Authority and Deliberative Moments : Assessing Equality and Inequality in Deeply Divided Groups

The notion of equality is central to public deliberation, but few researches have examined how participants construct interactions in face-to-face group discussion involving unequal conditions of authority. This study analyses discussion between slum residents and police officers in Brazil, focusing on both reciprocal and hierarchical relationships in the flow of deliberation. It contributes to explain that the expression of authority is far from straightforward. Looking at a range of authority sources (expertise, functional position, tradition, life experience) that serve to situate and re-situate participants in relation to each other in discussion dynamics helps clarifying what goes on in deliberative moments. Findings reveal that personal experiences prevail in deliberative moments whereas functional credentials predominate in non-deliberative ones. Yet, the case demonstrates that functional authority is not necessarily dominative but can be combined with certain behaviors (such as empathetic imagination, search for commonalities and self-criticism) that lead to reciprocal interactions. This study provides important insights for organizing deliberation more effectively in contexts of fear, mistrust and resentment. Author Biography Rousiley C. M. Maia, Ph.D. in Political Science (University of Nottingham, UK), is a Professor of Political Communication in the Federal University of Minas Gerais, Brazil. She is the author of Deliberation across Deeply Divided Societies (with J. Steiner, M. C. Jaramillo, and S. Mameli, Cambridge University Press, 2017), Recognition and the Media (2014, Palgrave McMillan), Deliberation, the Media and Political Talk (2012, Hampton Press). Some of her recent publications have appeared in European Political Science Review, Political Studies, Journal of Computer-mediated Communication, Journal of Communication, Representation, Journal of Political Power, Journal of Public Deliberation, and in several Brazilian journals. Danila G. R. Cal, Ph.D. in Communication Studies (Federal University of Minas Gerais, Brazil), is an Assistant Professor at the Federal University of Pará. Her master’s thesis, which focused on public discourses regarding child domestic workers, received several awards in 2008. Her Ph.D. dissertation was awarded the Compós Prize, National Association of Graduate Programs in Communication in 2014. She is an associate research in the Research Group EME/UFMG. She is the author of the book Comunicação e Trabalho Infantil Doméstico: política, poder, resistências and her recent publications have appeared in Journal of Political Power, Galáxia, Fronteiras, E-Compós. Janine K. R. Bargas is a Ph.D. student in Communication Studies the Federal University of Minas Gerais, Brazil. She has experience in projects about traditional communities and human rights. She is an associate research in the Research Group EME/UFMG. Her dissertation is about digital media and political organization of Quilombolas communities. Some of her publications have appeared in Bulletin of Human Sciences of the Museu Paraense Emílio Goeldi and Logos. This article is available in Journal of Public Deliberation: https://www.publicdeliberation.net/jpd/vol13/iss2/art7 Vanessa V. Oliveira is a Ph.D. Candidate in Communication Studies at the Federal University of Minas Gerais, Brazil. She was a Fulbright visiting scholar at the University of North Carolina at Chapel Hill, U.S., under the supervision of Jürg Steiner. She was a junior researcher at National Truth Commision of Brazil. Her master thesis on the media controversy regarding the human rights policy in Brazil received an award from the Brazilian Association of Political Communication (Compolítica). She is an associate research in the Research Group EME/UFMG. Some of her recent publications have appeared in Information Journal of Communication , Galáxia, Compolítica. Patricia G. C. Rossini, Ph.D. in Communication Studies (Federal University of Minas Gerais, Brazil), is a postdoctoral research fellow in the School of Information Studies at Syracuse University, USA, where she collaborates with the Center for Computational and Data Sciences. Patricia is also an associate researcher in the Media and Public Sphere Research Group (EME/UFMG). Her work has appeared in books such as the Handbook of Research on Advanced ICT Integration for Governance and Policy Modeling (2014), A Digital Janus (2014); Navigating Cybercultures (2013), International Journal of Communication, and in several Brazilian journals. Rafael C. Sampaio, Ph.D. in Communication Studies, is an Assistant Professor at the Department of Political Science in the Federal University of Paraná. He is an associate research in the Research Group EME/UFMG. His writings have appeared in the Journal of Community Informatics, Journal of Latin American Communication Research, Brazilian Political Science Review, and in several Brazilian journals.


Introduction
The literature on informal talk, political discussion and deliberation is growing at a rapid pace (Conover & Searing, 2005;Maia, 2012Maia, , 2017;;Marques & Maia, 2010;Moy & Gastil, 2006;Walsh, 2004Walsh, , 2007;;Wyatt, Katz, & Kim, 2000).Conscious of this tendency, researchers are now quite cautious about specifying the features of different group-affiliations and the social conditions and circumstances enabling argumentative discussion (Black, 2008;Grölund, Bächtiger, & Setälä, 2014;Steiner, 2012;Wojcieszak & Mutz, 2009).The topic of authority is central to political theory.However, it is rarely operationalized in empirical research.Although important studies have looked at the dynamics of face-to-face group discussions involving unequal conditions (Gerber, 2015;Karpowitz & Mendelberg, 2014;Mendelberg, Karpowitz, & Oliphant, 2014;Pedrini, Bächtiger, & Steenbergen, 2013;Steiner, 2012;Steiner, Jaramillo, Maia, & Mameli, 2017;Walsh, 2007), these studies have not yet examined moments when conversation is shaped by hierarchical relationships.How authority is enacted in the flow of conversation remains poorly understood; and our knowledge of the sources of authority that constrain or enable dialogue is limited.This paper contributes to filling this gap.Contrary to scholars who assume that authentic deliberation cannot be established between members of highly uneven authority, we argue that "moments" of constructive dialogue are possible and productive for deliberation.By focusing on intergroup communication between slum-dwellers and police officers in Brazil, we examine what sources of authority are mobilized in both deliberative and non-deliberative moments and how participants unequal in power can construct reciprocal relationships.To develop this study, we conducted six discussion groups with slums inhabitants and police officers in Brazil.This study applies the concept of "Deliberative Transformative Moments" (DTM) (Jaramillo & Steiner, 2014;Steiner et al., 2017), designed to assess how certain elements affect the flow of discussion.This analysis contributes to explaining the complexity of the notion of equality in the dynamics of face-to-face group discussions.Against a static view of power relationships, our analysis clarifies a range of authority sources that serve to dynamically shape relationships in conversational contexts.Our findings reveal that sources of authority based on life experiences predominate when deliberation is at a high level, whereas functional credentials prevail when deliberation is low.We suggest that the role of authority on deliberation is far from straightforward; functional authority is not necessarily dominative or coercive, and it can be combined with behaviors (such as empathetic understanding, search for commonalities, and self-criticism) that lead to reciprocal interactions.This article is organized in the following way.First, we briefly review the literature on inequality in inter-group deliberation.Next, we explain why discussion between police officers and slum residents in Brazil constitutes a particularly interesting case for studying authority in deliberation.In the third section, we present our methodology.In the remaining sections, we discuss our findings and provide examples to illustrate how powerful participants either enlarge or restrict communicative space among speakers.Implications of this study for moving empirical research further away from simplistic notions of equality are discussed in the conclusion.

The problem of equality in group discussion across deep divisions
Deliberation revolves around the controversial notion of equality; and scholars have different views on what type of equality is most conducive to deliberation (Karpowitz & Raphael, 2016;Knight & Johnson, 1997).Classical deliberative democrats stress the importance of meeting the criterion of political equality as a moral principle of mutually treating one another as equal arguers.According to Cohen (1997, p. 73), "the members recognize one another as having deliberative capacities, i.e., the capacities required for entering into a public exchange of reasons and for acting on the result of such public reasoning."In articulating and defending their views, participants should rely not on asymmetries created, for example, by socioeconomic resources, political power or social status, but only on what Habermas (1996, p. 305) calls "the force of the better argument." Habermas has been repeatedly criticized for assuming that power can be separated from public discourse.The most common criticism is that Habermas' "ideal speech situation," which is abstracted from inequalities of social power, masks exclusion and domination (Connolly, 1991;Montag, 2000;Williams, 2000;Young, 2002).A central objection is not just that power and ideological domination cannot be resolved through rational discussion, but economic dependence, political inequalities and social status asymmetries block reflexive communication.Skeptics emphasize that these differences-defined along the lines of gender, race, class, ethnicity, sexual preference, and so on-prevent people from being equal speakers and deliberators, making it impossible to establish authentically impartial dialogue between powerful and oppressed groups (Montag, 2000).
Those who advocate that deliberation is possible across unequal conditions or deep divisions have yet to explain how authority can be overcome in practical processes of dialogue.In this paper, we are interested in observing how authority intersects communication and deliberation.Based on previous studies on face-toface group discussion, we assume there are "moments of deliberation" in the flux of conversation (Laden, 2012;Steiner, 2012;Steiner et al., 2017), and in divided groups "confrontation and communication can be symbiotic" (Ugarriza & Caluwaerts, 2014, p. 207).We assume that members of disadvantaged groups with less authority and status in society are likely to speak less and receive less attention than members of powerful groups, and, consequently, their concerns tend to be undervalued (Karpowitz & Mendelberg, 2014;Mansbridge, 1983;Williams, 2000;Young, 2002).We further expect that the lack of respect in the eyes of those with authority is likely to generate negative expectations (such as low self-respect and low self-esteem) in people with low authority, who suffer continued exclusion or oppression (Honneth, 1996(Honneth, , 2001;;Maia, 2012Maia, , 2014;;Mendelberg, 2002;Mendelberg et al., 2014).
We seek to unpack what goes on in moments when the interlocutors' attunement is based on reciprocal as well as on hierarchical relationships (Laden, 2012).We define moments of reciprocity when there is an equal possibility of communicative exchange among group discussants; and they are open to the views and experiences of others, considering them to be entitled to accept and reject claims.We conceive hierarchical relationship when participants direct, command or license others; emphasize social distinctions and authority to ascend over the others, and determine unilaterally the space of conversation.Our study illustrates a special case of deep division.As in other divided groups, poor community residents and the police demonstrate mutual distrust and intolerance, while sharing a story of confrontation and violence.However, one group is characterized by high authority and the other by profound vulnerability.
Authority-usually defined as expectation to influence-is a complex and elusive concept.We endorse the view that authority is relational and it carries an inherent ambiguity (Laden, 2012;Parsons, 1963).On the one hand, the probability of those invested with authority to exert influence is always present, even if this property is not explicitly exercised.On the other hand, authority is inseparable from the interlocutor's interpretation of the interaction.While some forms of authority remain oppressive and coercive, other forms can be nondominative and compatible with freedom.In this last situation, individuals do not obey, but are free to accept or reject claims based on reflective considerations (Habermas, 1996;Laden, 2012;Warren, 1996).According to Habermas (1996, p. 119), "communicative freedom exists only between actors who, adopting a performative attitude, want to reach an understanding with one another about something they expect one another to take positions on reciprocally raised validity claims." To explore this issue, we follow Laden's distinction between authority of command and authority of connection.He contends that "authority of command" is backward-looking, since credentials, i.e., "the condition for being invested with authority" (Laden, 2012, p. 57) are given a priori.In other words, the credentials derive from an existing social structure and therefore must exist prior to one's capacity to command: Whatever it is that establishes them [the credentials] must be prior to the exercise of the capacity they authorize: do we already stand in a relationship that gives me the authority, the superior position, to command you?One reason the source of my authority to command must be backward-looking is that only then can the question of my authority remain properly normative, by being independent of its being effective.(Laden, 2012, p. 68) In contrast, authority of connection is forward-looking; and the final status of this type of authority cannot be established before one has spoken.This kind of authority depends on the conversation partner acknowledging the other's right or capacity to issue a claim in a symmetrical way and also acknowledging the statement as something they would say as well.In Laden's words: My normative capacity can at times remain indeterminate until you either acknowledge or refuse what I said.My authority to speak this way at least here and now, need not be something that is established ahead of time, but can be constituted, as it were, after the fact.Thus, it must at least be possible for the credentials that ground the authority of connection to be forward-looking.(Laden, 2012, p. 68-69) Since authority is essentially related to positions and roles, we find it productive to assess participants' credentials in our research.As a speaker may mobilize different sources of authority during discussion, such as expertise, functional position, tradition, life experience, and so forth (Saward, 2009), it is unclear how reciprocal and hierarchical relationships are produced.
With these considerations in mind, we ask the following research questions: 1) Is it possible to observe deliberative moments in discussions among members of deeply divided groups?
2) What sources of authority are mobilized in different moments of deliberation?
3) How do group participants build both reciprocal and hierarchical relationships?
4) How do group participants react to expression of authority?

Case Study
The very idea of deliberation between slum residents and the police in Brazil is hard to attain due to a history of atrocities and conflicts that creates power inequality in several ways.Poor community residents are subject of systematic human rights violation and abusive use of force.The rhetoric of "war on crime" to combat drug gangs and illegal arms trade, while resonating with the fears of elites and middle classes, has the result of criminalizing residents of slums.These individuals are typically seen as "marginal" and "dangerous," as a threat to society.In large Brazilian cities like Belo Horizonte and Belém where discussion groups for this study were held, the police are more prone to use abusive force and shoot against people living in poor neighborhoods in comparison to other localities in the cities (Zaluar & Barcellos, 2013) 1 .Violent police incursions into slums are usually the cause of death of many victims and of psychological 1 In Rio de Janeiro, the police shoot ten times more in poor neighborhoods than in other city locations; and they are two times more prone to use force against slum residents (Zaluar & Barcellos, 2013).
traumas.In addition, some police officers are members of militias, which are eager to profit from illegal trade.Studies in this field indicate that several governmental programs (such as initiatives based on communitarian police models) implemented in the last two decades either failed or were interrupted due to criminal networks that bring together illegal organizations, empowered community leaders and state agents (Arias, 2006(Arias, , 2013)).This ongoing strife has generated mistrust and fear between poor community residents and the local police-a context that can be considered far from ideal for conversation and deliberation (Dryzek, 2005;O'Flynn, 2007).Thus, this case poses a specific test for examining whether deliberation can occur among members of these groups; and opens up practical possibilities for observing how authority directly intersects dialogue (and deliberation).

Group discussion procedures
To develop this exploratory study, we organized six discussion groups-three groups in Belém (Para) and three in Belo Horizonte (Minas Gerais).These cities are located in the north and southeast of the country, respectively.To contact participants, we had the support of the social projects "Rede Escola Cidadã" in Belém and "Fica Vivo" in Belo Horizonte, as well as of community leaders and staff of the local police.We told participants that they would discuss the relationship between slum residents and the police.Altogether, 76 people participated in the discussion groups-46 males and 30 females.The groups had a minimum of 10 and a maximum of 17 members; and were composed by 1/3 of police officers and 2/3 of community residents, approximately.All police officers wore uniforms, as they were allowed to participate in our group discussion during working time.Since all police officers were lower ranks (with the exception of one lieutenant), we decided to observe only the authority relationship between these subjects and the slum residents.While we are aware that gender is important in shaping deliberative discussions, sorting out implications of gender roles would add greater complexity to the theoretical and analytical framework of this study.Differently from other researches on deliberation, such as Deliberative Polling (Fishkin & Luskin, 2005), participants did not receive briefing material, and the moderator did not intervene to encourage deliberative engagement.The moderator played a passive role to enable group discussants to freely interact with each other and speak their minds without external incentives.Before and after discussions, we applied questionnaires about socio-demographic, political and psychological issues to all participants.
To facilitate conversation, participants were asked to draw a picture and were then invited to present themselves and their drawings.To start the discussion, the moderator asked: "How to create a culture of peace between poor community residents and the local police?"If conversation came prematurely to an end, the moderator only repeated the aforementioned question or re-introduced the last topic of dialogue into the group.After roughly one hour, they had a quick coffee break and then were invited to discuss the minimum age of criminal responsibility in Brazil.Discussion lasted for two-and-a-half hours, on average.In this paper, we examine only the first part of the discussion.We used pseudonyms for participants following research ethical norms.

Transcription and coding
Our analysis is based on the transcriptions of the aforementioned topic from all discussion groups (164 pages; 48,079 words).We imported the text into NVivo qualitative data analysis software (version 10) for coding and analysis.The unit of analysis is the speech act-every time a participant spoke, his/her speech counted as one speech act, even in case of brief interruptions.Altogether, there were 334 speech acts.The content analysis was processed by NVivo's version 10.A researcher and a doctoral student carried out the coding.The coders underwent training sessions for approximately four weeks.To produce reliability tests, each person independently coded 60 speech acts (18% of the overall speech acts) chosen from the six discussion groups (in equal proportion in each group).The coders then met to discuss discrepancies in coding until reaching desirable agreement percentages (minimum of 96% pairwise agreement).This percentage was calculated through NVivo Comparison Query Coding.
To analyze the discussion flow, we employed the concept of Deliberative Transformative Moments (DTM) (Jaramillo & Steiner, 2014;Steiner et al., 2017).This methodology, by focusing on forms of communication and types of behaviors that can affect the quality of deliberation, was designed to assess smallgroup discussions, and it has a less demanding approach to fulfill deliberative norms in comparison to the Deliberative Quality Index (DQI).Following several scholars (Black, 2008;Polletta, 2007;Polletta & Lee, 2006), the creators of DTM consider that deliberation should encompass broader forms of communication, including storytelling and humor.Each speech act is classified according to one of the following four categories: 1) The speech act stays at a high level of deliberation: This category is used when the preceding speech act was at a high level of deliberation and the current speech act continues at this level.The coding of the current speech act is easiest if it fulfills all the criteria of good deliberation (Steiner, 2012) 2 .However, deliberation can still remain at a high level even if speakers do not fulfill all these criteria, and discussion nevertheless continues to flow in an interactive way on a particular topic, with the actors listening to each other with respect.Deliberation also stays high when an actor introduces another topic, giving reasons why the topic is linked with the issue assigned to the group.
2) The speech act transforms the level of deliberation from high to low: This second category is used when a speech act disrupts the flow of the discussion.The topic discussed so far is no longer pursued; and it may also include speech acts that are incoherent or confusing.Under these circumstances, it is not easy for the other participants to continue the discussion in a meaningful way.
3) The speech act stays at a low level of deliberation: This third category is used if the preceding speech act was at a low level of deliberation and the current speech act stays at this level.It means that participants are not able or interested in resuming the discussion.In our case, this would mean, for example, that the speaker is unable or unwilling to put on the agenda a topic relevant for the discussion of building a culture of peace.Instead, the speaker brings up unrelated topics or stories or presents confusing or incoherent statements.
4) The speech act transforms the level of deliberation from low to high: This fourth category is used if the preceding utterance was at a low level of deliberation and the current speech act brings the discussion to a high level, meaning that participants are successful in adding new aspects to the discussion or to formulate a new topic that is relevant for the discussion.
DTM is a qualitative methodological approach that seeks to tap into the flow of discussion, and it requires a qualitative analysis that is context-sensitive.Since a speech act is analyzed as a whole in relation to a previous utterance, it is not possible to identify particular deliberative dimensions (as prescribed by DQI, for instance).The aim here is to understand how the flow of communication is or is not disrupted by off-topic discussions, disrespectful remarks, irony or sarcasm, stories, spoiling leaders, and confusing statements.
Although the DTM does not provide means for scrutinizing the deliberative qualities of each isolated speech act, this method enables researchers to look at discussions as a whole.Insofar as these limitations are accounted for, we believe that the DTM presents a valuable contribution to analyze the interpersonal dynamics in small-group discussions and can help scholars to identify the types of behavior that can disrupt or enable a deliberative discussion.
Sources of Authority.To further understand the impact of authority on group discussions, we observed the type of credentials that underpin the speech acts.To operationalize this concept, we track different sources of authority as formulated by Saward (2009) (described in 1, 2, 3), and we add two other categories for research purposes (described in 4, 5).We propose that: a) backward-looking credentials have the following origins: 1) Expertise and special credentials, i.e., the possession of an amount of knowledge in a specific field, which is significantly greater than that of ordinary people.Such distinctive knowledge and insights are not necessarily connected to academic titles.Examples are: "I'm in the last semester of Law School and I studied this issue"; "I'm also a teacher; I teach in the slum"; "As I work with statistics in the police, I have this information." 2) Deep roots and tradition, referring to longstanding group identity and experiences, historically repeated and culturally re-strengthened.The following utterances can be considered as examples: "I'm the fourth generation of this community, so this problem has been here for almost sixty years.""When this community started out, it was full of life." 3) Broader interests and wide popular support, when the speaker claims to represent and stand for the will of many ("the force of numbers").For example, "When I represent the community…" 4) Functional/legal role, referring to a position held or a function carried out in an institution-in our research, authority in the police officers' speech acts was frequently aroused from their functional role.Consider the following examples: "The police is very well trained"; "I'm a police officer and I'm aware of that"; "The police study the neighborhood where they are located."5) Life experiences, referring to personal knowledge obtained by something that one has lived through; reporting one's experience, situation, and personal history (like what one has experienced) to validate one's claim (Adams, 2014; see also Black, 2008).Examples are: "As a community resident, who has suffered several episodes of police violence, I know that situation"; "Based on my experience, I can say that most police officers are disrespectful and racists"; "I can tell you about the place where I live and work." When authority based on backward-looking credentials is enacted, we coded that it can be: i) accepted, ii) contested, and iii) ignored.
b) Authority of connection, as already pointed out, "need not be something established ahead of time, but can be constituted, as it were, after the fact" (Laden, 2012, p. 69).It was coded when the speech act was endorsed in reference to a sense of community (i.e., to an "us" as something that both could say together) (Laden, 2012).Examples are: "As the lady said, what we want here is education.(…) Education is like the foundation of a house"; "So I think that is through education, also like he said."

Results
The discussion was characterized by an equitable distribution of speaking time between the two groups of participants.The community residents had a slightly higher share in the total speaking time (57%, 2 hours, 54 minutes) than the police officers (43%, 2 hours, 12 minutes).Proportionally, however, the policy officers were more active speakers.Since our discussion groups were composed by 1/3 of police officers, these individuals were responsible for 47% of the speech acts (more than their fair share-33%) while the slum inhabitants issued 53% of the speech acts (less than their fair share-66%).Most of the time, there was an appropriate length of speakers' contributions, and no group dominated the general discussion.
The first research question asked if it was possible to find deliberative moments during discussion in groups unequal in power.To answer this question, we examined different deliberative moments during discussion, as captured by DTM categories.A high deliberative level (DTM 1) corresponded to 58% of total speech acts.If moments when deliberation changed from low to high are accounted for (DTM 4: 7.7%), this percentage rose to 65.7%.The level of deliberation was low in 34.2% of the speech acts (DTM 2: 8.3%; DTM3: 25.9%).
Despite the imbalance of authority between group discussants, these results suggest that members of divided groups were frequently able to engage in deliberative interactions.If deliberative dialogue is welcome even between groups unequal in power, we should further examine what enables speakers to shape the environment of discussion in a mutual way.
Our second research question inquired into the sources of authority mobilized by each category of participants in different DTMs. Figure 1 shows that prior-tospeech credentials (backward-looking) prevailed at all conversation levels.Backward-looking credentials and authority sources may be tradition, professional functions, expertise, and experience.We identified very few occurrences of authority linked to "expertise," "wider interests," and "tradition" in the investigated material; and, thus, for clarity purposes, we did not include them in any of our Figures (they appeared in 1% of the speech acts analyzed).

Figure 1 -Type of credential in each type of DTM (%)
Note: FL stands for forward-looking and BL for backward-looking credentials.
To duly understand the relation between authority and the levels of deliberation, we observed the credentials behind the speech acts in each DTM. Figure 1 also shows a significant variation in the sources of legitimacy put forward by these categories of speakers in different moments of discussion.Forward-looking authority and authority based on experience prevailed when deliberation was at a high level-DTM1 (15.2% forward-looking, 14.2% experience, 8.9% experience combined with functional vs 12.7% functional).Conversely, functional authority is mainly present in DTM3, i.e., when deliberation was at a low level (23.8% functional vs 19.5% experience, 3.6% experience combined with functional).We used the Fisher's Exact Test3 and the result was a p-value of 0.01499, which indicates that there is a significant relationship between type of credential and deliberative moments, with life experiences predominating when deliberation is at a high level and functional credentials prevailing when deliberation is low.
The next step consisted of identifying the credentials behind the speech acts of each group.Figure 2 reveals that police officers issued considerably more claims with some kind of authority (63.8% out of this actor's total speech acts) in comparison with community residents (38.7%).As expected, the police officers backed up their claims more frequently on functional authority (32.5% functional vs 6.1% experience).Interestingly, we found that a noteworthy share of speech acts in which authority derived from the functional role was accompanied by authority of experiences these subjects have lived through (17.8%).The community residents based their propositions chiefly on personal experiences (18.6%) and forward-looking authority (18.1%).It is also worth noting that, as Figure 3 shows, when we move from low to high deliberation (DTM3 → DTM 1), there is a decrease of functional authority standing alone (40.8% → 26.6%) and an increase of statements that included both functional and experience sources of authority (6.1% → 20%) by police officers.To answer our third question, we will describe reciprocal and hierarchical relationships in the next section.
The empirical examples will help to explain the aforementioned nuances and how functional authority can be weakened in practical processes of dialogue.Note: FL stands for forward-looking and BL for backward-looking credentials.
Our fourth question pertains to the participants' reactions to the expression of authority.Our results displayed in Figure 4 reveal that distinct types of authority provoke different reactions in group discussion.Whereas claims backed up by personal experiences are typically accepted (88.9% accepted, 11.1% contested, none ignored), claims based on functional role stimulated more diversified reactions (61.1% accepted, 30.5% rejected, 8.3% ignored).Hybrid claims encompassing both experience and functional authority also incited varied responses (54.1% accepted, 37.5% rejected, 8.3% ignored).However, our analysis of participants' responses to authority taking into consideration distinct DTMs seems inconclusive.It should be noted that the distributed occurrence of these data became very low.In this respect, we could not find a strong co-relation as distribution was displayed rather randomly.

Figure 4 -Kind of reaction toward each type of authority credential (%)
Note: BL stands for backward-looking credentials.

Deliberative moments across deep divisions
Broadly speaking, our findings suggest that members of divided groups were able and willing to engage in deliberation.It may come as a surprise that deliberative moments prevailed.Given the fact that police officers hold a superior position, it is reasonable to assume that they would predominantly shape the interactions in hierarchical terms and determine unilaterally the environment of conversation, independently of what others may think or do.Clearly, a possible explanation is that the arranged and somewhat artificial environment of our discussion groups may have motivated participants to engage in dialogue and "listen to the other side" (Bächtiger & Gerber, 2014;Setälä & Herne, 2014).Previous studies have also demonstrated that participants tend to avoid vigorous confrontation and systematic contestation in discussion groups when no consequential decision-making is reached at the end (Bächtiger & Gerber, 2014;Fishkin & Luskin, 2005;Karpowitz, Mendelberg, & Shaker, 2012).
Deliberation requires the norm of equality and reciprocity.While this may seem an obvious fact, our study demonstrates a complex interplay of sources of authority that dynamically situate and re-situate participants in the space of conversation.In general terms, our findings corroborate our expectation that hierarchy would affect the community-police discussion; and participants with lower status would participate less in discussion (Gerber, 2015;Karpowitz & Mendelberg, 2014;Maia, 2012Maia, , 2014;;Marques & Maia, 2010;Mendelberg et al., 2014;Pedrini et al. 2013;Young, 2002).Both groups counted on different types of "authoritative standings" (Laden, 2012;Warren, 1996) that "authorized" them to participate in the discussion.As expected, the police officers were more active speakers and tapped mostly into sources of authority derived from stable, predetermined, functional positions in social hierarchy.Significantly, community residents mobilized mainly experiences to back up their claims.
However, our analysis points toward a more specific link between the sources of authority that shape the interactions between speakers.When discussion was at a low level of deliberation, police officers issued claims backed up mostly by functional credentials to frame the issue under discussion, pass judgements or make proposals in a unilateral way (Habermas, 1984(Habermas, , 1996;;Warren, 1996).When deliberation was at a high level (DTM 1), participants put forward credentials predominantly related to one's stories and life experiences.Interestingly, this does not mean that functional roles were completely erased.In deliberative moments, the police officers' functional roles appeared frequently aligned with individual and biographical experiences-what enabled these subjects to build other kinds of interactions.Differently from credentials which clearly come from socially and institutionally-based hierarchies that regulate behaviors and actions, personal experiences are not firmly established, and participants could engage in types of thoughts and feelings that did not depend on authority granted by formal education, institutions or expertise.Thus, a more symmetrical communication could be built.
Community residents relied almost exclusively on experiential knowledge obtained in specific contexts to speak with their own voices (Phillips, 1996;Warren, 1996).Life stories and narratives were frequently brought into debate by these subjects to define the conflict at stake and discuss what should be done to build a culture of peace.They particularly voiced experiences of humiliation, disrespect and harm to explain their concerns to the police officers, when the latter attempted to neglect or even ignore police violence.Community residents presented a much higher proportion of authority of connection than the police officers, insofar as they could align individual and collective communicative understandings or judgments more often.

Reciprocal and Hierarchical Relationships
Considering that reciprocal and hierarchical interactions are context-dependent, it is useful to illustrate how participants build these interactions in practice.
The following examples clarify how different types of credentials help either restrict or enlarge communicative space among speakers.

Reciprocal relationships
We have argued earlier that the probability of those invested with authority to exert influence is always present, even if this property is not explicitly exercised.
We have detected three different behavior patterns that help the most powerful put their functional authority aside and build reciprocal relationships.In these interactions, participants demonstrate openness toward each other or signal that they can equally engage in a joint conversation.These behavior patterns are: a) empathetic understanding; b) search for commonalities; and c) reflective selfcriticism.
Empathetic understanding.Previous studies have found that role-taking and empathy can increase constructive dialogue even in highly conflictual situations (Morrell, 2010).According to Habermas, "empathy, [is] the capacity to transport oneself by means of feeling across cultural distance into alien and prima facie incomprehensible life conditions, patterns of reaction, and interpretive perspectives" (Habermas, 1992, p. 269).According to Morrell (2010), empathy is multidimensional; it intertwines affect and cognition, projection and understanding.In our groups, unequal in power, empathy emerged as a subtle interplay of acknowledgment of needs in others as well as vulnerabilities that members of both groups can have.A good example is seen in Group A, when a police officer says: Josué, police officer: We don't see all slum residents as criminals or delinquents, not at all.I've several friends who live in slums, and personally, I see a minority [who are criminal], but everyone is harmed by that minority.So, we cannot make general judgments.When we come to work, we're prepared (…) we face difficulties, and still several policemen die.Do you understand?There's some fear regarding that minority and the reaction of people in the slum (...)You don't need to be afraid of the police.The police are not your enemy.(DTM 1, Group A).
The police officer does not issue a command based on his functional position.Through empathetic imagination, he projects community residents' harm and damages caused by unfair and distorted images.Josué reports bonds of friendship with some of these subjects; and somehow he shares the conception that such stereotypes are outrageous.Instead, presenting the police members as allpowerful creatures, Josué reports that they are susceptible to fear and suffering in many ways.He is also sensitive to slum residents' fear and negative feelings-an emotional state that usually obstructs open communication between members of deeply divided groups (Dembinska & Montambeault, 2015).
A similar demonstration of empathy is expressed by Paulo, a community resident, when he asks, "How [do you] break the stigma of being a community resident in the eyes of the police and the society and the stigma of being a cop in the eyes of the community?"Here, empathetic imagination is also invoked in a two-sided way: Paulo, community resident: What I think as more important for building a culture of peace is to break down the stigmas.I found, for example, the proposal of this group discussion very useful.If I saw her [pointing at Hanna, a police officer] in the street in that uniform, I wouldn't expect her to draw flowers [the drawing made by participants in the beginning of the group discussion].Because, as a community resident, I can say that this uniform reminds me of fear.I wouldn't imagine that he [pointing at the police officer Apoena] would draw a cross and think of Jesus, that is, the importance of Jesus' love for all of us here [a reference to Apoena's selfpresentation of his drawing].When he leaves his home, he thinks of how to transmit joy to his colleagues and to the people he meets.In addition, if I met any of the five [pointing at the police officers] without those uniforms, our relationship would be different.Similarly, when they come here, they face a huge stigma from slum residents, too.I think this group discussion establishes the possibility of breaking this stigma.We can collectively create actions to achieve this.(DTM 1 -Group A).
Paulo suggests it is hard to see the police officers as individualized persons with needs and concerns, unless slum residents remove "stigma"-mental models that reduce groups of people to oversimplified or fixed characteristics.By employing perspectival thinking, Paulo assumes the position of the police officers and acknowledges that stigma and fear are similarly present on both sides.He appeals to group participants to suspend their suspicion in order to constructively address difficulties across their divisions.Both speakers in the above excerpts combine empathy and understanding to open space for a truthful communicative, though a difficult exchange.
Search for commonalities.Deliberative scholars (Dembinska & Montambeault, 2015;Kanra, 2012;Luskin et al., 2014) have emphasized that members of divided groups may acknowledge they have enough in common to enable a constructive dialogue.Situations of reciprocal relationships are seen when police officers stress other aspects beyond their sphere of control in order to build bridges with the less powerful.
Ricardo, police officer: You [pointing at a poor community adolescent] can speak up.Feel free, don't be afraid and don't worry.We know there are many flaws [in the police institution], right?But you can be sure that we are doing everything we can.(…) We [the police] need to be perfect, but we aren't perfect, right?! We also feel tired, sleepy and hungry; we cry, we've feelings, but not everyone can see that.We're trained to ignore these feelings as much as we can, but a lot of us can't; we get emotionally involved with the cases, sometimes the death of a child, but we cannot be sentimental.We need to be made of steel because the community and society do not accept us being weak.(DTM 1, Group E).
Wiliam, police officer: Behind the uniform, we're humans as well, right?
We have family, kids.(...) I work in the community and I work in the center as well.My gun is always ready to be used.I've been in a shooting, I've seen my colleagues die, and I don't want this neither for me nor for my family.I'm a good citizen too and I also pay my taxes.I understand that ninety percent of the problem of the community is not a police matter (…) the community lacks infrastructure, sanitation, transportation, housing (...) Sometimes I feel helpless because our job is not to provide housing to citizens; our job is not to provide urban mobility, schools, sanitation, or transport.Our constitutional duty is to provide safety, qualified detentions, and prevention.The police are trying… we should be able to have a conversation like we're having here today.(DTM 4, Group C).
Instead of exercising ascendency over the group, both police officers attempt to scrutinize commonalities with the slum dwellers.Holding his authority at bay, Ricardo invites an adolescent to speak his mind, showing sincerity and respect.While acknowledging that the police are imperfect and there are many flaws that cannot be swept under the rug (Moellendorf, 2007;O'Flynn, 2007), Ricardo exposes his intimate thoughts, bodily needs, emotional or physical suffering, and embarrassing aspects present in everyone.By distancing himself from his functional role and dropping his guard, this police officer manifests different areas of vulnerability to others and finds a measure of equality with slum residents.Hence, he leaves room for the conversation to keep going.William says that "behind the uniform" they are persons who could build other kinds of interactions-intimacy and care in the family, cooperation in the neighborhood and solidarity as citizens.Both police officers acknowledged the police's merits as well as downsides and showed an interest in the slum inhabitants' worries and needs.They fostered the slum dwellers' sense of worth and enhanced their authority in the discussion; and allowed space for further contributions.
While this paper focuses on the police officers as authoritative agents, it is worth noting that in many occasions, community residents also recognize that members of these divided groups could understand one another across differences.
Luiz Gustavo, community resident: I had this usual view that the police are bad, brutish, and violent.(...) My first job opportunity was granted by a police lieutenant.That's when I started to understand the other side of the coin, not just as a community resident, but as a citizen.I think that, when a police officer leaves home, it's not a cop who is leaving: it is a human being.A family man who's leaving his mother, his child, and his wife-to take care of public safety and do a job that he has chosen (…) I think that in the community, as well, there are residents who are violent and avoidant [to the police], and I think that is a mirror (…).(DTM 1, Group A).
This community resident tries to relativize the negative views towards the police officers by adopting wider lenses to understand the set of relations these individuals establish in the intimate sphere, at work and in society.Also, here the view of one's biography and a messy admixture of social roles go beyond the police functional role (Gutmann & Thompson, 1996;Honneth, 1996Honneth, , 2001;;Walsh, 2007).Luiz Gustavo says: "We need to demystify the view that cops and the community can't coexist.I think we have to start working for that to happen."By showing respect and willingness to learn from the police officers' perspectives, this participant leaves room for establishing a new or not yet fully accomplished relationship between them.
Reflective self-criticism.Another situation of reciprocal relationship is found when the police officers adopt self-reflexive attitudes toward their superior standing and the sources of structural hierarchy.In Group F, a police officer brought particular situations of the exercise of power into discussion: Sérgio, police officer: The police has a strict disciplinary regime, which means the hierarchy determines that the orders are top-down.But this discipline cannot be blindly followed; you have to preserve your "I" (...) Of course, I can be punished for that [taking my own direction], right?And here comes our great dilemma: Stay or leave the institution?Is it something that I believe in?Is it something that I believe can be changed?I believe in participatory, democratic management, so that society can express their desires and what they expect from the police.But many sectors and those in high ranks-the secretary of public security, delegates, and colonels-do not see with good eyes the intervention of civilians.They say others want to teach them how to do their job (…) but it's not like that; there are things we don't know (DTM 4, Group F).
This police officer takes distance from his functional role and invites other participants to think along with him about "internal" dilemmas of an authorityholder.Instead of lecturing or defending the police institution from an externalobserver perspective as experts often do, Sérgio motivates group participants to perceive the police officers' internal doubts and pressures to conform to expected behavior.He encourages others to adopt a performative attitude to assess the system of rules and uncontrollable influences shaping the circumstances police officers find themselves in: "We need to make a good decision regarding someone's life in seconds, whereas a judge, a prosecutor, a jury, would have days, months, or even years to decide."Through an open-to-dialogue posture, Sérgio requests slum residents to question what is demanded for their mutual benefits.By putting the conflict between police officers and slum residents into a new perspective as part of a joint conversation, he creates opportunities for the less powerful to raise objections over the power-holders and the authority system alike.
To summarize, the police officers, in the cited excerpts, did not insist in their functional authority.Very frequently, the functional credential remained as a condition of self-identification, but these individuals employed different mechanisms-such as empathetic understanding, search for commonalities and self-criticism-to relate to the community residents in different domains of experience and to keep the space of communication open.Typically, in these moments, the police officers showed respect to their interlocutors.Their attitudes appear to evoke in the other discussants an expectation of a "response" rather than a "reaction."Hence, the police officers suggest that community residents could articulate their claims and disappointments and say what is important in relation opinions issued.Rather, she places herself as an office-bearer with authority as well as someone who is an authority on the subject under discussion.She claims that objective knowledge (not personal experiences) and accurate information "about the entire state police force" are needed for carrying out a proper discussion.She then characterizes slum residents' views as foolish.She says "It's interesting, isn't it?It's an extremely simplistic view, attributing a social problem to a professional class."In this way, Hanna disrespects previous opinions as well as participants in the discussion, and by no means considers others' opinions as equally valid.Once her instructions are not accepted without questioning, she threatens to leave the discussion: "I'm not here for this [to be criticized], and I'll walk out if this doesn't stop."Here the policewoman's superior standing is used to create a gap between Hanna and the other participants; and the interaction is marked by a reaction-sparking communication.
Reporting or Lecturing.While examples above show a relatively aggressive attitude towards the other participants and/or to their arguments, here we identify a form of hierarchy rooted in politeness, provision of relevant information and rational arguments.However, the speech is not a dialogue, but some kind of reporting or lecturing towards an "audience."Police officer Guilherme's lengthy statement is a good example: Guilherme, police officer: Good morning, everyone, again.And Marcia [the moderator], I ask everyone, in relation to that first question, to correctly define what we are discussing.Are we discussing the police's efficiency in the community or are we discussing a culture of peace with the police?Because we need to see if there's a real problem regarding the culture of peace with the police.We can look upon this [issue] in another way, through the complaints about the police's misguided actions in order to define this correctly.(...) Kaique, efficiency is one thing, and the culture of peace and the type of work done by the police in the community is another (...), whether it is effective or not, we're discussing one thing, but whether it is peaceful or not (...), that's very different.Then, I'd like to, if you allow me, to talk about the culture of peace with the police officer.
When we talk about peace, I cannot, honestly, (…) see the problem of peace with the police as being a problem.(DTM 2, Group C).
Police officer Guilherme presents good reasoning in a polite and professional way.He attempts to define the problem at stake and make useful distinctions, but he does not submit what he says to collective scrutiny.Previous studies have found that framing the discussion can produce domination insofar as this operation prevents participants from presenting their views or basing their judgments on a wide range of relevant reasons (Calvert & Warren, 2014;Mendelberg, 2002).Guilherme refuses to accept, even to acknowledge, the several dimensions of the problem evoked by the other participants, including the moderator; and thus he shapes unilaterally the environment of conversation.His unimpeachable certainty is professed when he concludes there is no conflict between the slum residents and the police, and therefore, there is no problem to be discussed.
Under this circumstance, Guilherme creates an atmosphere that tends to obstruct other participants' autonomy to reject his claims.Such an authoritarian posture precludes the possibility of bringing a divergent view into the discussion; and it also prevents participants from eventually adjusting where they stand as a consequence of what is said.He thus closes the possibility of jointly constructing a conversation.This attitude undermines the equality norm of discussion, leading to a painful silence in the group.Bickford regards silence as an indicator of nonverbal refusal to engage with the others' claims (Bickford, 1996).Participants could hardly speak in order to open new directions in the conversation, let alone seek to build a bridge between their differences.Even if using different means (commanding, reporting, lecturing), they are not truly open to consider what other people say.The space of conversation is restricted, and is likely to be defined by the more powerful speakers.

Reactions towards Authority
Whereas one may expect that authoritarian behavior constrains deliberative possibilities, this dynamic is not straightforward since it also depends on how other participants react, as well as on what they say.Furthermore, we should keep in mind that deliberation occurs in groups, and, thus, we need to pay attention to the ongoing interaction produced by many participants and how they react to authoritarian speech acts.
Our fourth question inquired into the responses of group participants to different types of authority.Our findings revealed that claims with authority resting on personal experience were largely accepted.In contrast, claims based on functional authority (even when this credential was accompanied by personal experience) provoked more diversified reactions, being rejected or ignored more frequently than experiential knowledge.Since experiences are not closely related to strict division of labor, discussants may acknowledge that this type of "authoritative standing" has no hegemony over them.Therefore, they may be more receptive to it.By contrast, discussants may resist a functional stance of power because it may block their autonomy to engage in critical examination of themselves and others-interpreting, questioning, and making demands through mutual communication.
Indeed, deliberative theory indicates that hierarchical relationships are never fully closed, but always subject to challenge (Habermas, 1984(Habermas, , 1996;;Laden, 2012;Thompson, 2008).This study corroborates the view that speakers with greater authority can easily end a conversation that raises objections or present reasons to revise a certain proposition (Laden, 2012).However, our analysis urges closer attention to moments when the exercise of authority does not undermine deliberative quality of discussions.For instance, in Group A, community inhabitants resisted police officer Hanna's attempts to impose her viewpoints on others.They promptly rebutted the claim that a "qualified" discussion would not occur through so many personal stories.Many slum residents kept on telling their personal stories.Thus, the relationship dominance-submissiveness could not take place and deliberation was sustained by group discussants.Now consider the second case of authoritarian relationship.In the course of conversation, police officer Guilherme showed politeness and rational argumentation; he called other participants by their names and demanded attention to what he said.Yet, Guilherme unilaterally framed the problem to be addressed, controlled the themes that could be talked about and often made long interventions that minimized other participants' chance to speak.In this situation, dominant behavior was complemented by submissive behavior.Participants were not able to re-configure the relationship in order to shape the conversation in a mutual way; and we hardly found any contestation.Thus, deliberation remained at a low level.
Prior scholarship suggests that people in disadvantaged positions hold back their voices in public discourse, due to a number of factors: the vulnerability of members from disadvantaged groups in the face of powerful ones; the lack of confidence and inhibition to challenge authoritative behavior; the avoidance of embarrassment and attempts to minimize the conflict; and the prioritization of bonds of connection with others (Conover & Searing, 2005;Gerber, 2015;Karpowitz & Mendelberg, 2014;Karpowitz, Mendelberg, & Shaker, 2012;Maia, 2014;Pedrini et al., 2013;Walsh, 2004;Warren, 2006).Our study clarifies how different sources of authority used by group discussants contribute either to enlarge or reduce the space of mutual communication among them.Yet, since authority is embedded within social interactions as such, these relations should be seen in the light of the individual potential to re-open the space for reciprocal communication.

Conclusion
Dynamics of reciprocal and hierarchical relationships within face-to-face discussions in groups unequal in power has not received systematic attention in studies on deliberation.This study focused on how slum residents and police officers structured conversation more or less spontaneously.It examined different moments of deliberation, sources of authority enlisted in each moment, and reciprocal and hierarchical interactions among participants.Our conclusion can be summarized in two major findings.
First, while it might be fairly obvious that reciprocity is conducive to deliberation and authority tends to obstruct it, we moved beyond this general insight.Our study provides empirical evidence supporting the view that some forms of authority are coercive while other forms can be non-dominative and attuned with freedom (Habermas, 1984(Habermas, , 1996;;Laden, 2012;Warren, 1996).Whereas our findings demonstrate that authority is mostly based on pre-existing social and institutional hierarchies, a close analysis of types of credentials helps us to grasp a nuanced interplay of forms to exert influence.In general terms, claims backed up by experiential knowledge favored mutual discursive engagement across group lines, while claims based on functional roles tended to block this type of communication.This paper offers important insights into how we can move beyond the more familiar, fixed, or constraining view of authority to assess the ways individuals situate and resituate themselves to reciprocate an invitation to dialogue.Paying closer attention to the combination of sources of authority and behaviors that help to create reciprocity pushes the view of deliberation further away from simplistic notions of equality.
Second, this study might add a layer to the growing literature concerned with the mechanisms required to build deliberation in groups unequal in power.Horizontal and reciprocal interactions were built when the police officers: a) engaged in perspective-taking, projecting, and understanding the disadvantaged subjects' frustrations, resentments, and fears; b) made a self-description of themselves that diluted their ascendency over others or produced a narration to build bridges to other participants; c) displayed a respectful stance toward those with less authority and an openness to their insights, perceptions, needs, or interests; d) ensured inclusiveness and voice for those with lower authority; and e) allowed critical reflection, including criticism on sources of authority themselves.We expect that these behaviors can be incentivized by moderators in deliberative forums.In many circumstances, the aforementioned mechanisms could help participants invested with authority and those in lower position to mutually shape the space of discussion, in spite of the fact that reciprocal relationship might be achieved only by some speakers and in certain "moments." Like any empirical study, this comes with several limitations.We should be cautious about making generalizations out of our investigation of six discussion groups.Future research could further analyze different types of conflict and seek to unpack how individuals challenge illegitimate authority in distinct situations.While we are aware that the composition of the group and its procedure shape the types of interactions, we could not specify the sorts of impacts that ethnic and gender composition, for example, played in the group dynamics or in individual attitudes in this study.Further research could also investigate different roles performed by the moderator or the emergence of leadership among group discussants in reciprocal and in hierarchical relationships alike.Despite these limitations, this study suggests that scholars and forum-organizers should pay more attention to different moments within discussion as well as the constitution of reciprocal invitation to dialogue in contexts torn by deep conflicts.

Figure 2 -
Figure 2 -Type of credential used by each category of participants (%)

Figure 3 -
Figure 3 -Type of credential used by the police officers in each type of DTM (%)