Salience and Skewness Preferences

Whether people seek or avoid risks on gambling, insurance, asset, or labor markets crucially depends on the skewness of the underlying probability distribution. In fact, people typically seek positively skewed risks and avoid negatively skewed risks. We show that salience theory of choice under risk can explain this preference for positive skewness, because unlikely, but outstanding payoffs attract attention. In contrast to alternative models, however, salience theory predicts that choices under risk not only depend on the absolute skewness of the available options, but also on how skewed these options appear to be relative to each other. We exploit this fact to derive novel, experimentally testable predictions that are unique to the salience model and that we find support for in two laboratory experiments. We thereby argue that skewness preferences—typically attributed to cumulative prospect theory—are more naturally accommodated by salience theory.

In democracies, innovative political institutions have opened up scope for direct public participation often in the form of talk: citizens talking to the state and mutual talk among citizens on matters concerning community development. A prominent example is the Indian gram sabha, or village assembly, which occurs in a highly stratified context. This paper undertakes a talk-centered analysis of the gram sabha with a focus on examining the oral participation of women in general and women affiliated with microcredit self-help groups who have access to an associational life. The qualitative analysis of 255 gram sabha transcripts from four South Indian states finds that women associated with microcredit self-help groups employ a wider variety of narrative styles and utilize a more multilayered structure to convey their messages compared with all women taken together. Thus, the difference is not so much in the numerical instances of talking or in the types of issues raised, but rather in the quality of participation. The paper makes an important theoretical contribution by proposing the concept of oral democracy as an alternative to deliberative democracy, and urges an analytical focus on the oral or oratory competency of subordinated groups as they participate in these important institutions.

INTRODUCTION
Notions and practices of inclusive, participatory democracy in the real world have witnessed the rising diffusion of 'deliberative democracy'. The genealogy of 'deliberative democracy' in the modern history of ideas traces back to the seminal works of Habermas (1984Habermas ( , 1990 on communicative action as a form of organizing democracy and the public sphere necessary for accomplishing this new vision and its salutary implications for moral consciousness. The Habermasian notion of deliberative democracy is anchored in a pair of governing assumptions: rational argumentation, a singular form of communication, is the only admissible mode of dialogue; and the "ideal speech situation" of discursive equality, that imagines a moral community of equal citizens, prevails among participants. This assumption of discursive equality allows proponents to conceptualize this system as guaranteeing fairness in deliberations and eventual decision-making. Deliberation is envisioned as having a preference-altering effect on its participants, increasing tolerance and liberal-mindedness, and moving divergent opinions to a consensus through giving and hearing reasons. This idea of a talk-centered deliberative democracy received further elaboration in the works of liberal political theorists (Cohen 1989;Mansbridge 1990;Fishkin 1991;Bohman 1996;Dryzek 2000;Gutmann and Thompson 2004;Fung 2004).
Around the world, practical attempts continue to be made to incorporate 'deliberative democracy' into models of participatory (community) development to make decisions more representative and more moral and consensual. In these forums participation takes the form of public deliberations in civic settings, i.e. verbal/ oral communication of ideas, arguments, and justifications in front of a purposeful gathering of people. Notably, such deliberative forums have been constitutionally mandated in India (known as gram sabha, i.e. village assembly), the world's largest democracy, being held in thousands of villages around two to four times a year as part of the central government's mandate of decentralized governance and devolution of powers (Rao and Sanyal 2010). They are also present in Brazil, where participatory budgeting has been pursued for the last few decades, beginning with successful experiments in Porto Alegre and then diffusing to other municipalities (Baiocchi 2005;Baiocchi, Heller, and Silva 2011). Recently, such deliberative forums have emerged in Canada and U.S., where they are being experimentally used for state and provincial civic and administrative purposes (see Fung, Wright, and Abers 2003;Delli Carpini, Cook, and Jacobs 2004;and Gastil and Levine 2005). Public discussions in these assemblies usually focus on allocation of public finances to community infrastructure and resources, on the redistributive task of allotting state subsidies and benefits to citizens, or concern decisions that affect the entire community.
Considering such deliberative systems (Mansbridge, 2015), a fundamental question is whether all citizens have equal capability, or competency, to deliberate, i.e. to engage in reasoned argumentation in a public forum. As Heller and Rao (2015) point out, the assumption of discursive equality takes for granted that, once given equal structural opportunity for participating in a public discussion on civic matters, this overrides differences in deliberative capabilities that may potentially result from social and economic inequalities. Yet, descending from the lofty heights of utopian political theory and taking a sociological view, can reveal the possibility that "durable inequality" (Tilly 1999) might affect deliberative capabilities through persistent disadvantages (e.g. poverty, lack of education, gender subordination) and exclusion (e.g. subordinated groups being historically debarred from appearing and participating in mainstream public sphere institutions and never having spoken in public). Filtering through the lens of durable inequality thus makes us wonder how social stratification might inflect oral competencies. It also makes us wonder whether such oral competencies can be cultivated by experiences of participation in other associational spheres. This paper has three closely aligned goals. First, it aims to re-theorize deliberative democracy as oral democracy, democratic inclusion based on the equal right to speak, without the weighty assumption of reasoned argumentation as the only appropriate form of talking and the only one that moves us closer to democracy. The empirical goal in support of this theoretical aim is to undertake a "talk-centered" (Eliasoph 1996) analysis of democracy. This analytical strategy is befitting because the public sphere, as envisioned or rhetorically invented by Habermas, and which has inspired practical experimentations "designates a theatre in modern societies in which political participation is enacted through the medium of talk" (Fraser 1990, 57). Analyzing talk directs our attention to a set of key issues that are easy to overlook unless we conceive of talking publicly on public issues and talking to the state (since many deliberative forums are state organized) as a capability, or competency. It directs attention to citizens' oral competency, i.e. their capability of speaking in public in civic settings in a manner that is effective in eliciting a response, generating a discussion, and reaching a decision.
Second, it specifically focuses on analyzing women's oral participation in the Indian gram sabha that was instituted as part of political reforms to implement decentralized governance and participatory development. These are village assemblies that are tasked with a plethora of important governance and community development functions. These include oversight of government funded public works; giving input and ratifying redistributive decisions such as beneficiary selection for government sponsored schemes and subsidies aimed at below poverty households and other disadvantaged groups, like the schedule castes and tribes; and when necessary, demanding accountability from village-level officials on public budget and public works and their responsiveness to public needs. These assemblies are supposed to be attended by the general public and by elected panchayat leaders and government bureaucrats and staffs from different departments. This interface between the lay public and public officials, who symbolize the state, makes oral participation in the gram sabha an exercise of talking to the State, and the participants also experience it as such. The significance of women's participation in civic institutions is easily understood when we think of the exclusion of women from civic and political institutions and from the public sphere worldwide in the past centuries and specifically during the colonial and pre-colonial periods in India. Even in the contemporary world, many civic forums continue to be male dominated. Therefore, viewed from the perspective of gender as a "durable inequality" (Tilly 1999), analytical focus on women's oral competency in institutions of democracy, such as the gram sabha, gains added sociological significance.
Third, it investigates one potential source of influence on women's oral competency in the gram sabha. This is women's membership and participation in microcredit groups, also known as 'self-help groups' or SHGs, which number over one million in India. Starting in 1985, NGOs in India (primarily in the South Indian states) started forming women's groups primarily to provide financial services (thrift and credit) to the poor. Over the decades, a noncommercialized group-based microcredit has taken root in India through institutional collaboration between a diverse set of organizations (including multilateral agencies such as the IFAD 1 in the early days and The World Bank during recent times, local NGOs, state and national governments, and public sector banks (NABARD 2 ) and the Reserve Bank of India). This groupbased model of microcredit is distinguished by its "associational mechanism" (Sanyal 2014(Sanyal , 2009) that requires women's mandatory participation in weekly or bi-weekly group meetings at 6 the neighborhood level and monthly meetings at the cluster level. This raises the question, whether women's participation in microcredit SHGs influences in any way their oral competency in participating successfully in the oral democracy of the gram sabha. The relevance of this question is supported by a voluminous literature exploring the relation between associationsim (its civic values) and democracy. 3 This includes contemporary research on two Latin American cities (São Paulo and Mexico City) that reveals a significant association between individuals' participation in associations and active citizenship, and shows this association to be higher for lower classes, with a striking increase (in one city) for women (Houtzager and Acharya 2011).
To summarize, the motivating goals of this paper are the following: 1) To propose the concept of oral democracy as an alternative to 'deliberative democracy' and to gain new analytical purchase on talk-centered democracy.
2) To analyze women's oral competency, or more specifically, their oratory competency, 4 i.e. what women talk about and how they talk, in the gram sabha.
3) To analyze whether membership in microcredit SHGs has any impact on the frequency, issue focus, or manner of women's talk at the gram sbaha.
The rest of the paper is divided into several sections. Next we explain why we need an alternative theory and elaborate on the concept proposed. Then we move on to the empirical section, which is divided into context, data and methods, general findings, and in-depth discussion of state-level findings from one of the four states studied.

RETHEORIZING DEMOCRACY: ORAL DEMOCRACY & ORAL/ ORATORY COMPETENCY
Deliberative democracy has come under critique from some political theorists who have begun to challenge the classical view (see Mansbridge, 2015 for a recent review and response). While some have called for an expansionary revision of the original conception and its attendant assumptions, a few have launched a more trenchant critique and proposed alternative theories of democracy. Overall, these revisionist accounts and critiques have revealed the exclusionary origins and the "bourgeois masculinist" (Fraser 1990, 62) conception of deliberative democracy and the "public sphere" and have come to recognize modes of presentation of facts, opinions and experiences that fall outside the ideal-type format of reflexive arguments. For instance, later scholars have come to acknowledge formerly inadmissible things like: emotions -that may infuse rational argumentation, ignite conflicts, make people act outside the parameters of narrow rational self-interest and even have negative consequences (Mansbridge 1983;Elster 1996;Mouffe 1999;Sanyal 2015); and storytelling -a discursive style that is found to be present in public sphere deliberations and that has positive implications for making democracy inclusive of disadvantaged groups and minority points of view (Young 1996(Young , 2000Sanders 1997;Polletta and Lee 2006). In this section we discuss the notion of "subaltern counterpublics" (Fraser 1990, 67) that has been proposed as a critique of deliberative democracy. We then elaborate on oral democracy, discussing why we need it and what distinguishes it from existing alternatives. Nancy Fraser has made a fundamental theoretical contribution by calling for "an alternative, post-bourgeois conception of the public sphere" (Fraser 1990, 58) and by proposing a theory of democracy in stratified societies envisioned around the proliferation of "subaltern counterpublics". These "counterpublics" are constituted by members of historically subordinated groups, among them women, who have all been excluded from the bourgeois public sphere and have resorted to creating their own associational life and parallel discursive arenas. Here "members of subordinated social groups invent and circulate counterdiscourses, which in turn permit them to formulate oppositional interpretations of their identities, interests, and needs. (Fraser 1990, 67)," ultimately aimed at reducing their disadvantage in official public spheres.
Subaltern counterpublics "help expand discursive space" beyond the dominant mainstream public sphere and force a "widening of discursive contestation" by making everything open to be challenged, discussed and defended and have a "publicist" orientation (Fraser 1990, 67). Their emancipatory potential lies in the dialectical relationship between the scope they provide for withdrawal and regroupment and of training in agitational activities. Therefore, it is the "contestatory function" of subaltern counterpublics that makes them valuable for enhancing democracy's reach in stratified societies (Fraser 1990, 67).
Fraser's alternative conceptualization is inspired by an incisive critique of Habermas's liberal model of the bourgeois public sphere and its shortcomings -particularly its blindness to status inequalities; its failure to recognize gender which was deliberately used as an organizing principle for the public sphere that was historically constructed as a masculine site; and its idealized separation of the public sphere from the state, which invalidated private interests as an appropriate theme of talk. Neither of these conceptions -free of inequalities and free from the state -are supported by the practical reality of the public sphere in the twentieth and twenty-first centuries.
Given this paper's focus on women's participation in democracy, let us pay special attention to Fraser's discussion of gender as a principle of exclusion from and construction of the public sphere. Drawing on the revisionist historiography of Landes, Eley, and Ryan, Fraser highlights the inequalities and gender and class based exclusions that were the reality of the Western European contexts from which germinated the rhetoric of publicity and accessibility. Landes (1988) has shown that, in France, a masculinist ideal of the public sphere emerged in direct and deliberate opposition to salon culture where women participated actively and which the republicans stigmatized as feminine. Eley (1992) has shown that in England and Germany, women were excluded from the liberal public sphere and from the associational life that had 9 sprung up in the form of clubs and voluntary organizations. These associations were restricted to bourgeois men and seen as a sphere where they could cultivate themselves as a distinct class separate from the former aristocracy and from the lower classes. Thus the construction of the public sphere happened in conjunction with and deliberately designed to facilitate a new bourgeois class formation. And participation in the public sphere was treated as a mark of class distinction in Bourdieu's sense (Fraser 1990, 60). "This process of distinction, moreover, helps explain the exacerbation of sexism characteristic of the liberal public sphere; new gender norms enjoining feminine domesticity and a sharp separation of public and private spheres functioned as key signifiers of bourgeois difference from both higher and lower social strata (Fraser 1990, 60)." As a result of this exclusion women had to seek out alternative ways of participating in a public and political life. In North America, in the nineteenth century, elite bourgeois women created exclusively female voluntary associations, including philanthropic and moral reform societies (Ryan 1990) in which they transformed themes such as domesticity and motherhood, commonly perceived as private themes, into arenas for public action having a lasting hegemonic effect on the gender ideology of other classes. And working class women participated alongside men in supporting male-dominated working class protest activities. Thus Fraser justly critiques Habermas's failure to see the irony of how a discourse of publicity celebrating accessibility, rationality, and equality was used with the strategic aim of constructing class and gender distinctions.
The other constitutive assumption that is increasingly untenable is the public sphere being an institutionalized arena of discursive interaction among citizens on common affairs that is fully separate and distinct from the state, a separation deemed necessary to allow for the voicing of critical perspectives on the state. As mentioned at the start, some states have adopted models of decentralized governance and participatory democracy, which has meant that the state has played a catalytic role in creating discursive spheres for the discussion of common affairs by common citizens. For example, the municipal government's adoption of 'participatory budgeting' in Porto Alegre, Brazil, has created citizen's assemblies and opened up scope for public-minded discussions (Baiocchi 2003). Another compelling example is the gram sabha (village assemblies) in India. It is an example of a discursive sphere that has been inaugurated to facilitate conversations, decision-making and scrutiny among citizens on common affairs, but is intimately linked to the state. It has been created through state action; bureaucrats, administrators, and local level political leaders who are representatives of local governments, preside over and participate in these discursive exchanges; in fact one of its aims is to facilitate discursive exchanges between the state and its citizenry at the village level; allocation of state funded private subsidies and decisions of state-funded public goods are taken in this forum.
However, as the empirical evidence from our research shows, this intimate link with the state does not preclude the possibility of critical discourse about the state -of its negligence, inefficiency and corruption -being voiced in these discursive forums. In fact, a key function of the gram sabha is to facilitate the critical scrutiny of the state by opening up channels for public accountability of political leaders and state officials. As Fraser also notes "with the emergence of "welfare state mass democracy," society and the state became mutually intertwined" (1990, 59).
But, unlike what Fraser tells us, that has not necessarily meant in all cases that critical scrutiny of the state has been compromised; in some cases it has been enhanced.
A final set of criticisms by Fraser and also by Mansbridge has been about the disallowing of 'private interests' from the public sphere. Fraser (1990) has argued that 'private' and 'public' are cultural classifications and rhetorical labels that are strategically used in political discourse to treat some interests and topics as valid and to devalue and delegitimize others. Mansbridge (1998) also raises this concern when she argues that assuming that deliberation must be deliberation about the common good narrowly restricts deliberation to talk framed from the standpoint of a single, all-encompassing "we" that might, after all, not be inclusive of the interests of marginalized groups and individuals. The inadmissibility of 'private' interests prevents them from being expressed, understood and debated over, which goes against the central premise of deliberative democracy that all interests and concerns should be publicly expressed and debated to yield a consensus regarding what is common good.
With these critiques in mind, and building on but departing from the genre of validating and revisionist extension accounts emerging from valuable contributions by sociologists like Baiocchi, Heller and Polletta, in this paper, based on empirical evidence drawn from the gram sabha, we introduce the concept of oral democracy. This concept is proposed not as a utopian ideal, but as a representation of some prominent models of "actually existing democracy" (Fraser 1990). It is also proposed as an analytical frame to guide our investigations into empirical instances of talk-centered democracy without the encumbrances of the theory of deliberative democracy. Oral democracies, as they exist, are geared toward giving all citizens the equal right of talking to the state and the equal right to speak to each other as members of a polity. The goal is also to engage in an interactive and participatory process of decision-making and scrutiny on matters of community welfare and development. While deliberative democracy theory focuses on mutual conversation among citizen-peers outside the purview of the state, thereby, neglecting to focus on citizens' interactions with the state, political systems based on oral democracy recognize that talking to the state is an important opportunity that has been inaugurated by this talk-centered model of democracy. Indeed this study of Indian village assemblies shows that when such participatory governance initiative is organized by the state and integrated within the state's political and governance institutions, citizens talk not only to their peers, i.e. have mutual conversation, but also talk to state officials, local bureaucrats, elected leaders of village level political institutions and experience their verbal participation as talking to the state. As evidence from village assemblies shows, now so called "subalterns" (Spivak 1988) can speak to the state in a pubic setting in their own "vernacular voices" (Hauser 1999) and without having to rely on mediating middlemen. Leaving its substantive implications aside for a moment, we must recognize that this is an important measure in which democracy has been procedurally enhanced, and that procedural enhancement must precede substantive enhancement. Therefore, examining how citizens talk to the state should be recognized as an important analytical agenda in the study of oral democracy.
Talking to the state can take various rhetorical forms, ranging from the sophisticated use of reason and factual information to narrating personal travails and verbally petitioning the state for public and private goods. In this regard the model of oral democracy intends to make an important sociologically minded intervention by arguing that the analytical focus should be on revealing the entire array of rhetorical styles in use and to examine if particular styles are associated with particular advantages and disadvantages experienced by particular groups in a stratified society, i.e. with "durable inequalities". 5 Another related focus should be on understanding how various rhetorical styles are socially evaluated by citizen-peers and by authoritative figures of the state. 6 Therefore, rather that a priori privilege only one rhetorical style and deem all others as detracting from democracy, the focus should be on revealing the plethora of oral styles and oratory strategies in use and understanding their correlates (i.e.

association with inequalities) and conversational consequences (i.e. whether a statement is recognized and responded to or is ignored and belittled, whether it generates a conversation).
Focusing on oral competency, without the presumption of reasoned or reflexive argumentation, takes us back to the historic roots of deliberative democracy, a new model of politics that arose from the prominence of rhetoric in Athenien politics (Hauser 1999). In ancient Athens, rhetorical skills were understood to be essentially political skills because speaking persuasively and effectively could influence public policy. Rhetoric was also recognized as having inventive power by having the potential to create new political visions and realities (Hauser 1999). 7 With the rise of rhetoric a new kind of politics emerged where direct rhetorical appeal was used to generate public consensus, which increasingly came to be seen as a more legitimate form of political authority than noble birth and elite dictums. "Public deliberation became more genuinely the method by which serious decisions were made, and "universal" participation led to a new form of political power emphasizing rhetorical skills over noble birth.
It valued each citizen as capable, in principle, of contributing to a solution through his participation in public deliberation (Finley 1962). (Hauser 1999, 15-16)" Consequently, due to the rising prominence of participatory democracy in the political context of ancient Athens, rhetoric, the art of public speaking, became a subject of formal study and training as part of civic education in Greece during the fourth century b.c.e. The sophists, a new group of professional travelling tutors, emerged in the fifth century b.c.e. to provide professional training in the art of public speaking and oratory skills. The Elder Sophists regarded deliberative exchanges and encomiastic performances as the polity's method for clarifying vague or poorly understood problems, uncovering new ways to frame issues, resolving impasses, and discovering shared ground for communal action. The general view of the times was that public speaking was a skill, a political skill that could be cultivated through training. It was this tradition and vision of democracy that Aristotle built on later by categorizing rhetoric as one form of ethical politics and recognizing it as a technique or art of formulating arguments in deliberative settings. In light of this historical perspective, we must recognize oratory competency within the sphere of oral democracy as a cultivable political skill, and focus on the direct and indirect influences, like participating in associations, which might impact positively or in any other fashion citizens' oratory competency.
Finally, viewing from the analytical perspective of oral democracy, we must not rule out private and group interests as detracting from democracy precisely because of the reasons argued by Mansbridge and Fraser that have been discussed earlier. We would add to this that petitioning the state for meeting individual and group interests has been an old strategy in colonial and precolonial times in India. But during colonial times, petitions had to be made in written form, making it less accessible to non-literate members of the society. But now with talk-centered democracy opening up the scope for making such petitions orally in the gram sabha, it has become a widely accessible mode of approaching the state. So, although such petitions may appear to be motivated by narrow interests (whether they are justifiable or not is a separate question), we must recognize that under oral democracy presenting private interests has become widely accessible to all groups. Consequently, there is increased publicity of private interests, and they must be publicly negotiated against competing interests in order to be fulfilled. All of this lends transparency to the allocation of state subsidized benefits for citizens, and fosters public scrutiny and input, which are fundamental goals of democracy. Therefore, in analyzing the demands and concerns citizens bring up in institutions of oral democracy, the focus should be on deciphering the substantive issues on which demands and concerns are expressed and the unit on behalf of which a demand or concern is expressed:

CONTEXT: 'Gram Sabha' and 'SHGs' in South India
This study focuses on India's four South Indian states, Andhra Pradesh (AP) 8 , Kerala (KL), Karnataka (KN), and Tamil Nadu (TN). Ideally, in order to analyze women's oral participation in the gram sabha and if SHG membership has any influences on oral competency, we would focus on states that have a record of strong institutional implementation of the gram sabha and have adopted women's SHGs for poverty alleviation and women's empowerment. All states in the country taken together, these four states were thought to provide the most realistically ideal context for exploring this issue. Each of these states were frontrunners either in implementing the decentralized governance system that laid the foundation of civic participation via gram sabha, or, they led the way in mobilizing women into SHGs. Varshney (2000) has also noted that democracy in South India is deeper than in the North, with greater degrees of political participation, and this difference could be ascribed to different histories of caste reform movements. These states have among the longest running SHG programs in the country. Also, it is worth mentioning that these states have lower levels of gender disparity on a nationally comparative scale.
As laid out in this section, the four South Indian states differ in the structure and implementation of the gram sabha. At the time the data analyzed in this paper were collected (2003)(2004)(2005)(2006), Kerala was at one end of the spectrum with highly specialized gram sabhas and Andhra Pradesh was at the other extreme with its perfunctorily conducted and poorly attended gram sabhas. Tamil Nadu and Karnataka gram sabhas, on average, had the most public discussion and interaction between the public and state officials. 9 Dramatic variations in the quality of democracy are to be expected in a federal parliamentary system established over a social landscape of extreme stratification and diversity (Heller 2000). The commonality between all four states is that they have state-wide SHG programs that were adopted at different times with Tamil Nadu having the longest running program (since 1989-90) followed sequentially by Andhra Pradesh (six years later), Kerala (eight years later) and Karnataka (ten years later). These different histories set the stage for the extent to which women participate orally in the gram sabha and whether SHG membership makes a difference. These differences mean that we should expect few generalizable findings across the four states.
Andhra Pradesh has historically been the least effective in implementing decentralization via the federally mandated panchayats and gram sabhas. Its SHG mobilization on the other hand is very well developed. Launched since 1996 as a pilot project, SHGs were first formed in three districts. Subsequently the state government adopted this SHG model into its major poverty alleviation project, called Velugu (light), which was scaled up to provide statewide coverage. The program is reported to cover sixty-six thousand households that comprise thirty-four percent of households in the program and is widely recognized as a success.
Kerala was one of the first states to wholeheartedly adopt and implement the decentralization mandate, surpassing all other states. (Heller 2001, Heller andIsaac 2005). Gram sabhas in Kerala meet approximately six times a year. But apart from the 'special gram sabha' that provides scope for discussion, all other gram sabhas in a planning cycle are limited to selecting beneficiaries for government subsidies. Other gatherings, like the Ward level meeting of 'Neighborhood Groups', targeted at fostering discussion, occur prior to the gram sabha as a leading up exercise. 10 This structure and functional specialization is because of the four-staged format for the yearly planning exercise laid out during the years of the People's Campaign that has remained in place to this day (Heller, Harilal and Chaudhuri 2007). A study reports that the role of gram sabha in the planning process has largely been confined to need identification and beneficiary selection, and its role in seeking accountability and monitoring implementation is either non-existent or unclear, and that women's participation increased in the gram sabhas in the later cycle that were focused exclusively on beneficiary selection (Kurian 2010). SHGs in Kerala were launched in 1998 with the State government sponsored Kudumbashree (family prosperity) program that boasts coverage of 50% households and 37 lakh members. Women are formed into neighborhood groups, each having 20-40 members. The group members meet weekly and engage in thrift and credit related activities and also put together a 'microplan' based on their needs. These get consolidated at the panchayat level into a 'subplan' that becomes a part of the antipoverty program of the local government. Thus Kerala has a uniquely structured decentralized development planning exercise that has constrained the role of gram sabhas and moved the task of substantive discussion to smaller sized groups.
Karnataka has been a pioneer in promoting village-level democracy, starting as far back as the early 1980s, when the state streamlined the organizational structure of panchayats. This

DATA & ANALYTICAL METHOD
The data consist of the complete transcripts of 255 12 gram sabhas selected from two districts in each of the four South Indian states: AP: Chittoor and Medak; KN: Dakshin Kanada and Bidar; TN: Dharmapuri and Coimbatore; KL: Kasargod and Pallakad. All blocks from within the sampled districts were chosen to be the closest possible in their majority language to a block in the matching district of the neighboring state. This was done by choosing those places that had belonged to the same administrative unit during colonial rule, but had been transferred to different states in 1964 when a reorganization of states took place. Since language is a good proxy in these regions for cultural differences given the prevalence of caste and linguistic endogamy, language matching allows us to partially control for unobservable sociocultural differences.
A team of field investigators was sent out to these villages in 2003-06 timed to match the state-assigned schedule of holding gram sabhas. Team members, with the permission of the village panchayat, tape-recorded the entire proceedings and discussions at these meetings. Later, the responsible team member, who in each case was proficient in the local language and in English, transcribed the recordings verbatim into the local language and then translated that into These three codes were further classified into 'all women' and 'SHG women' categories depending on the identity of the speaker to understand the frequency with which certain types of issues were raised by SHG women, their narrative styles and quality of participation in comparison to all women (i.e. SHG affiliated and non-affiliated women). For women, whether they belonged to an SHG was usually discernable from their self-identification as a group member at some point during their speaking at the meeting. This type of group identification seemed to be the speech norm. Sometimes, the demands made by the women also revealed their group membership, as when they requested a building for their SHG group. We used this information to classify women as belonging to SHG groups. Another way to verify the presence or absence of SHG groups was through the speeches made by government and panchayat officials. If SHG groups were present, then there was a strong likelihood of these officials referring to these groups in their speeches. Given the speech norm of group identification, there is reasonable ground to take women who spoke up without ever identifying themselves as group members and made individual demands to be non-affiliated with SHGs. Therefore, in our writing we sometimes refer to 'non-SHG members'. However, since we cannot totally rule out the possibility that all women who did not self-identify as SHG members were truly not affiliated with SHGs, we have created the all-inclusive 'all women' coding category. We have followed this more conservative coding strategy that represents what our data can reveal more authentically. The qualitative coding software Nvivo was used for systematic coding.

FINDINGS
To begin with, there are some limited but important outcomes that can be generalized across all four states. Women identifiably associated with SHG groups employ a wider variety of narrative styles and utilize a more multilayered structure to convey their message compared to all women taken together. For example, in both Karnataka and Tamil Nadu (where women's oral participation is highest), SHG members use the descriptive narrative strategy with the highest frequency, often combining it with a command or an assertion of need and concluding with a request for action. The combined narrative strategies involve the presentation of context, description of the issue and the establishment of a link with a problem, which is concluded by formally requesting or commanding problem resolution. The use of this mixed-strategy is qualitatively richer than that frequently used by all women taken together. Non-SHG women use single narrative strategies and do not frame whatever they say as common issues. Non-SHG members use 'complaints' and 'queries' extensively, but in a mutually exclusively way. Overall it was found that non-SHG participants use a one-style strategy much more frequently than a multistyle strategy. This finding is congruent with our expectation that there might be some qualitative differences in participation between SHG women who have access to an "associational mechanism" (Sanyal 2014;2009) and are, therefore, more likely to have consistent opportunities of honing their oral competency in group meetings where they need to discuss and decide on loan allocation and repayment issues and present their case to fellow members and group leaders and to the program staffs.
The instances of oral participation by non-SHG women are more in numbers than that of identifiable SHG members. But this is due to the fact that, overall, among all female attendees, women belonging to SHGs are proportionately less. Additionally, there is the slight problem of not being able to identity all SHG members (due to the fact that some of them may not selfidentify by their group identity before speaking since there is no formal rule). Also, another important factor (emerging from Karnataka) associated with high levels of participation from non-SHG-affiliated women turned out to be their occupying positions of some authority and power, however small, and relatedly having designated responsibilities toward the community (these women were coded as 'de facto leaders'). These positions are mostly in the government sponsored rural daycare centers (anganwadi, balwadi), management of communal toilets, and in vocational training facilities and local health centers. The common denominator of these sectors is finance, administration, women and children, and these are reflected in the frequency table of issues raised for women coded as 'de facto leaders.' However, although greater in volume, the vocal participation of non-SHG women and those who were not de facto leaders or some kind of position-holder was often substantively confined to seeking a government benefit for their household.
Thus, thinking in terms of oral competency, the main difference between SHG and non-SHG women's participation (not considering de facto leaders) was in the quality of participation rather than in the numerical instances of oral participation. SHG members in their verbal statements and interjections presented the context for the problem, used a public goods framing, and showed their awareness that panchayat and government officials were accountable. These features meant that in our coding system, they were coded as displaying a higher quality of participation at a higher rate than non-SHG women's participation. Non-SHG women (who comprise the overwhelming majority of the 'all women' category) were divided almost equally between the high and low qualities of participation. In Karnataka 82.35% 13 of women from SHG backgrounds displayed a high quality of participation, while a smaller 51.44% of 'all women' were coded into the same category. In Tamil Nadu 59.82% of women from SHG backgrounds displayed a high quality of participation while a smaller 51.53% of 'all women' were coded into the same category. This fact denotes an increase in the quality of participation of SHG members relative to their non-SHG counterparts. In Kerala and Andhra Pradesh, it was not possible to meaningfully code the quality of participation because public participation was very low to nonexistent because of the contextual factors described before which curtailed free public discussion in gram sabhas.
While the finding regarding the quality of SHG women's participation matched our expectation, the finding that nearly half of verbal interjections by non-SHG women also displayed a high quality of participation was unanticipated. In explaining this pattern we must consider a number of factors that might have a positive influence on women's quality of participation independent of SHG membership. The literacy level is an important factor among them because we can expect literacy to make a positive difference to the quality of participation because of its role in making villagers more aware and articulate. From the village-level literacy data 14 , we see that villages from which gram sabhas were sampled in one district in Karnataka had uniformly high literacy levels and in the other district had largely medium literacy. This fact combined with the long history of the panchayat system and gram sabhas in this state and the vocal participation of de facto female leaders (female ward members; school headmistresses; female staffs of government departments, like the women the children dept.), who are a significant presence in gram sabhas in medium to high literacy villages in Karnataka, might explain why women not affiliated with SHGs would be able to participate in gram sabha discussions in a qualitatively superior way at least half of the times. In Tamil Nadu, the literacy level in villages from which gram sabhas were sampled in the two districts ranged from medium high to medium low. So here, too, literacy could be having a marginally positively influence on the quality of participation of non-SHG women. In the case of Tamil Nadu, with its long-running state sponsored SHG program (which would have completed roughly 16 years by 2006), there is also the strong possibility of a spillover effect of SHGs on the presence and participation of non-affiliated women in public sphere institutions. We must also remember that these South Indian states have lower levels of gender disparity, which means women are freer to associate and interact with others, which might give women greater facility with being articulate. It might also be the case that women have improved in their skill of presenting their arguments and demands in the course of the decade long period of attending gram sabhas.
Lastly, one of the unforeseen patterns to emerge from our data was the construction of de facto female leaders, i.e. women with no formally assigned leadership role who nonetheless rose above their individual or immediate group-based responsibilities, and used their oral competency to present, argue and influence gram sabha members on a particular issue of public relevance.
These women not only used a mixed variety of descriptive, command, complaint and request based narrative styles, they also often had the unintended consequence of motivating more women to follow up right after them -a pattern we call 'cluster talk' -increasing women's engagement with public goods issues. In Tamil Nadu and Karnataka, where gram sabhas are structured to allow public discussion, 12 out of 63 and 3 out of 29 women who were coded as de facto leaders, respectively, were also part of an SHG. The fact that most of these female de-facto leaders were not SHG members, indicates that women have the scope for gaining oral competency through other mechanisms, like an experience of employment and holding positions of some authority in their fields of work that may be related to serving the community (e.g., women who manage government run daycare centers for children).
In the following section we discuss the state-level effects in Tamil Nadu as an in-depth example since this emerged as the only state where there was a combined presence of a talkcentered gram sabha and a robust SHG program that had been running for a decade by the time of data collection (13-16 years in Dharmapuri and about 10 years in Coimbatore). 15

Issues raised
About half of the gram sabhas recorded in Tamil   Marking a greater difference with all women, our analysis revealed a consistent pattern of SHG women raising multiple issues at a time. In the excerpt below, we see an example of two SHG leaders (both schedule caste women and the only two women to speak up at the meeting) raising problems facing the community, speaking one after the other as a strategy of informal collaboration to emphasize their issues, and each of them raising more than one issue (separate issues raised by each speaker are numbered).
[ (iii) Employment opportunity should be provided for the ladies. (iv) Light (electricity) facilities should be done for us. (v) In the crematorium area, roads should be laid.

Mr. S. Karuppan, Panchayat President:
Village people's grievances with basic amenities like transportation, water, and also street light facilities were expressed by the women's self-help groups. And they insisted that we address the problems. For all these activities, if the government provides the sufficient funds, we will immediately take actions for addressing these problems…

Mr. S. Kathirivel, Vice President:
For laying those roads we have requested the MP and MLA.
They have given their word to us, and we have given the estimate for it.

Mr. S. Karuppan, Panchayat President:
There are 150 children who go to school. We have given an estimate for Rs.7,50,000 for laying roads. The MLA has told us that we will not be getting that much as our share. We will be allotted funds for laying only one kilometer, which is around Rs.5,00,000.
The first SHG leader raised three issues followed by the second SHG leader who raised five issues, two of which (italicized ones) were raised exclusively by her (not raised by men who spoke preceding these women). Of these two issues, one was specific to the group while the other was relevant to improving community hygiene. Note the panchayat president's affirmation of SHGs by making a special mention of their active participation in the gram sabha and his informative response as he explains the resource constraints and the action that has been taken.
Cluster talk is another important feature that emerged from our analysis of issues raised.
This pattern is observed across all the gram sabhas recorded in Tamil Nadu, and it is most prominent following the verbal interjection of a SHG President, who is a leader with perceived authority. This pattern raises the frequency of an issue being mentioned by women and is a likely explanation for the relatively high percentage of civil works issues mentioned by all women.
The following extract provides an example of cluster talk (separate issues raised by the speakers are numbered).
[ Next, the burial ground is not proper. The situation is such that the dead bodies have to be put in pits where water is stagnating. We need a proper place for burial.

President:
In the past, years ago, they used that place for tying horses... Now that place has gotten narrowed in many ways. Even then we will arrange to have a compound built for this place. I request all your cooperation to complete that work with the fund that we will be receiving within ten days. Fund of Rs.1lac is allotted for L.G.Pudhur…We have submitted a petition for the drinking water problem. When the officers come on a specific date you all must come and participate. We will go and do indefinite strike at the water board office. But there must be no disturbance to the public.

Sundrammal (OBC), SHG: (i) There is no burial ground in our village. (ii) No roads. (iii)
Drinking water is a big problem. All these must be provided to us. On behalf of the group we will give ten percent of the expense needed for this.
President: You say that you accept this and, not only that, you say that you will pay ten percent of the expenses! I give this work to you. So it will be like that you also got a work… Sundrammal SHG: (i) There is no proper water facility in our village, (ii) no road facility, (iii) no burial ground facility, (iv) no lights. We request the president to do all these quickly. (v) Water is stagnating and we can't walk. There is a terrible stench.

Padma (OBC) Barathi women's group:
There is no proper employment opportunity. I request the president to do some arrangement for that.

… man speaking
Sundrammal SHG: There is no way to go to the burial ground. We don't know what to do. You must clean the pathway for going to Kizhahumaan. Can't bear the smell of the ditch.

Ms.Amsaveni (OBC):
Though we mentioned our problems in the past sabhas they were not resolved.

Ms Subhashini (OBC):
Water is stagnating like a lake. There is no way to go (to the fields for open defecation) and if we relieve ourselves in gardens belonging to others, then they object to it.
For how many days can we go in others' garden. We have given application twice and thrice, but didn't get any solution. Can't bear the stench from the ditch. You must correct it. You haven't done anything even though diwali and pongal went by. You must lay roads for all the places.
You didn't lay roads for us though we too asked.

…men speaking
Sundrammal SHG: They (referring to ST households) take water by cutting the pipe. If we tell the Tahsildar about this he says that this is not our place, it is a public place. You have to correct it.

…ST man speaking
Padma (OBC) SHG: They take water by removing the pipes. We don't get water.

Ms. T. Amsaveni, (OBC): (vi)
The path we use to go between our village and the mill is very dark. There are no streetlights. We can't take the children. If this proceeds we will do a procession (in protest) on the road.

Sundarammal SHG:
There is no place to put the dead in the burial ground. It is full of water.
There is no latrine facility. If we go to other people's fields (to relieve ourselves), they scold us.
They also put the dead in the water. We will definitely do a road procession (in protest) on behalf of the women's club.

…men speaking
Ms. Jayamani (OBC): There are no roads in our village. No water and bus facilities. The streets are all dark. There are snakes. We can't bring children out in the nighttime. You must correct it.

Ms. Pushpa (OBC):
There is no water facility, no streetlights and no bus facility. The auto drivers don't want to come here (to transport passengers). We came here five years ago and, in this time, there have been no improvements. You must take immediate action for these.

Ms. Santhakumari (OBC):
There are no proper road facilities that would allow pregnant women to travel. You must build (water) tank and provide us water. Only one or two pots are available for each house.

Ms. Dhanalakshmi, (OBC):
I request you to provide water facility, road facility and power facility for us.

Padma SHG:
The water problem is severe. Only five pots are available everyday. Improve the water supply.
President: According to our accounts, there should be one pipe for every 10-15 houses in this Panchayat. But here, a connection has been drawn for 4-5 houses. Even then you are giving petitions. We are telling you lies that the collector didn't give order. You have to understand this.
So we (panchayat officials) suffer between the two sides. You say that water is not enough and the government asks why tax hasn't come through this? What is the alternate solution for this?
You tell me. You build new houses and then you say that you didn't get power connection immediately, but it is enough if you are given a pot of water. Later you ask for five pots of water and then ten pots. Tell me, what should be done for this? They (some people) take water by cutting the pipe. When we asked about that, you came to my home and made problems. So here onwards you must not take water by removing the pipe. If anyone does so, then we must solve the problem through police intervention. Secondly, we have built half in the Irular colony in M.G.R. Nagar. In that place the lands are above and the houses are below. The water dam has been completed partially. It will be completed quickly. At present the path to the burial ground is being paved in Sathya Nagar. Rs.25, 000 has been given to the tribal community for this. We thank the government for this.
In this extract, a woman who is an SHG leader raises four distinct issues dealing with the poor condition of public facilities, which are common concerns, in her opening speech. It is noteworthy that, despite there being double the number of men than women, she is the first member of the public to speak up at the meeting as soon as the panchayat president concludes a lengthy status report on various public works projects in the village. The vocalization of multiple public concerns all at once carves out a symbolic space in the gram sabha for SHG affiliated and non-affiliated women to repeatedly adumbrate these shared demands. Another important pattern that is observed frequently in TN gram sbahas, and one captured here, is the way in which nonaffiliated women begin to add their voice to the demands introduced by SHG women in a choruslike manner. In this excerpt three SHG members (Vijayalakshmi, Sundarammal, Padma), of whom one is particularly dominant, raise a set of issues, and they are subsequently joined by six non-affiliated women (Amsaveni, Subhashini, Jayamani, Pushpa, Santhakumari and Dhanalakshmi) who repeat the same demands, at times adding their own narrative of inconvenience. This pattern arguably leads to the higher focus on public-good issues that are captured in Table 1.0.
In speculating on the factors that generate this pattern of cluster talk, we can point to a few possibilities. One possibility could be that the close-ties that are fostered by SHG membership and association over shared interests (Sanyal 2014) manifest themselves in the clustered talk of SHG members who feel the need to vocally support their leaders in a public forum by repeating and emphasizing the issues and commands voiced by the SHG leader. In addition, it is possible that the vocal participation of an agent with power (SHG leaders) and the agent's utilization of their power by signaling their SHG affiliation can momentarily carve out a discursive space for other members of the group and for non-affiliated women, thereby allowing them an opportunity of honing their oral competency.

Narrative Style & Structure
The major difference between SHG and non-SHG women in terms of narrative structure and style is the simultaneous use of multiple styles by SHG women as opposed to using a single narrative style common among non-SHG women in communicating their problems. 19 While the 'descriptive' code (describing a problem, giving it context and inviting a response or discussion on the subject) is most popular amongst both categories of women, SHG women typically use it in conjunction with a command for resolution or an assertion of need, and bolstering these with a complaint about negligence or a reinforcing request for remedial. Among SHG women, the 'command' (arguably the most assertive style any citizen may use) is the next most frequently used style followed by 'need', and closely followed by equal use of 'complaint' and 'request'.
Unaffiliated women tend to either only describe an issue or follow up an earlier statement of problem by another speaker by lodging a one-line complaint or much less frequently by a command or a statement of need. Among these women complaint, command, need and request are the next most frequently used narrative styles, with complaints far exceeding (nearly double compared to utilization of other styles) commands, needs, and requests. The narrative structures of SHG and non-SHG women can be portrayed by these models (Figures   1 and 2).
The following excerpt illustrates a typical example where an SHG member raised multiple issues at once using a multi-style strategy. (Separate issues raised by the speaker in the first segment are numbered consecutively and marked alphabetically with 'a' for the first time an issue is mentioned. In the second segment, when the same woman expands on these issues with further details, they are marked with the same number and with 'b' or 'c'.) [Kattakaram, Bargur, Dharmapuri]

Mr. Kannan (BC) President:
Lists all the works he has done for the village and then invites the public to participate by declaring, "Anybody from ladies club can talk." for them to walk all the way and say that they won't come. Children also cannot go (to school).

Ms
In the evenings also buses are not plying properly. They said tar road has been sanctioned but till now it has not been done. Two months have gone by. (iii:b) Buses should be able to come at least twice, that is, in the morning while going to school and in the evening while returning. If they miss the bus they have to walk back. Since they take a long time to walk back they remain absent the next day. Again the same problem repeats itself the next day. For four to five days, buses are not plying properly. In case of emergencies it is very problematic. Some have bicycles, and they can use that to go places but most people depend on bus only. So they can't go to places further away. So we need bus facility definitely. That is very important. Or else school will be stopped in Kattakaram. The place will not be developed in any field. The panchayat will get a name if it does it. So, you have to take care of this. We also will co-operate. You yourself come directly to see. In today's position, you itself come and see it.
The The narrative style used by SHG women is effective in delivering greater information grounded in factual and contextual descriptions and packaging them in a manner normatively acceptable to panchayat leaders, i.e. framed as an issue of public inconvenience and lack of village development. By presenting multiple facts surrounding a problem coherently and supporting them with commands, statement of needs, requests and complaints, a participant is able to effectively communicate the urgency of a problem and its persistence over time, revealing a history of negligence. As a result, we see the observed effect of the authorities making a serious attempt to respond to the commands and complaints of SHG members, either by promising action or explaining why the problem remains unresolved (shortage of funds at the panchayat level is the most frequently cited reason in TN). SHG women's ability to employ a complex narrative structure in a male-dominated public forum (at least going by attendance) is of great significance because it is a public display of public-speaking skills and could potentially influence positively how men view women's effectiveness as participants in this civic forum.
We can speculate on the factors associated with the multilayered narrative style of SHG women, associations that can be rigorously explored by future quantitative and qualitative studies. It is possible that SHG leaders have some education, and this could make them effective communicators. In addition, getting habituated into the practice of public speaking at group meetings fosters the cultivation of their oral competency. As a result, these women are better able to fashion their arguments in a way that can influence public discussion and local government action. It is also possible that SHG women's narrative style may be a result of an induced mandate by SHGs regarding member participation. It is possible that SHG members have internally agreed to follow a narrative blueprint in these public meetings in order to achieve perceived narrative efficiency by presenting multiple issues at the same time. Alternatively, the institutional staffs (usually NGOs) that support SHGs (through training and 'capacity building') might be coaching SHG women to present their demands and problems in particular narrative styles. Concomitantly, the complex structure could be the assimilated result of greater information, deeper engagement with issues, and an organic deepening of women's ties with civil society via their SHG membership.
In the context of narrative styles a final important observation concerns the manner of introduction used by many SHG women. Many of these women announce their own and, importantly, their SHG's name before they start to speak about issues of concern to them. While on the surface this may appear to be a trivial stylistic routine followed by group members, it arguably denotes a strategic tactic to tap into the SHG identity. The construction of identity around a women's group and the public expression of such identity reveals the social importance of belonging to SHGs. The repeated demonstration of SHG affiliation through direct vocalization of membership in the gram sabha hints at the perceived social and political value of SHG membership for women. One deduction that can be made from this pattern is that selfidentification for SHG women extends to them greater respect, authority and legitimacy in the eyes of the public and the political leadership present at the gram sabha.
In contrast, non-SHG women, with their simpler narrative styles and lack of details are less effective in forcing a response from panchayat authorities. Non-SHG women receive little acknowledgment and cursory responses at best from authority figures. Consequently, it is difficult to make deductions on the impact of non-SHG women's participation at the gram sabha, apart from the normative changes resulting from women participating in the first place.
But this, by itself, is significant because it arguably represents women gaining oral competency and being able to move beyond formal inclusion to substantively participate in political institutions and development planning. This does not mean, however, that the goal of political inclusion under oral democracy has been accomplished. There is much room for improvement.

Quality of participation
To analyze the quality of women's vocal participation in the gram sabha, we developed a binary code to differentiate the participation of women into 'high' and 'low' categories. In every event of a woman speaking 20 , we looked at the following factors to make an assessment on the quality of a woman's participation: a) whether the vocal interjection exceeded or was limited to a single sentence statement; b) whether the participant engaged in the issue deeply enough to substantiate a problem or need with appropriate reasoning including a reasonable amount of description and context to the problem or whether she only made a simple statement (a command, complaint, demand or request); and c) whether the participant demanded or prompted a discussion on an issue or made a statement that did not lead to further discussion on the topic. While these categories are difficult to delimit using exact boundaries, a consistent interpretation of women's participation based on these factors allowed us to form a general assessment of the quality of women's participation. This difference in the overall quality of participation is a cumulative function of the issues brought up at the gram sabha and the narrative style employed to communicate the issues.
The focus on common community concerns, which generally leads to a discussion and prompts a response from panchayat officials, and the utilization of a complex narrative structure using multiple styles combine to give SHG member's participation a qualitatively deeper and richer character than that of non-SHG women.
The 'coding density' analytic on NVIVO helped support our finding regarding the higher quality of SHG member's participation at the gram sabha. Coding density is determined by the total number of codes utilized in analyzing a particular event of participation. The greater the number of codes that are relevant to and utilized for analyzing an event of participation the greater the coding density. An increase in coding density in our data is explained most by the number of issues raised and the styles utilized within a particular event or excerpt of participation. Simply put, this means that SHG women, per segment of coded participation, raised a greater number of issues and used a greater number of narrative styles than their non SHG counterparts. The high coding density associated with SHG women's events of participation is an objective corroboration of our subjective assessment of their higher quality of participation at the gram sabha. However, women in general are able to meet the demanding bar of high quality participation at least half of the times they verbally participate. This is an encouraging finding with respect to women's inclusion into civic discussions and their substantive political inclusion.

CONCLUSIONS
This paper began with the aim of re-theorizing 'deliberative democracy' into oral democracy to gain new analytical purchase on talk-centered democracy, with a special interest in investigating women's oral participation in the gram sabha, and any potential impact that associational membership in SHGs might be having on it. Through in-depth analysis, we hope to have demonstrated what a talk-centered analysis of participatory, oral democracy would look like and why a talk-centered analysis is important. Talking in civic forums is a form of public speaking, and it is essential to acknowledge it as an oral or oratory competency. To a large extent, this competency is developed through consistent access and repeated participation in venues where individuals are required to vocalize their opinions and arguments, and make and defend their claims and do so in the presence of others, even in the presence of multitudes of people and visà-vis state officials and elected political leaders. This is not an intrinsic quality, or an endowment, that is randomly distributed across individuals in society, but possibly influenced by a plethora of factors, including class, caste, gender, education, associational exposure, and broader contextual factors like gender (dis)parity and political regime. While a quantitative study can better adjudicate between these factors and examine which of these are strongly associated with oral competency, in this first-ever qualitative study on this topic we have analyzed women's oratory competency in the gram sabha and additionally investigated whether women's access to SHGs, which have opened up through the 1990s, makes any difference in how women are able to verbally express their problems and needs in these forums and talk to the state.
Analysis shows that although the formal institutional structure of participatory, oral democracy was the same across all the states in India, there was significant heterogeneity in the practice of actually existing oral democracy because of state-level differences in political context and emphasis. Where talk-centered democracy actually functioned in a way to foster talk, significant differences were found in women's oral participation and in SHG women's oral competency. These differences were likely associated with differences in literacy levels (particularly where SHGs were weak and new) and in the length of time and robustness of the SHG program. Keeping these differences aside momentarily, the general finding is that, compared to all women considered together, SHG women have a more complex narrative style that combines a wide variety of narrative styles in the same speech, and they have a higher quality of participation. These contribute to keeping the pressure of responsiveness and accountability on local governments. All women considered together lagged behind SHG women in their quality of participation, however, they did not lag as far behind as we had anticipated.
This undermined our original expectation of a stark difference in oral competency between SHG and non-SHG women. This is not to say that associating in SHGs does not have enough of an impact on women's oral competency, rather, we should entertain the possibility that longrunning presence of SHGs might be having a positive spillover effect on non-affiliated women (through the peer-effect of observing SHG women talk in these forums) and may consequently raise their oral competency.
We would like to tie the analysis and arguments presented in this paper into discussions on deepening of democracy (Moore 1966;Heller 2000). Shifting our focus from the "making of democracy" to the "deepening of democracy" (Heller 2000, 487)  In analyzing the democracy literature, Heller has noted that the first generation of scholars studying democratic transition highlighted the role of elite political actors, and they were followed by a second generation of scholars who focused on historical explanations, where the role of the subordinated classes gained analytical precedence. 21 These studies pioneered a focus on working classes, their mobilizational tactics in pushing for democracy, and on labor movements around the world (in South America, Southern Europe, South Africa, Brazil). 22 Building on this second-wave focus on the role of subordinated classes, Heller has called for an analytical focus on conceptualizing "deepening democracy" (2000,487). Accordingly, he has astutely argued that we need to "disaggregate democracy" and "look beyond the macroinstitutional level of parliaments, constitutions, and elections, and we must investigate instead the intermediate-and local-level institutions and consultative arenas located in the interstices of state and society where "everyday" forms of democracy either flourish or founder (Heller 2000, 488)." Within this framework, prominence is given to analyzing the strength and vibrancy of civil society, including the social patterns of associationism (social movements, unions, collectives, etc.), as this is seen as a crucial component of an effective democracy because it can foster a political culture of providing public input into governance and demanding accountability from governments (Heller 2000).
Our analysis takes the focus on deepening democracy even deeper by shifting the focus to the microinstitutional level of participatory or direct democracy and by analyzing patterns of talk within it. This is the level at which the publics predictably engage with the state and their citizen peers, provide direct input into governance and keep on the pressure of accountability. This is the level at which we can observe democracy directly at work and observe the political culture in action. Examining the issues citizens raise and the manner in which they talk to the state can tell us something important about the relationship between citizens and the state, and deepening democracy should be about making that relationship more egalitarian for all groups in society, making the state accountable to the publics, and more responsive to citizens needs and suggestions. Analyzing talk and citizens' oral competency in making claims, complaints and commands of the state can help us evaluate how egalitarian that relationship is and observe the relationship under transformation as citizens gain more oral competency through various pathways. In furthering a "political sociology of democracy" (Heller 2000, 488) the next wave of studies should focus on institutions of actually existing oral democracy, which may have been originally established with the aim of promoting the utopian ideals of deliberative democracy or to inaugurate participatory planning, and engage in a talk-centered analysis that can reveal the extent to which citizens, particularly subordinated groups, can take advantage of these institutions and their relationship with the state. 13 For Karnataka, this includes women identified by the field observer in the transcript as SHG members and women who were not identified in the transcript as such (because of variation in field observer transcribing strategy) but demonstrated direct affiliation to SHGs by mentioning their SHG or SHG-related demand in the content of their speech (coded as 'SHG mention'). We can safely assume that women mentioning their SHG or a related demand (like a building or finances for the group) belong to SHGs.
14 The literacy data have been computed from the 2001 census of the Govt. of India (GOI). It relies on the government definition of literacy, which is simply the ability to write one's own name. It is a very basic measure and quite possibly inflated. 15 In Karnataka, the Stree Shakti program at the time was relatively weaker due to weak implementation in the first three to five years of the program. 16 In some of the fifty-two villages in TN, gram sabhas in two different cycles were observed and recorded, such that we have eighty-nine gram sabhas in the data set.
17 Physical infrastructure for community use such as road repairs, installation of streetlights, water pipelines & borewells, construction of public-use buildings and sanitation facilities 18 The percentages report the frequency with which an issue was raised as a percent of all issues raised by women in Tamil Nadu. 19 A key observation for most women in Tamil Nadu is that a problem is rarely structured in individualized terms using singular pronouns of 'I,' and 'me,' but done so using plural pronouns of 'we' and 'us' when referring to those affected by a problem. This statewide observation, encompassing both SHG and non-SHG members, indicates the standard norm is to speak in collective rather than individualistic terms. This could be an effect of the more than two decade long SHG movement or a predisposition due to the cultural speech norms of the region.