other mediated through norms and values, including patriarchal control, familial care and casteist endogamy (Kapadia 1995;Guha 2013).
In agriculture, for example, interrelating materials include inanimate things such as clothing and tools. Materials also include (farm) animals, soils and plants. With these materials, farmers and workers work with seasons, topographies and technologies like tractors and canals (Arora and Glover 2017). These technologies are far from neutral or apolitical. Power in the form of social hierarchies of gender and caste can be partially built into designs of technological materials such as irrigation infrastructures and heavy machinery (Shah 2003;Cockburn 1985). The social and the material are thus interwoven in constituting assemblages that act, in which all relations are sociomaterial, cutting across modern categorical divides of nature and culture (Latour 1993). As conceptualised here, agency is sociomaterially relational, even when it is attributed to an individual human or organisation (Law and Callon 1997;Arora et al. 2013).
Performed within sociomaterial assemblages, agency is defined as action that makes a difference in/to the flows of life (Latour 2005). Thus, agency includes mediating, transforming, knowing and governing the intercoupled flows of ideas, symbols, values, interests and materials. Our conceptualisation of 6 agency encompasses any action that brings about the smallest change in flows of life. This includes actions to produce knowledge that aim to represent reality, ostensibly without intervening in it (Hacking 1983). Our relational process approach, as distinct from a structural one (Giddens 1984), thus helps account for agency's dynamism, diversity and distributedness. Yet we do not overlook structures expressed as casteist and gendered norms -in the form of prescriptions and proscriptions -and values that underpin care and control . Our approach to structures is fluid, building on Giddens' (1984) structuration approach. Here, agency is constituted through relations with structures such as gendered and casteist norms, but in diverse and uncertain ways, and structures are conformed and transformed in equally uncertain ways through situated agency.

Situating agency and uncertainty
In our approach, agency is situated (Haraway 1991) in two interconnected ways. First, action is constituted and performed within relations of specific sociomaterial assemblages (Latour 2005). As they are performed, actions can (re)orient relations in specific assemblages (Callon 2007), potentially bringing about transformation in structures situated in the assemblages. This is discussed in more detail later.
Second, situatedness points to the importance of admitting uncertainties (Arora 2019). Unlike risk that is quantifiable based on probabilities attached to the occurrence of predictable events, uncertainty points to incomplete knowledge because possible events cannot be fully characterised (Wynne 1992;Callon, Lascoumes and Barthe 2009). While the importance of uncertainty is recognised for weather patterns and climate disruptions (e.g. FAO et al. 2018), it is often marginalised or even concealed in other arenas of modernist knowledge production (Arora 2019; Stirling 2019). For example, modern technologies may be promoted to work without problems, in accordance with design manuals, while suppressing or marginalising associated uncertainties that can later manifest as social, technical or ecological problems (Arora 2019; Stirling 2019). In this paper, we treat uncertainties as central by approaching the assembling of agency as always underdetermined. Accounts of sociomaterial agency produced using our approach are always incomplete. They are therefore characterised by uncertainty.

Ways of relating
We distinguish between two contrasting ways of relating: control, and care. Control, defined as constricting or directing others' actions and possibilities, is often conditioned through structures of gender and caste in rural India. Relations of control are thus central to constituting 'exclusionary' forms of agency. The latter forms of agency may be viewed as power (see the review above of the social exclusion literature).
In contrast, relations of care based in the nurturing of others (and of oneself) play a central role in enabling the agency of resistance and livelihood-building. The latter forms of agency can be geared towards struggles for freedom by the 'excluded', from power that governs their actions (Sen 1999;Foucault 2001).
As power, the agency to exclude is centrally constituted by intersecting hierarchical relations with structures of gender, caste, class and development 'expertise'. For instance, if not resisted effectively, the power to exclude in the form of violent oppression and expulsion, particularly against Dalit people (Jose, Sadath and Varghese 2013;Jodhka 2016;Arulselvan 2016), is constituted by caste-based relations of control. In our approach, because humans and nonhumans contribute to actions, power is exercised not just over other humans but also nonhumans in nature, often through the deployment of technologies. The power to control and exclude is never absolute. Subjugated people resist, or sidestep, power -and things can be recalcitrant (Foucault 2001;Latour 2000), meaning they often do not behave as expected through relations of control.
Correspondingly, agency may be oriented towards sustaining relations of care. Relations of care involve solidarity, attentiveness, nurturing, concern for wellbeing, mutualism, friendship and love (Tronto 1993). They often lie at the heart of supportive relations with members of family or social movements for justice, and with people who help develop skills and knowledge (e.g. through collegiality and apprenticeships). Through identification and solidarity, relations of care are central to constituting the agency of freedom (Glissant 1969, in Roget 1989. They can also involve close engagement with tools and with materials in nature, like soils and plants, which are required for skill and knowledge development (Arora 2019; Puig de la Bellacasa 2017). Forms of agency, such as learning and skilling, are constituted by relations of care with tools and other materials. Relations of care can thus play a central role in enabling the agency of building livelihoods by the 'excluded' (cf. Hanrahan 2015;Bosher et al. 2007).
While distinguishing between relations of control and relations of care (as outlined above), we must account for the ways in which they coexist and intersect (Murphy 2015). For instance, everyday control associated with patriarchy means relations of care at home are asymmetrical rather than mutual. Across India, women perform much of the caregiving for the disabled, for the elderly and for children (e.g. Addlakha 2020; Ahlin and Sen 2020). In general, intersecting relations of control and of care can work together in assemblages to constitute different forms of agency. For example, patriarchal and caste-based relations of control can play a central role in constituting agency to exclude women's voices in participatory development (Elias et al. 2020). Yet to silence the 'oppressed' caste women in development interventions, powerful actors may depend on relations of care within their own families, caste groups and professional communities.

Relational agency of the 'excluded'
As noted above, the situated agency of 'excluded' people can take many forms. In the following, we focus on two broad forms of agency for freedom (Sen 1999) emphasised in studies on social exclusion. First, we explore 'excluded' people's struggles to resist sociomaterial relations of control that constitute exclusion. Such struggles may be triggered by people's experiences of exclusion. Crucially, resistance may be supported through sociomaterial relations of care involving members of one's family, political party or social movements.
'Everyday forms of resistance' (Scott 1985) can often go unnoticed (or are tolerated) by members of the dominating castes. Such resistance can, for example, be part of 'traditional' rituals in which Dalit people express opposition to caste dominance (Rao 1996;Gorringe 2016). In some situations, particularly through alignments with wider social movements (Mines 2005 In addition to direct resistance, we highlight 'excluded' people's actions to build their livelihoods. This agency often thrives when people attempt to circumvent particular relations of control (Jakimow 2012). Exclusion can attempt either to fix the place or to demarcate the paths within which subjugated humans and objectified nature can move (Glissant 2010). To avoid being locked in place, excluded people often migrate to urban areas, where exclusionary power constituted by rural relations of control (around landownership, caste and gender) may be partly circumvented. However, migration can also entangle people in other assemblages that enact the power to control and exclude in other ways (e.g. urban employers offering exploitative wages and oppressive working conditions) (Breman 2007;Mosse 2018;Vijaybaskar 2020; also see Abu-Lughod 1990).

Narrating sociomaterial relations through life histories
To further develop our conceptual approach, we rely on the life history narratives of three people, two women and a man, residing in two villages of Tiruvannamalai district, Tamil Nadu. These narratives are based on five to eight open-ended interviews with each person. Interviews were conducted in people's homes, farms and other locations, such as workplaces, primarily by two of us over the course of seven months (October 2017 -April 2018). We had intermittent follow-up interactions through short visits until December 2018. All interviews were conducted in Tamil and the audio recordings were translated into English for transcription.
Life histories help access marginalised perspectives by approaching people's memories, not as chronological 'facts' but rather as reflections on events and processes (Portelli 1981: 99;de Heering 2016). The stories we narrate below are not deterministic explanations of social exclusion or of effective strategies to confront it. Instead, they are uncertain accounts of agency constituted and performed within assemblages, as remembered and considered significant by people. Stories include accounts of exclusion and marginalisation alongside valiant attempts to resist caste-based and genderbased relations of control. They include people's relations of care with family and ecologies, as well as negotiations with state and non-state actors.

Xavier
Xavier is 77 years old. He emphasises that, as a Dalit, caste is central to the relations of control shaping his agency. However, relative to other Dalit people in his village, Xavier has developed an extensive network with agricultural extension officers and panchayat members. His daughter-in-law and son are also both gainfully employed.
Xavier remembers building his livelihood through farm-labour, sharecropping and driving a bullock cart for transportation. Xavier's wife has also worked as a farm-labourer. His sharecropping experience was particularly exploitative. While Xavier bore all expenses and labour, the dominating caste landlords appropriated half the harvested crop. This exclusion from retaining the fruits of his labour, through sharecropping relations of control, was central in shaping Xavier's agency to struggle for freedom by saving money and buying his own parcel of land.
Constituting Xavier's agency to acquire land, centrally, were relations of care with members of his family. Among these were relations with his parents, from whom Xavier inherited a small piece of uncultivable land that he sold a few decades ago for 42 rupees, with which he bought a piece of gold jewellery. By selling this piece of jewellery and by relying on all of his family's (his wife's, son's and daughter-in-law's) savings, Xavier purchased a small parcel of cultivable land over a decade ago.
Xavier was only able to afford a piece of dry land. Without irrigation, he could not cultivate, so he submitted a request for an electricity connection. A few years passed before the electricity was finally connected. Xavier spent substantial amounts of money to get the connection and dig a bore-well for irrigation. Central to assembling this agency were relations of care with workers who did the digging (using machines), a supportive junior engineer from a government department, an investment of tens of thousands of rupees, electricity, a pump and -most significantly -groundwater.
Crucially, Xavier's relations of care with groundwater are asymmetric. While the groundwater reservoir helps Xavier cultivate his fields, Xavier cannot simply reciprocate and be attentive to replenishing the groundwater. His agency to do so is constrained by the wider relations of control over land and life, which have been promoted through agricultural industrialisation since the Green Revolution, driven by modern expertise developing technologies in electrical, mechanical, chemical and biological laboratories (Cullather 2004;Sharma 2019). The Green Revolution's relations of control, mediated by modern technologies, constitute farmers' agency to extract groundwater in the region rather than to harvest rainwater for irrigation (Janakarajan 2004).
Through modern relations of control that enable the extraction of groundwater using the bore-well and electricity, Xavier is now able to cultivate paddy. In this process, Xavier engages with chemical pesticides bought from a local shop. The shopkeeper offers guidance on how to use the chemicals and Xavier applies the pesticides to multiple crops including beans, groundnut and lentils. At first sight, Xavier's relation with pesticides and the local shopkeeper selling them may appear to be one of care for his crop, but this relation of care is realised through the deployment of pesticides geared towards the control of 'pest attacks'. These relations of control towards nonhuman insects, turned into pests, make Xavier dependent on chemical technologies developed in laboratories and on a local shopkeeper's expertise.
In contrast, in assembling other forms of agency, Xavier relies centrally on relations of care with his ecological surroundings. For example, some years ago Xavier nurtured a few fledgling coconut trees by collecting water from leaking municipal supply pipes. In turn, the leaves from the trees helped Xavier cultivate his chilli saplings by sheltering them from sunlight. The same leaves were crucial in building the roof of a shed on Xavier's farm. The coconut trees also contribute food for his family. These relations of care between Xavier and the coconut trees are mutualistic.
Xavier remembers contesting an election to become the village panchayat president, which he lost by a small margin. This election, and his lifelong membership of a Dravidian political party, may be understood as attempts to resist historical marginalisation produced by casteism. Xavier's agency to contest in the election was assembled through a loan of 25,000 rupees from a nephew. Just like other forms of Xavier's agency, contesting the election was characterised by uncertainty. The election also tied Xavier into another relation of control -indebtedness. To circumvent this relation, Xavier repaid his debt by relying on his land and property rights. He transferred three-fourths of an acre of land in the nephew's name.
Xavier emphasises particular values he cares about. First, he believes in social progress that makes people more 'civilised' over time. Such progress for Xavier does not depend on the agency of social movements. Referring to the local Ambedkar movement (Gorringe 2005), he says 'the eradication of untouchability [in Tamil Nadu] is due to cultural and political civilisation rather than this movement.' Second, Xavier attaches himself to values of self-discipline (cf. Pandian 2010), due to which he has never succumbed to alcoholism and has worked hard all his life, even for meagre wages and often doing multiple jobs simultaneously. He even gave up smoking beedis years ago.
Inherent to assemblages of relations of control and care that constitute Xavier's agency are the uncertainties of inadequate rainfall, low crop prices and his old age, which often makes it hard to do extended work in the field. Rather than suppressing such uncertainties to convey an impression that Xavier is in control of his livelihood, Xavier embraces the uncertainties as unavoidable facts of life. Perhaps the biggest uncertainty is that of death itself, which Xavier also appears to take in his stride: on one day, when three of us visit him, Xavier refuses to accompany his wife to an astrologer. He says: 'My time is good, I do not need a fortune-teller.' However, at the same time, Xavier talks about the recent passing away of a close friend. He also narrates a dream from three or four nights ago involving the lord of death assisted by three policemen. Xavier says the lord of death sported 'a big moustache, riding a buffalo. He asked his assistants to call me. I told them that I have more days to live.'

Madhavi
All three of Madhavi's daughters are married. Her husband passed away a few years ago and Madhavi now lives alone. Madhavi and her husband did not inherit any land, nor were they able to buy any later. Madhavi belongs to the Yadava caste. Yadavas own significant parcels of land and are second only to the Mudaliars in terms of economic and political dominance in Madhavi's village.
Madhavi recalls how she was ridiculed for not having a male child. Such ridiculing as an exclusionary form of agency is assembled through relations of control associated with patriarchy. Madhavi highlights this role of patriarchal norms in privileging sons over daughters: 'The last rites are to be performed by a male child. He only has to carry the fire for cremation.' Yet Madhavi also articulates resistance against patriarchal relations of control by asking if daughters can also carry out the last rites. She observes: 'Some girls are now boldly doing just that!' Over four decades ago, at the time of her marriage, Madhavi remembers bringing a cow from her village as dowry. From this one cow, she and her husband gradually built up a herd of 15 cows. However, governed by patriarchal relations of dowry as part of their daughters' weddings, Madhavi and her husband were forced to sell 14 of the cows. Madhavi eventually gifted the last cow to one of her daughters.
Madhavi and her daughters share a bond of care. Madhavi notes how the daughters offer her financial support every time they visit. They also phone her regularly to inquire about her health and wellbeing. If her voice sounds feeble on the phone, they ask her to consult a doctor and to skip work. These relations of care play a crucial role in constituting Madhavi's agency to build her livelihood.
Madhavi's livelihood is built on farm-labouring and a few other jobs. She emphasises the difficulty of finding farm-work in some months, particularly when there is little rain. On one day when we meet her, she has waited for a few hours along with 50 other people to find a bit of farm-work. She earned a meagre wage of 30 rupees that day. Madhavi asks how anyone can afford to buy nutritious food on such a wage. She highlights that her survival depends crucially on the few kilogrammes of rice accessed every month, without payment, from the public distribution system.
Madhavi narrates how mechanical harvesters are displacing farmworkers such as herself. Yet, within this displacement, Madhavi highlights the role of uncertainty by invoking rainfall patterns. She observes that, when rainfall is adequate and the crop is plentiful, farmers look beyond machines and employ farmworkers to help harvest. In this story, sociomaterial relations of control and care intersect to constitute farmworkers' agency to build a livelihood. On one side are relations of control, in which labour value is exchanged with landlords (who are generally men belonging to the dominating castes) mediated by labour-displacing technologies like harvesters. On the other side are relations of care with the crop, which fall into place when there is adequate rainfall. Beyond these uncertain attempts to build a livelihood undertaken by Madhavi (and other farmworkers), these intersecting relations of control and care constitute the agency of farmer-employers to harvest a crop.
Crucial in the assemblage constituting Madhavi's agency to build her livelihood is her old-age pension from the state. While the pension is crucial for Madhavi's agency to care for herself, the process of securing this pension is loaded with relations of control (involving state bureaucracy and officials). According to Madhavi, applying for the pension begins with submitting proof to the local village agricultural officer that the applicant owns no land and has no family members providing for her. If the village agricultural officer gives his signature of approval, the approval of two sub-district officials is required. To move an application through these multiple levels of bureaucracy, bribes often have to be paid to each officer involved. To circumvent these relations of control with the state, Madhavi opted to apply through a broker, paying him 4,000 rupees. The broker helped fill in and submit all the forms, collect the required signatures and secure the pension.
In addition to her old-age pension and farm-labour, Madhavi relies on two occupations for her livelihood. First, Madhavi cleans the kitchen and utensils at her village's milk centre. This form of agency depends on Madhavi's engagement with milk measurement vessels, kitchen sinks, water and cleaning materials. She also encounters dairy farmers and the milkmen who purchase milk from them. For this job, Madhavi receives a small salary. She also receives 200 millilitres of milk in the morning and the same quantity every evening. Relations of care, such as those with the milk and with other nonhumans and humans she works with, are central to constituting Madhavi's agency to sustain her livelihood and wellbeing.
Second, Madhavi works as a spiritual healer. She says: 'I am possessed by my family god and I cure people who are sick using a mantra.' She is known in her village for getting rid of the 'evil eye'. In addition to cultural beliefs and values linked to such practices, Madhavi's agency as a healer relies on careful engagement with materials such as turmeric, lime, a small pot-like vessel and water. This form of agency as a healer is central to sustaining Madhavi's relations of care with many other non-Dalit women in the village.
Madhavi describes how she resists exclusionary forms of agency shaped directly by patriarchal relations of control. In one such situation, before the birth of her third daughter, Madhavi's husband left the village to live with his lover. He took Madhavi's gold earrings with him. To raise her daughters as a single parent, Madhavi had to put in more hours of work as a farm-labourer. She was also subjected to cultural control by the village community. For example, while her husband was away, Madhavi and her daughters were asked to stay away from village events meant for families.
Madhavi's husband returned to the village three years later. To appease Madhavi, he sent gifts, which she declined to accept. Madhavi also sent her husband back out the door each time he came to visit her house. She thus intensely resisted her husband re-entering her life. This resistance was eventually broken after a village headman intervened. According to Madhavi, the headman said her husband had 'realised his mistake and asked [Madhavi] to be silent and accept him.' This situation may be seen as a backlash to Madhavi's resistance. As a form of exclusionary agency, this backlash was constituted by relations of control in which the headman, embodying village-level patriarchy, aligned with intrahousehold patriarchy.

Asai
Asai is about 65 years old. Her husband passed away over a decade ago. She is mother to four daughters and a son, and grandmother to 16 boys and girls. She feels responsible for the families of her children, saying she carries 'the burden of five families' on her mind.
Asai has a title to 1.2 acres of unirrigated land, but the land is effectively under her son's control. The same is true for the house she built. Her son occupies it now with his wife and children. Asai has no access to the house. Her exclusion from her land and house is just one situation among many in which she faces the brunt of patriarchal relations of control. Other instances include her landowning father excluding her from any inheritance; her alcoholic husband selling a part of their land to buy alcohol; and bearing the burden of repaying the loans taken by her son and husband. Intersecting with patriarchy, relations of control associated with caste are crucial to Asai's exclusion and marginalisation as a Dalit woman.
Probably the most critical form of exclusion is exemplified by a situation that occurred a few decades ago. Asai was then pregnant with her first child. Dalit people were barred from accessing commons such as the village tank. On a day when Asai and some other members of the village Dalit community were catching fish from the tank and collecting firewood nearby, they were seen by some dominating caste villagers who decided to 'punish' Asai and her companions in public by forcing them to do 20-30 sit-ups while holding their ears. Such sit-ups were commonly used by teachers to discipline children in schools. Asai was subjected to this humiliation of repeatedly squatting and standing while she was pregnant.
A few years after this incident, Asai remembers how she was made to lose her job as a cook in the government-run childcare centre: the village balwadi. Many caste villagers had asked their children to refuse the food cooked by Asai because she was Dalit. Under pressure from caste villagers, government officials then decided to terminate Asai's appointment. Asai believes this expulsion, driven by intersecting relations of caste and state-based control, was meant as further 'punishment' for her political activism against casteism. By this time, Asai was actively engaged with Ambedkar Makkal Iyakkam (AMI), the Ambedkar People's Movement.
'I was determined to gain our rights over the tank,' says Asai about her resistance. During a meeting organised by AMI, Asai took the microphone to announce that the village tank was located on common land so it should be accessible to all inhabitants. However, caste villagers punished Dalit people for accessing the tank for fish or firewood and also appropriated all firewood from the tank area. Money from auctioning this firewood was used only for celebrating their festivals. The Dalit community did not receive its share. To challenge and change this, Asai and others in AMI collected signatures on a petition. Asai travelled to deliver this petition to the government's Block Development Office, emphasising that Dalit people wanted a share of the money generated through auctioning firewood to help organise community events at the church in the Dalit part of the village.
Upon returning from the Block Development Office, Asai was beaten by her husband. He demanded that Asai go back to say that Dalit people did not need a share of the auction money. Her husband feared Asai (and her family) would be attacked by dominating caste people. The husband's agency of patriarchal violence, intersecting with the threat of casteist violence, failed to shake Asai's resolve to struggle for what she believed to be Dalit people's rights.
When government officials arrived to oversee the auction, Asai was called to meet them. While recognising Dalit people's rights to their share of the auction proceeds, the officials asked Dalit people to give up their share because dominating caste villagers needed the money for a special fire festival that year. Asai retorted that caste people had enjoyed the auction money for many years. She added: 'We are also conducting the Mary festival at our church. Please ask them to give up the money this time.' And so the auction money came to be shared, and controlling relations of caste were transformed in Asai's village. Eventually Dalit people were able to access the tank for fishing and for other uses, but only that portion of the tank adjoining their residential part of the village.
This story illustrates Asai's agency to transform casteist relations of control central to the exclusion of Dalit people in her village. The sociomaterial agency of resistance involved the mobilisation of many Dalit women and men in the village, as well as the wider AMI.
It is important to emphasise that Asai's agency as an activist was assembled within a heterogeneous assemblage of sociomaterial relations. Through this assemblage, while being entangled in casteist, patriarchal and state-based relations of control, Asai was enabled by many relations of care. These included relations with AMI activists from outside her village; relations with petitions and signatures; relations with fish and firewood; relations with the church and religious festivals; relations with values of equality against the casteist fallacy that some people are superior than others; relations with a spirit of entrepreneurship -Asai started small businesses selling vegetables and cooked snacks; and relations with modern technologies, such as the microphone and modes of transportation.
In addition to resistance, Asai describes her attempts to circumvent power and build a livelihood. She does the labouring work of transplanting rice, the wages for which often depend on the number of saplings handled by a worker. Asai notes how workers in a nursery compete with each other to collect more saplings. Comparing this competition to a fight, Asai notes: 'People are not fighting over property, they are struggling for wages.' She observes how agricultural production is oriented by power to depress wages for farm-work.
First, Asai highlights how modernising technologies are adopted to displace labour. She observes how water for irrigation is extracted from a well at the flick of an electrical switch, displacing the hard labour of pushing a wooden water wheel until one's 'legs shivered with strain'. Similarly, the use of synthetic fertilisers does away with the necessity of keeping cows and buffaloes for manure. The farmlabour involved in caring for cattle and in producing manure has been displaced by synthetic fertilisers produced in factories located in (peri)urban areas.
Second, Asai points out how a large share of the depressed wages earned by farmworkers pays for their children's education; for industrial goods such as clothing, detergents and toothpaste; for modern healthcare; and for motorised transportation. A farmworker thus has little left to spend on food, which means the price of food is kept low. Low food-crop prices squeeze small and mid-sized farmers' incomes, constraining their ability to pay decent wages to farmworkers. This is particularly true in the last two decades of chronic indebtedness among small and mid-sized farmers, also due to the costs incurred for expensive chemicals and seeds used in modernised agriculture (Arora 2012;Vasavi 2012).

Concluding reflections
That technologies and nature are mobilised as resources to marginalise and dominate people is well established in studies on social exclusion. These studies also acknowledged that social networks help constitute the agency of resistance and livelihood-building by the 'excluded'. However, the role of engagement with technologies and nature in constituting 'excluded' people's agency is underemphasised. Nevertheless, as we have illustrated, these engagements as relations of care are crucial for constituting agency to build livelihoods and wage resistance.
As noted by Asai and Madhavi, relations of care established by workers with modern technologies in learning to perform industrial agriculture can be subsumed within wider relations of control targeted at local ecological and social processes. These wider relations are often engineered by modernising expertise and by caste-based and gender-based landownership. The observations by Asai and Madhavi corroborate existing feminist scholarship on politics of care, which holds that caring relations are often conditioned by wider structures that engender the power to exclude (e.g. Murphy 2015). This conditioning of care by control can be so strong, as in the case of patriarchy at home, that caring turns out to be anything but egalitarian and mutualistic (cf. Addlakha 2020).
The conceptual approach we have developed and illustrated can be useful in mapping such sociomaterial complexity arising out of intersections of relations of care and control, which not only constitute exclusion but also, crucially, diverse forms of agency involved in resistance and livelihooddevelopment. Our concepts, such as ways of relating -control and care -are practical, but abstract. They can find relevance in different contexts in ways that are highly sensitive to local and trans-local conditions. Such sensitivity is crucial because sociomaterial processes of exclusion are extremely varied, even within a single region like Tamil Nadu. Our approach can thus help address questions that are directly relevant for policies and politics to apprehend situated forms of social exclusion. Such questions include the following. a) What kinds of assemblages, such as those manifesting in Asai's activism, can serve as fertile ground for constituting forms of agency to struggle for freedom from casteist and gendered exclusion? How do public policies and civil society strategies strengthen sociomaterial relations of care developed in such situations? How can the state contribute to effectively suppress and resist backlash by those who dominate through caste, gender, ethnicity, and religion? b) In constituting the agency of resistance, how do social movements such as AMI forge relations of care in which local concerns are kept central? How can policy makers move beyond standardised solutions to emulate the caringly situated support provided by such social movements? c) In building livelihoods and enacting resistance, constituted centrally by people's relations of care with their surroundings, people produce 'subaltern' knowledge beyond dominant development paradigms of modernisation, like the Green Revolution. How can policies and civil society strategies promote such knowledge to open up alternative possibilities for transformation towards sustainability beyond modernity in India?
Finally, as our approach promises to help produce accounts of agency that are uncertain, we conclude with a note about political action under conditions of uncertainty. While uncertainty can align with power to marginalise and oppress the 'subjugated' -as noted by Xavier -uncertainty is an unavoidable feature of sociomaterial life, particularly in a contemporary world riddled with multiple 'global' crises, from the COVID-19 pandemic and climate disruptions to rapid biodiversity destruction. By admitting uncertainty, we can highlight the need of political action to intensify relations of care to support 'excluded' people struggling for freedom from casteist, gendered and other forms of oppression. By foregrounding uncertainties otherwise suppressed in promoting development that favours the dominant, our approach can also help foreground possibilities of holding 'exclusionary' power to account and open up spaces for alternatives led by the 'excluded' that are urgently needed for transformations to sustainability.