Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Politicizing deliberative democracy : strategic speech in deliberative systems Calvert, Aubin 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


24-ubc_2013_fall_calvert_aubin.pdf [ 1.81MB ]
JSON: 24-1.0074072.json
JSON-LD: 24-1.0074072-ld.json
RDF/XML (Pretty): 24-1.0074072-rdf.xml
RDF/JSON: 24-1.0074072-rdf.json
Turtle: 24-1.0074072-turtle.txt
N-Triples: 24-1.0074072-rdf-ntriples.txt
Original Record: 24-1.0074072-source.json
Full Text

Full Text

POLITICIZING DELIBERATIVE DEMOCRACY: STRATEGIC SPEECH IN DELIBERATIVE SYSTEMSbyAubin CalvertB.A. (Honours), Queen?s University, 2007A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYin THE FACULTY OF GRADUATE STUDIES(Political Science)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)August 2013? Aubin Calvert, 2013AbstractWhen using language to resolve conflicts and make decisions, people access democratic resources inherent in the practice of communication. Making a claim implicitly appeals to another?s capacity to agree to that claim autonomously?without being coerced or bought, and based on considerations he or she takes to be valid. Deliberative democracy describes political arrangements that harness this potential as the basis for collective decision-making. To the extent that it empowers individuals, however, actors have incentives to use language strategically to influence the very judgments deliberative democrats hope will be governed by carefully weighed reasons. In political contexts, language is often a tool for political ends, bypassing rather than engaging capacities for autonomous judgment. Deliberative democratic theories respond mostly by imposing the normative condition that deliberative speech should not be strategic. But the cost of this normative line is to depoliticize the theory, since it fails to engage much?even most?of the universe of speech in politics. Where democratic institutions channel politics?characterized by conflict and competition?into communication, we should expect speech to be strategic. Yet it is still possible for such speech to underwrite democratic autonomy. To establish a better understanding of strategic speech and its implications for democracy, I develop an analytic framework for conceptualizing the force of language. Under the model of communicative influence, the democratic implications of strategic language use depend not on intentions, but on how language produces pragmatic consequences, shaping the processes by which actors reason towards judgment and action. iiThe model generates propositions about what common features of political communication?narratives, loaded words, and exaggeration, among others?entail for the quality of political judgments. It also systematizes the specific anti-democratic hazards strategic speech that result from the frame-based, subject-based, and institutional ecologies of discourse that condition communicative influence. A democratic theory with analytic capacity around strategic speech can identify institutional interventions into these ecologies that promote autonomous judgment by targeting these specific hazards of strategic speech, without trying to work against the incentives and motivations that make problems political. The result is a politicized theory of deliberative democracy. iiiPrefaceThis dissertation is original, unpublished work by the author, Aubin Calvert.ivTable of Contents....................................................................................................................................Abstract	 ii......................................................................................................................................Preface	 iv.....................................................................................................................Table of Contents	 v.........................................................................................................................List of Tables	 viii..........................................................................................................................List of Figures	 ix..................................................................................................................Acknowledgements	 x..........................................................................................................Chapter 1: Introduction	 1.........................Communicative and Strategic Action in Deliberative Democratic Theory	 5.................................................................................Deliberative Talk and Its Limitations	 10....................................................Current Responses to Limitations on Deliberative Talk	 16..........................................................................Communicative Influence and Autonomy	 30..................................................................................................................Chapter Outline	 33Chapter 2. Habermas and Brandom: From Illocutionary Force to Deontic Scorekeeping..................................................................................................................................................	 42....................The Pragmatics of Communicative Rationality and Communicative Action	 43..........................................................Robert Brandom?s Model of Deontic Scorekeeping	 49...................................................................From Thin to Thick Models of Language Use	 56.................Chapter 3. The Model of Communicative Influence I: Inferential Structures	 61..................................................................................................Communicative Influence	 63..........................................................................Dimensions of Communicative Influence	 65....................................................................................Types of Communicative Influence	 78.......................................Summary and Comparison of Communicative Influence Types	 94.........................................................................................................................Conclusion	 97............Chapter 4. The Model of Communicative Influence II: Discursive Ecosystems	 100v...................................................................Conceptualizing the Discursive Ecosystems	 103.....................................Effects of Discursive Ecosystems on Communicative Influence	 117...................................................................Strategic Actors and Discursive Ecosystems	 130.......................................................................................................................Conclusion	 131........................................Chapter 5: Communicative Influence and Democratic Goods	 133.................................................................Discursive Democratic Goods and Autonomy	 135........................................................................Communicative Influence and Autonomy	 139................................Evaluating Discursive Ecosystems from a Democratic Perspective	 146.......................................................................................................................Conclusion	 156.............................................................Chapter 6. Patterns of Communicative Influence	 158..........................................Narrative Patterns: Testimony, Anecdotes, and Story-telling	 160................................................Comparative Patterns: Metaphor, Analogy and Example	 168.......................................Evocative Patterns: Persuasive Definitions and Loaded Words	 175.....................................................................Deceptive Patterns: Exaggeration and Lies	 182......................................................................Avoidance Patterns: Evasion and Hedging	 187.......................................................................................................................Conclusion	 193Chapter 7. Institutional Responses to the Anti-Democratic Hazards of Modification ....................................................................................................................................Effects	 196.The Logic of Institutional Interventions: Direct, Indirect, Site-level, and System-level	200...................................................The Anti-Democratic Hazards of Modification Effects	 205...................Direct and Corrective Interventions for Frame-Based Modification Effects	 206...................Direct and Corrective Interventions for Target-Based Modification Effects	 215.......................................................................................................................Conclusion	 222Chapter 8. Institutional Responses to the Anti-Democratic Hazards of Magnification ....................................................................................................................................Effects	 224........................................................Anti-Democratic Hazards of Magnification Effects	 225vi............Direct Interventions for the Anti-Democratic Hazards of Magnification Effects	 228.........................................................Corrective Interventions for Magnification Effects	 232.......................................................................................................................Conclusion	 235Chapter 9. Institutional Responses to the Anti-Democratic Hazards of Selection  Effects................................................................................................................................................	 237................................................................Anti-Democratic Hazards of Selection Effects	 239.................................................Direct Interventions: Preventing Pathological Selection	 244...........................Corrective Interventions: Balancing Communicative Influence Types	 253.......................................................................................................................Conclusion	 265.......................Chapter 10. Communicative Influence and Democratic Representation	 267............................Deliberative Democracy and Representation: Mansbridge and Disch	 268............................................................................................Anticipatory Representation	 273..............................................................................................Gyroscopic Representation	 278.................................................................................................Surrogate Representation	 282.......................................................................................................................Conclusion	 289.......................................................................................................Chapter 11: Conclusion	 291.........................................................................................................................Bibliography	 301AUBIN?S DISSERTATION - COMPLETE DRAFT CHAPTERS 1-10	viiList of Tables...........................................Table 2.1. Summary of Brandom?s Deontic Scorekeeping Model	 56.........................................................................Table 3.1. Types of Communicative Influence	 78........................................................Table 3.2. Types of Non-rigid Communicative Influence	 79...............................................................Table 3.3. Types of Rigid Communicative Influence	 85........................................................Table 3.4. Summary of Communicative Influence Types	 95.............................Table 4.1. Ecological Features that Condition Communicative Influence	 104...............................................Table 4.2. Subject-Based Features of Discursive Ecosystems	 108............................................Table 4.3. Summary of the Features of Discursive Ecosystems	 117...................................................................................Table 4.4. Types of Ecological Effects	 118...........Table 4.5. Institutional Ecological Characteristics of Common Discursive Practices	 125..............................................Table 6.1. Summary of Patterns of Communicative Influence	 160................................Table 6.2. Narrative Patterns of  Communicative Influence: Summary	 168...........................Table 6.3. Comparative Patterns of  Communicative Influence: Summary	 174................................Table 6.4. Evocative Patterns of  Communicative Influence: Summary	 181...............................Table 6.5. Deceptive Patterns of  Communicative Influence: Summary	 187...............................Table 6.6. Avoidance Patterns of  Communicative Influence: Summary	 192..............Table 7.1. Direct and Corrective Responses to Frame-based Modification Effects	 215...............Table 7.2. Direct and Corrective Responses to Target-based Modification Effects	 222.................Table 8.1. Key Functions of Institutional Interventions in Magnification Effects	 228.............................Table 8.2. Direct and Corrective Interventions for Magnification Effects	 235Table 9.1. Selecting for (?) or against (? ............) Rigidity, Reciprocity, and Transparency	 252.....................Table 9.2. Balancing Relationships between Communicative Influence Types	 261.......................Table 10.1. Representation Types and Hazards of Communicative Influence	 273............................Table 10.2. Representation Types, Democratic Hazards, and Correctives	 287viiiList of Figures.....................................................................Figure 2.1. The Model of Deontic Scorekeeping	 50...................................................................Figure 2.2. Implicit and Explicit Communication	 58...........................Figure 3.1. The Three Dimensional Structure of Communicative Influence	 77ixAcknowledgementsIn preparing this dissertation, I had many people helping and encouraging me, dedicating their own time to making it the best it could be. Most of all, I would like to thank my superb supervisor, Mark Warren, whose dedication and guidance were more than I could have asked or hoped for. Mark?s constant enthusiasm for the project kept me inspired, and I consider myself privileged to have had the opportunity to work with a supervisor so committed to his students? success.Committee members Chris Kam and Mike Neblo, were instrumental in helping me develop these ideas. Some of the biggest breakthroughs were the result of their thoughtful suggestions at crucial points in the process. I would also like to acknowledge the support of the other faculty members at UBC, who were always willing to make time to offer encouragement or feedback?especially Barbara Arneil and Bruce Baum. I also owe thanks to Dick Johnston for letting me work with him on the undergraduate honours program over the past two years. Mike Burgess of the Centre for Applied Ethics gave me great opportunities to apply this work in practical settings through interesting collaborations, for which I am very grateful. Finally, thanks to Jonathan Tomm, Sean Gray, and Kelsey Seymour for coming up with excellent insights and comments. To my wonderful friends: thanks for keeping me busy?with awesome adventures, ski days, and volleyball tournaments?and happy?with your excellent listening abilities, your hilarious commentary, and your words of encouragement, understanding, and support. Finally, I am most grateful to my parents, Bob and Ann Calvert, without whom I would never have had the opportunity to accomplish something like this, nor any of the other experiences that I value most.xChapter 1: IntroductionSpeech is essential to democracy. Democratic systems valorize speech as a means of resolving conflicts and making collective decisions. Language and communication enable self-determination by creating opportunities for people to be motivated by reasons they can both know and accept. Through language, people articulate their choices, beliefs, and values, which is necessary if a political system is to be responsive to its citizens. Communication allows people to coordinate social action, to express their identities, to establish interpersonal relationships, and to build trust. For all its democratic potential, however, speech has its own hazards. Speakers can use words strategically: as tools to suppress some considerations and to favour others, to undermine one another?s credibility, to misrepresent their positions, to deceive, and to manipulate. Rather than replacing money and power as a means of exerting political influence, speech can become their vehicle. Between these two extremes?on the one hand, the democratic ideal of speech that channels the autonomous motivating force of reasons and, on the other hand, outright manipulation?lies a range of communicative forms and functions, each with its own democratic possibilities and pitfalls. The aim of this dissertation is to provide a theory that identifies these possibilities in ways that can guide democratic responses.The problem of strategic speech can be formulated as follows. The benefits democratic theorists associate with speech?notably the ways speech can empower those affected by a decision to make autonomous judgments on the basis of reasons they are free to accept or reject?are premised on the absence of strategic orientations. The dominant assumption in deliberative democratic theory is that in order for speech to be consistent with democratic goods, it must follow the model of communicative action (Habermas 1984). 1Political actors represent their positions and the reasons behind them genuinely, and they should be prepared to justify their claims on grounds others can at least potentially accept. Within the currently accepted framework in deliberative democratic theory, as soon as speech becomes strategic (words become tools in the service of ends beyond the immediate circumstances of communication), the democratic goods of talk-based politics are compromised.At the same time, the circumstances under which strategic considerations pervade communication are precisely those cases in which it is most important for speech to function democratically: politicized situations with high stakes and conflicting interests. Such situations are often accompanied by mistrust and disrespect, and by the temptation for elites to use language to sway public opinion in their favour. This is not to imply that strategic speech is limited to fraught circumstances. Even friendly disputes may be characterized by selective uses of information or forceful logical structures, both of which deviate from the model of a pure exchange of justifications. Moreover, in neither case does speaking strategically necessarily entail deceit and manipulation. The point of departure for this dissertation is J?rgen Habermas?s conceptualization of language, which remains the most comprehensive account of the positive democratic potentials of speech. As I explain in detail below and in Chapter 2, Habermas?s theory necessarily entails the exclusion of strategic speech from these positive potentials. Because it is tied to Habermas?s model of language use, most deliberative democratic theory currently lacks the tools to differentiate among the full range of ways that actors with strategic intent?understood in the minimalist sense of orienting themselves to securing some outcome?use language to pursue their desired ends. And yet this strategic intent defines most political 2situations. Moreover, there are good reasons to think that the differences between being deceitful, using loaded words, and equivocating or hedging are normatively important, since each has implications for people?s ability to make autonomous judgments about politics. Likewise, the apparent differences between loose rhetorical appeals and tight syllogisms, or between invoking an issue frame and appealing to credibility, raise questions about the appropriateness of treating identically all forms of speech that are in some sense ?strategic.? The insufficiency of Habermas?s distinction between communicative and strategic action?both in terms of capturing variation among forms of strategic speech and in terms of the applicability of deliberative models to politicized situations?requires a theoretical response that can render insights about the democratic value of speech consistent with the realities of strategic intent.To the extent that deliberative democrats, following Habermas, explicitly or implicitly conceptualize all forms of speech along the single dimension of communicative action to strategic action, a second problem follows. Standard responses by deliberative democrats to deviations from ideal communication generally take two forms. The first merely discounts the potential for such conditions to produce deliberation or communicative action. The second attempts to alter the conditions that produce strategic action in order to move speech towards the ideal, often by downplaying or seeking to remove the incentives (or the actors) that render a situation political in the first place. Theoretical accounts of the democratic benefits of communication rely, implicitly or explicitly, on actor motivations to separate those forms of speech that preserve autonomy from those that undermine it. As a result, democratic theory lacks the basis for working with the strategic elements of speech that are pervasive in politics. This leaves deliberative democratic theory with no means of securing the democratic 3benefits of communication except eliminating the very strategic elements that follow from the structure of political problems. This dissertation offers a theoretical response to the ?problem? of strategic speech in deliberative democratic theory and practice. The approach I develop here represents a departure from the dominant solutions offered within the framework of deliberative democracy. The model of communicative influence I develop will provide a framework to connect the existing normative goals of democratic theory to a more refined set of distinctions among the ways communication exerts influence. One of the goals of the framework is to eliminate the implicit dependence on assumptions about actor intentions, which will allow the model to apply equally to conditions of conflict and cooperation. The analytic framework of the model of communicative influence yields a set of suggestions for institutional responses to the political conditions that select for different types of speech. Rather than trying to mirror or approximate conditions for ideal dialogue, these institutional responses seek to tap into the democratic potentials of strategic speech. Based on the theoretical framework I develop in the first half of the dissertation, I suggest that institutional responses can target the contextual factors that render strategic speech influential, either directly preventing hazards associated with communicative influence, or correcting for their consequences. One set of responses might include the ways political systems can position competing actors?representatives, interest groups, courts, and the media, for instance?to challenge one another when they use their status to exceed acceptable influence. Another might be to design sites of discourse with incentives that select for types of speech that counterbalance one another to secure broader, system-level democratic functions, ranging 4from building empathy and tolerance to mobilizing public opinion towards making or challenging a collective decision.I begin this introductory chapter by locating the roots of the problem of strategic speech in Habermas?s theory of how social actors use language to coordinate action on the basis of communicatively rational agreement. While I return to issues in the philosophy of language in detail in Chapter 2, I start here with a preliminary formulation that traces Habermas?s assumptions about communication into deliberative democratic theory. Second, I turn to deliberative democratic theory itself, identifying the psychological and incentive-based limitations on ?deliberative? speech. Third, I review the literature in deliberative democracy that seeks to respond to these limitations. I suggest that an ongoing attachment to assumptions about the distinction between communicative and strategic action hinder these responses. Fourth, I sketch the foundation for the model of communicative influence developed in Chapters 3 to 5 and differentiate it from existing approaches. Fifth, I describe the core normative commitments behind the model of communicative influence. Sixth and finally, I outline the dissertation as a whole. Communicative and Strategic Action in Deliberative Democratic TheoryThe problem of strategic speech has its roots in Habermas?s work in pragmatics and philosophy of language. It is worth explaining Habermas?s pragmatics at some length for two reasons. First, his theory remains central and useful to understanding the positive democratic potentials of language and establishing a connection between communication and democracy, even though, as originally formulated, it is tied to a flawed understanding of language itself. Second, specifying the conceptual problem at its foundation enables me to justify the need to 5step outside of existing frameworks in order to offer a theoretical framework capable of dealing with strategic speech. I make use of the precision and abstraction afforded by pragmatic concepts like speech acts, assertions and inferences because understanding how people use language strategically implies an understanding of what it is to use language in any way. I have chosen to focus on pragmatics over two comparable alternatives. First, going deeper into language, to the level of semantics, is unhelpful because the problem of strategic speech is not an issue of the rules of reference and grammar, but rather of how actors use language to achieve ends. Moreover, I accept the premise that pragmatics is prior to semantics?that meaning derives from how language is used in discursive practice. Second, approaching strategic speech at a higher level of generality, such as the study of rhetoric, does not give a precise enough leverage as to how persuasive speech generates the force that it does. Habermas uses pragmatics to advance a theory of social action grounded in discursive practice, the rationality of which, he argues, cannot be understood in terms of instrumental rationality. According to Habermas, when people use language to communicate, they necessarily incur commitments. This type of communicative action, in which people try to reach understanding on the basis of validity claims they raise and suppose can be redeemed, has the advantage that it relies for its effectiveness on people coming to agreement. That is to say, for communicative action to achieve coordination, people must accept one another?s claims (Habermas 1984). This idea of the voluntary, rationally motivating force of good reasons establishes a connection between communication and democratic autonomy, and grounds much of the deliberative democratic project discussed below. 6Communicative action and strategic actionCommunicative action is action oriented to reaching mutual understanding on the basis of validity claims. By raising a validity claim?making a statement that purports to be true, right, or otherwise valid?people are, in effect, doing something. To grasp what people are doing when they make claims, Habermas adopts from the speech act theory of Austin and Searle the distinction between the locutionary content of what is said, the illocutionary force of the speech act, and its perlocutionary consequences (Searle 1969, ; Austin, 1975; Habermas 1984, 286). The illocutionary component, which refers to what the speaker is doing in saying something?for instance, asserting, representing, commanding, prioritizing, warning, or identifying?is particularly crucial for Habermas?s theory (Habermas 1984, 320). In accepting a speech act offer?s illocutionary component, the hearer also commits himself or herself to the rightness of the command, the truth of the assertion, or the sincerity of the expression. These commitments represent the illocutionary force of the speech act, which describes how they can be binding on the ongoing interaction.Habermas defines communicative action in opposition to strategic action, that is action oriented to success. In communication, strategic action ?instrumentalizes speech acts for purposes that are only contingently related to what is said? (Habermas 1984, 289). As James Johnson points out, Habermas also loads onto the concept of strategic action the idea that it involves treating other people as resources to be exploited (1991, 190). For Habermas, the difference between communicative action and strategic action amounts to the difference between working with someone to reach understanding, and exerting influence upon them causally (1984, 286). As I explain below, Habermas?s use of speech act theory drives this 7theoretical distinction, which in turn prevents him from recognizing the ways in which strategic speech might tap into the rationally motivating force of reasons. The centrality of intentions to communicative actionIn distinguishing communicative and strategic action in this way, Habermas thus connects the meaning and normative significance of communicative action to the speaker?s intentions. In communicative action, speakers must ?unreservedly pursue illocutionary aims? (Habermas 1984, 305). Meaning derives from the illocutionary intent of the speech act, because this intention determines the relevant validity basis, and this validity (what it would take to justify the intended claim) in turn confers meaning on the speech act?s content. Habermas makes two assumptions that lead him to reject any normative potential for strategic speech. First, he assumes that the only form of normatively acceptable force is the illocutionary force of a claim that the hearer accepts as valid. Second, he assumes that the only time illocutionary force is legitimate is if it is self-sufficient, that is, if the speaker?s intention is only to secure the speech act?s acceptance as valid. Speech acts in which the illocutionary intent is not self-sufficient?that is, in which the illocutionary aims are not the whole purpose of the speech act?are strategic, not communicative. Such speech acts still make use of illocutionary force, but in order to secure that illocutionary force, the intent must be concealed. As a result, while the hearer accepts the validity of the speech act, it is not a rational (or autonomous) acceptance because the hearer has, in effect, been misled. Therefore, in instrumentalizing speech acts and treating communication as a strategic means, Habermas suggests that speakers also treat people as means rather than ends. Their agreement or their opinions are harnessed towards the speaker?s own ends in ways that compromise the autonomy of the hearers (1984, 288). 8Because perlocutionary consequences can only arise from accepting the illocutionary component of the speech act offer, Habermas claims that strategic action?an orientation to producing perlocutionary consequences?is parasitic on communicative action (Habermas 1984, 288). Strategic uses of language only work because the bulk of what goes on in discursive interaction is communicative. Parasitic forms of speech may reflect a narrow slice of how people use language. But it represents a particularly important slice for many of the most pressing political issues that do incentivize strategic behaviors that manifest themselves in language. Within Habermas?s framework, normatively acceptable influence in speech is limited to the illocutionary binding force of a rationally accepted speech act offer. In order to meet this criteria, the speech act must not be oriented to producing a specific outcome beyond its acceptance. This theoretical move blocks consideration of the positive potentials of strategic speech, prevents more fine-grained distinctions among different forms of strategic speech, and renders the idea of the rationally motivating force of reasons inapplicable to many of the most pressing political problems.An alternative account of the force of speechBecause Habermas?s conception of communicative action?s depends upon actors? intentions, it is not easily applicable in circumstances in which actors? dispositions do not already favour cooperation. Furthermore, it fails to capture the full range of ways actors use speech to produce consequences. To remedy these failures, we have to move away from Habermas?s problematic focus on communicative intent. This requires replacing the construct of illocutionary force, which is itself grounded in questionable aspects of speech act theory (Heath 2001, 47). I will set out such an alternative account of the force of speech in Chapter 92. My primary source is Robert Brandom?s pragmatic model of language use. According to Brandom, the problem with Searle?s typology of illocutionary acts, and similar philosophies of language, is that they are grounded in agent semantics: they attempt to define meaning in terms of how the speaker intends to be understood (1998, 147). I will explain how Brandom moves away from intentions by replacing agent semantics with functional semantics, in which meaning derives from the inferential antecedents and consequences of an utterance rather than intentional states (Brandom 1998, 147-51).1 In this way, Brandom?s account of language lays the groundwork for an account of the force of language that no longer prioritizes intentions.Deliberative Talk and Its LimitationsMany strains of deliberative democratic theory take up the claims about political life that follow from Habermas?s theory of language, applying them to democratic institutions and to the public sphere. There are three principle ways in which deliberative democrats have drawn on Habermas?s theory of language. The central claim is that, under the right conditions, the right kind of talk can secure a democratic basis for collective action and ensure that decisions reflect the considered judgment of all affected. Deliberative talk generally involves ?debate and discussion aimed at producing reasonable, well-informed opinions, in which participants are willing to revise preferences in light of discussion, new information, and claims made by fellow participants? (Chambers 2003, 309). 101 Both Brandom and Habermas argue for the primacy of pragmatics over semantics; i.e  that the meaning of language derives from what it is to use language. Semantic assumptions, though, are still a necessary part of a pragmatic theory, even if they follow from pragmatics.A second implication of Habermas?s work has to do with the democratic legitimacy of deliberative outcomes. Those deliberative democrats who take up the notion of the rationally motivating force of good reasons?in other words, the force of the better argument (Habermas 1984, 25)?see deliberation as ethically desirable insofar as it respects the freedom of interlocutors to accept or reject the claims that others put forward (Thompson 2008). The democratic impulse here is that agreement cannot be imposed, but must be earned on the basis of the validity of reasons, and that the conditions under which this agreement can be rational can be created or approximated. Deliberation is also said to be epistemically desirable, in that decisions are subject to vigorous testing on the basis of an expanded pool of reasons (Bohman 2006; Estlund 1997). This derives most directly from Habermas?s claim that when actors encounter a misunderstanding or disagreement, they move from communication, in which validity claims remain implicit, to discourse, in which claims must be justified in terms of their validity. Moreover, though discourse itself is only a small part of communication, the very idea of communicative rationality relies on the possibility of discourse as a ?court of appeal? (Habermas 1984, 17) within which validity conditions are tested. Thus, the possibility of discourse enables the motivating force of reasons and enhances the epistemic quality of those reasons through testing. At a practical level, deliberative democrats look for such opportunities within political systems. A third way in which deliberative democrats draw Habermas?s work into the realm of political systems is through the discourse principle (D), which states that ?Just those action norms are valid to which all possibly affected persons could agree as participants in rational discourses? (Habermas 1996, 107). By rational discourses, Habermas means 11Any attempt to reach an understanding over problematic validity claims insofar as this takes place under conditions of communication that enable free processing of topics and contributions, information and reasons in the public space constituted by illocutionary obligations (Habermas 1996, 107-08). In Habermasian terms, establishing the validity of norms using the discourse principle?in other words, settling rightness claims?is analogous to testing truth claims through discourse with reference to facts. Deliberative democrats apply these principles in practice to identify or create enclaves under which the validity basis of decisions can be judged with reference to the considered interests of those affected. Instances of genuine deliberation thus add democratic or self-determination benefits to the advantages of deliberative speech.Not just any form of talk can deliver these democratic benefits. Talk is deliberative when it consists of ?reason-giving? (Mansbridge et al. 2010, 498), generally in a reciprocal or dialogical format such that participants in deliberation are responsive to one another, rather than merely re-asserting a fixed position (De Vries et al. 2010; Stromer-Galley 2007). Reasons should at least be potentially convincing to a broad or universal audience (Bohman 1996), under conditions of respect, inclusivity, and equality (Gastil, Black and Moscovitz 2008; Knight and Johnson 1997). While there are ongoing debates and competing interpretations of some of these criteria,2the above discussion represents a broad consensus on the nature and advantages of deliberation. This conception of deliberation runs up against two sets of barriers. The first is based on incentives, and the second, on psychological constraints.122 For example, on the necessity of sincerity (Adler 2008; Lenard 2008; Warren 2006, 2008) or the necessity of consensus or decision-seeking (Thompson 2008; Sunstein 1999). For an overview of the key tensions in the field, see Neblo (2007).Incentive-based limitations on deliberative talkThe incentive-based limitations on deliberation are a result of the fact that political problems demand coordinated action under conditions of conflicting interests, high stakes, and uncertainty (Bohman 1996; R. Calvert and Johnson 1998; Warren 1995). By their nature, these conditions favour strategic action. Political actors are motivated to take the means that achieve their ends, and strategic speech can be a good way to get what they want, often at the expense of interlocutors? preferences. Hobbes was one theorist who recognized the problem of strategic speech clearly. The principal threat to rationality and civil society, according to Hobbes, is the corruption of meaning as a result of the fact that people use words in motivated ways (Pettit 2008; Ball 1985). Hobbes?s proposed solution was to generate external incentives to keep language use in line by establishing a sovereign who would guarantee all covenants, including those governing the meaning of words (Wolin 1960). Outside an Hobbesian state, in situations where we cannot count upon a sovereign to make, and enforce, rules governing how words are used, the balance of incentives will often favour the instrumental or strategic use of words. The incentives that select for strategic speech may be a function of the features of the particular collective action at hand (Pellizzoni 2001, 79), and particularly of the extent to which actors? preferences are misaligned, making communication more difficult (Bloom et al. 2001, 80). Incentives may also derive from the features of the institutions in which speech takes place. Do they promote competition or pre-determine when people will have opportunities to speak (Landwehr and Holzinger 2010)? Do they include or exclude powerful actors with a direct stake in the problem (Hendriks 2006)? 13Rational choice theorists have modeled the role of communication in cooperation problems, with results that call into question the possibility of deliberation as it is currently conceived. This literature suggests that, under many circumstances, ?talk is cheap? (Austen-Smith 1992, 46). Since, in many cases it costs nothing for people to make verbal claims, if it is in their interest to make the claim, they will do so whether it is true or not. As a result, words lose their ability to communicate meaning. In the absence of shared preference, costless verification, or costly signaling, private information cannot be credible (Austen-Smith 1992; Farrell 1995). Others have refined these models to suggest that there are some circumstances under which cheap talk can be informative. For instance, communication may be self-committing (where the structure of the game is such that the speaker would have an incentive to follow through if his or her proposal were believed) or self-signaling, (where speech is credible if one would want it to be believed only were it true) (Baglia and Morris 2002). The direct implications of these models may be limited to circumstances in which speech serves only to convey private information. They nevertheless represent a fundamental challenge to deliberative theories of democracy, in that they highlight how incentives set limits on what deliberation can achieve. Democratic theory needs to take account of these limitations, in part by recognizing how different forms of communication may be compatible with, or complimentary to, preference maximization. Strategic actors can also leverage the power of speech to direct attention to particular considerations (Hafer and Landa 2007); establish connections among propositions (List 2011; Patty 2008); alter the dimensionality of issues (Riker 1996); or coordinate around focal points on the basis of principles, precedents, and symbols (R. Calvert and Johnson 1998; Johnson 1993). Each of these possibilities 14suggests that the existing framework, which divides speech on the basis of either strategic or communicative action orientations, may be inadequate to understanding all the functions of speech under the conditions of politics, which demand some measure of strategic orientation from actors.Psychology-based limitations on deliberative talkA second set of challenges to the model of interpersonal communication behind much of deliberative democratic theory come from what research in psychology tells us about how people actually reason. To review, Habermas?s pragmatic account of communicative action takes the following form. When coordinating social action through the medium of language, actors, in effect, issue a warrant or guarantee that the claims they make can be justified. If a claim has any persuasive force, it is because the hearers have either accepted the speaker?s warrant, or the reasons that support it (and reasons for those reasons in a chain of justification).This model of raising and testing validity claims is valuable for drawing attention to the democratic resources residing in the basic social practice of communication, as I argued above. But it bears little resemblance to what research in psychology tells us about how people reason. Rather than deliberately tracing the validity of a claim through a chain of justifications, people reason according to predictable patterns that are not necessarily determined by the substantive content of claims. Psychologists and cognitive linguists know that we make some connections rather than others on the basis of symbols (Sears 2001); that we respond to frames that draw our attention to some considerations over others (Chong and Druckman 2007b); and that we are biased towards confirmatory evidence (Mercier and Sperber 2011). Our thoughts are constrained by heuristics (Kuklinski and Quirk 2000), 15structured by metaphors (Lakoff and Johnson 1981), and weighted according to our perceptions of a situation, including whether we stand to gain or lose (Mercer 2005). All things being equal, our judgments tend to favour simple explanations over those that are both more probable, and more complex (Lombrozo 2007). Work in informal logic similarly suggests that we are influenced by the ways arguments are formulated, including the use of ridicule (Van Laar 2008), selective illustrations (Perelman and Olbrechts-Tyteca 1969), and even personal pronouns to deflect blame or enhance credibility (Fetzer and Bull 2008).Thus, how we formulate and respond to arguments depends on much more than the logical or rational justifications of claims. The fact that interplay between language and pre-existing psychological tendencies and cognitive structures may determine the quality of the force of communication distances us from the notion that speakers articulate reasons that others simply accept or reject. These latent forms of psychological influence are worth exploring, in part because they may challenge the democratic value of even some well-intentioned forms of speech. As well, strategic actors can construct their arguments to take advantage of these tendencies, potentially extending their influence well beyond the substance of what they say. Both these possibilities suggest a need to revisit the idea that the force of the better argument rests solely with the acceptance of the validity of a claim. Current Responses to Limitations on Deliberative TalkMany deliberative democrats have recognized these limitations, and the deviations from deliberative ideals that strategic action entails. This section identifies several approaches that respond to the problem of strategic speech. These include, first, attempting to clarify the status of deliberative standards so as to make their criteria less stringent; second, 16developing typologies that distinguish between deliberation and other forms of speech; and, third, establishing institutions to approximate the necessary conditions for deliberation. In addition to these approaches, I focus closely on Nathaniel Klemp?s (2012) The Morality of Spin, which offers another take on the problems associated with strategic speech.Modifying the status of rational dialogueThree distinct theoretical strategies respond to the pervasiveness of strategic motivations in politics by modifying the status of the ideal of rational dialogue. The first strategy treats the ideal as regulative, the second specifies more inclusive criteria for what counts as deliberation, and the third attempts to identify the deliberative potentials of seemingly non-deliberative speech, namely, rhetoric. The ideal speech situation as a regulative idealOne response to the problem of strategic speech involves emphasizing that the ideal speech situation, a key component of communicative rationality, is a regulative ideal against which one can measure actual speech and its conditions. Treating the deliberative model as an ideal (Steiner 2008, 189) allows us to place specific instances of speech on a scale of being more or less deliberative (McCarthy 1990, 456), rather than making the binary judgment that all speech that does not mirror rational dialogue is simply not deliberative. But there are two problems with treating the ideal speech situation as a regulative ideal. The first is that its primary function is as a pragmatic presupposition. In order to engage in communicative action, actors must assume that all claims could, at least potentially, be justified under conditions of ideal dialogue. They must further assume that no force except the rationally motivating force of valid reasons?for instance, coercion, instrumental 17incentives, or personal power?holds sway. The ideal speech situation is only indirectly about correspondence between the ideal and reality. It can be projected onto actual circumstances in order to ascertain whether or not people have good reasons to adopt the performative attitude communicative action requires, but it would be wrong to assume this means that the purpose of any single political institution should be to replicate these conditions. Neither treating the ideal speech situation as a regulative ideal nor as a pragmatic presupposition we can project onto the world tells us anything about what happens when actors do not have sufficient grounds to attribute communicative intent to others. If reasons cannot have motivating force under such circumstances, the democratic potential of communication would be limited to relatively easy cases. Neutralizing limitations on deliberative speech by creating a space in which the only form of power is the persuasive force of reasons appears insurmountable in the face of the conflicting incentives that characterize most political situations. While we might succeed in creating small scale deliberative enclaves?the impetus behind designing and empowering mini-publics?that tells us little about most of political discourse, and whatever democratic potentials it may have.Relaxing deliberative criteriaIf the first approach is about loosening the way deliberative ideals apply to reality, the second is about relaxing the content of those ideals. For example, in Public Deliberation (1996), James Bohman suggests moving away from procedural accounts of deliberation, and focusing instead on instances of successful coordination on the basis of reasons. In their review of the field, B?chtiger et al. (2010) pursue a similar line of thought. They suggest 18distinguishing Type I deliberation, which conforms closely to the Habermasian standard, from Type II deliberation, which admits more flexible forms of discourse. Type II deliberation emphasizes relatively democratic outcomes, and seeks to attend directly to the real world constraints on normative ideals (B?chtiger et al. 2010, 33). Nonetheless, B?chtiger et al. also identify problems with the movement towards the less demanding Type II conception of deliberation. As the category becomes broader, deliberative democrats may no longer be able to claim many of the ethical and epistemic benefits attributed to deliberative talk. Loosening deliberative criteria without also offering a means of drawing distinctions between speech that serves deliberative ends and speech that does not represents another dead end in resolving the problem of strategic speech. Moving towards Type II deliberation entails giving up communicative action as a normative metric without replacing it with a new one. B?chtiger et al. recognize this problem, but their solutions involve reintroducing Type I standards, in the form of rational justification and sincerity criteria. By contrast, this dissertation will help resolve this problem by offering a more inclusive account of speech along with a way of specifying which forms of speech can secure the democratic benefits attributed to deliberative speech. To foreshadow the argument in subsequent chapters, replacing communicative action with an alternative pragmatic model yields a conceptual framework capable of making distinctions among forms of discourse without re-introducing the same Type I standards that smuggle in assumptions about communicative intent. Such a framework will be better suited to covering a broad range of types of communication. Like the Type II approaches B?chtiger et al. identify, it will focus on outcomes rather than intentions. In this case, the relevant outcomes are at the level of individual commitments, judgments, and choices.19Rehabilitating rhetoricA third way of modifying the status of the ideal of rational dialogue involves seeking the deliberative value in rhetorical speech. Rhetoric sits uneasily in a deliberative framework because it tends to be strategic, and would seem to admit both insincerity and demagoguery, both of which threaten deliberative goods. However, Simone Chambers and John Dryzek have both recently sought to integrate deliberation and rhetoric. According to Chambers, given the fact that mass democracy prevents most communication from being dialogical, deliberative democrats should distinguish between plebiscitary rhetoric, which is ?concerned first and foremost with gaining support for a proposition,? (2009, 337) and deliberative rhetoric that prompts thoughtful reasoning and avoids emotional triggers. This contribution, while useful, leaves a number of questions unanswered. What makes some instances of rhetoric and not others spark thoughtful reasoning? What counts as putting winning ahead of reflection? These questions reflect the fact that while operating within a framework of communicative action, it is very difficult to avoid using speaker intentions to identify democratically undesirable forms of speech and risks leaving politics, with its winners and losers, outside the scope of deliberative rhetoric. In Dryzek?s view, the democratic quality of rhetorical speech depends on its consequences. Does rhetoric facilitate reflexivity, generate legitimacy, or heal divisions (Dryzek 2010, 332)? While rhetoric may in some cases contribute to positive democratic outcomes, Dryzek?s theory does not provide the tools to establish systematic connections between these outcomes and forms or instances of rhetoric.What makes one instance of rhetorical speech promote the deliberative virtues of reflection and reflexivity (Dryzek 2010, 329), but not another? It may be possible to make more specific claims about the relationship 20between forms of speech, on the one hand, and responses at the individual level, on the other, but only if we re-think the pragmatics of communicative influence. Each of the three approaches outlined above reflects some effort to address the effects of strategy by altering the status of the ideal of rational dialogue. At the same time, they all recognize the importance of maintaining some basis for making positive normative claims about the democratic benefits of communication. As a result, the ideal of communicative action still skews their responses towards excluding strategic speech, and so moves deliberative democratic theory away from politics. If we continue to see strategic forms of speech only in negative terms, we will be unable to connect the normative insights of deliberative democracy to the conditions and practices of politics. The concept of strategic speech itself cannot capture normatively important variation in the ways speech can be oriented to securing outcomes because it is only defined on a single dimension of opposition to the ideal. Understanding how communication can enable or undermine democratic outcomes requires a closer analysis of how it works outside ideal conditions.Typology-based responses to incentive problemsA second general set of approaches to the problem of strategic speech addresses incentive-based limitations on deliberative speech by developing typologies. These approaches start from the premise that it does not make sense to apply deliberative criteria to all social interaction mediated through language. Just because actors use communication to coordinate does not mean that one should be looking for communicative action. Rather, sometimes coordination is clearly governed by instrumental rationality, which is associated with strategic action. In this section, I consider three such approaches.21Arguing vs. bargainingThe first, and perhaps most familiar typology is a distinction between arguing (the realm of communicative rationality) and bargaining (the realm of instrumental rationality, and by extension, strategic action) (Holzinger 2004; Risse 2000; Elster 1997). Arguing and bargaining envision very different connections between speech and outcomes. In argument, speech targets and changes beliefs. In bargaining, speech enables a mutual adjustment of demands and the identification of overlapping preferences. The claim here is that the presence of instrumental reasoning in communication does not represent a fundamental challenge to the normative account of deliberative talk, because deliberative standards are only properly applied to the realm of communicative action: argument, and not bargaining. The problem with this approach, at least as an answer to the problem of strategic speech, is that the distinction between communicative and strategic action does not map onto the distinction between arguing and bargaining. The former distinction has to do with action orientations, and the latter, the purposes and types of claims being made (Saretzki 2012, 177). The arguing and bargaining typology does not resolve the problem of strategic speech because strategic motivations can pervade both sides of the dichotomy. If strategic speech is defined broadly to include any communication where speech is oriented to outcome rather than mutual understanding, there is no reason to think that argument cannot be just as strategic as bargaining, although speech acts may be performing different functions.Four ways deliberative talk can resolve conflict of interest problemsSecond, in an effort to combine self-interest and deliberation, Mansbridge et al. (2010) suggest a four part typology for understanding the role of deliberative talk in resolving conflict. In convergence, actors reach an agreement for the same reasons; in 22incompletely theorized agreements, they reach agreement, but for different reasons; in integrative negotiations, parties find a way around conflict by exploiting different aspects of an issue for joint gain; and in cooperative negotiations, parties deliberate to settle on a fair process, eschewing power and strategic posturing (Mansbridge et al. 2010, 7-10). This typology does address the issue of incentives, but it does not speak to the exact problem of strategic speech. The problem of strategic speech instead concerns how actors can use speech to secure ends not by hammering out agreements, but by influencing the judgments and decisions of others. Because Mansbridge et al.?s typology is an answer to a slightly different question?namely, what kinds of agreements can actors reach in the face of conflicting self-interest?it does not challenge the assumption that actors proceed on the basis of reasons, albeit potentially competing or contradictory reasons. Coordinative and discursive typesClaudia Landwehr?s (2010b) typology comes closest to the specific problem of strategic speech I identified above. Landwehr?s typology applies to the communication that precedes political decisions. The features of communication relevant to Landwehr?s typology are whether or not it is discursive (characterized by dialogue and publicity) and coordinative (oriented to producing agreement on a course of action). These two dimensions yield a four part typology. Discussion is discursive but not coordinative. It involves an exchange of reasons that others might potentially accept, but not necessarily with an orientation to securing some outcome. Deliberation is both discursive and coordinative. Debate is neither. It is not meant to be dialogical, nor is it oriented to actually making a collective decision. Bargaining is coordinative but not discursive. Actors seek an outcome, but not on the basis of reasoned exchanges (Landwehr 2010b). Landwehr?s typology has the advantage of not 23crowding the modes and purposes of speech onto the communicative action/strategic action distinction. Because she avoids relying on the concept of communicative action, she is able to highlight some important differences between kinds of communication. For example, her typology illuminates key aspects of debate that differentiate it from discussion, on the one hand, and bargaining, on the other. Debate is not intended to elicit dialogic responses, nor does it provide a direct link between communication and decision. However, Landwehr?s typology falls short in resolving the problem of strategic speech. It is a step removed from the level at which communication actually operates. Speech works first on individual beliefs and judgments at the level of reasoning, and only second on collective decisions. I argue that we need a ?micro? account, where outcomes refer to the immediate connection between speech, on the one hand, and a target?s belief or action commitments on the other. The question of how these add up to collective decisions is a separate issue. It does not require a theory of the pragmatics of language like the one on which deliberative democratic theory currently founds its normative arguments for talk. Rather, it requires theories of aggregation and negotiation. The same rules that are said to enable deliberation?excluding forms of power and influence based in money or coercion, for instance?may also be necessary conditions for these other forms of talk-based coordination, but we still need to understand how strategic speech exerts influence, rather than lumping it in with other forms of instrumental action that unfold through communication.The three typologies discussed above capture distinct ways of connecting communication to a joint or collective decision. One problem that afflicts all of them, however, is that they are framed at the wrong level of analysis to understand the distinct problems posed by strategic speech. Our conception of strategic speech needs to be focused 24first of all on how individual speech acts induce others to take up commitments. Just as communicative action concerns the pragmatic orientation towards the validity basis of claims and their uptake or rejection, strategic action can be conceptualized at the same level: how do actors use words as tools to get people to take up claims? Distinguishing instrumental rationality, self-interest, or decision-orientedness in the ways people reach decisions through communication reflects a secondary stage of the process. This secondary stage should be primarily concerned with decision-rules, thresholds for agreement, resources for bargaining and the like. To understand the pervasive influence of strategic speech, a typology should capture the ways that actors use language to target the beliefs and judgments that precede aggregation and negotiation. Mini-publics: An institutional workaround for the problem of strategic speechPart of the problem of strategic speech is that politics undermines the conditions for genuine deliberation. Another response, then, is to create those conditions by designing institutions that foster good deliberation and discourage strategic speech. ?Mini-publics,? small scale deliberative bodies created by governments and civil society actors, are well known examples. Mini-publics are made up of a sample of the constituency relevant to a problem. Members are asked to learn about it, deliberate, and issue advice or recommendations. Mini-publics can range from relatively formal citizen juries empowered to make decisions, such as the Oregon Citizen?s Initiative Review, to informal venues that generate considered opinions, such as AmericaSpeaks? 21st Century Town Meetings. Variations on such processes have proliferated (see Fung 2003). Generally speaking, one of the unifying assumptions is that mini-publics can better approximate the conditions of ideal deliberation than the uncontrolled public sphere (Webler 1995). They typically enforce 25equality and mutual respect (De Vries et al. 2010; Gastil, Black and Moscovitz 2008), use facilitation to draw out an exchange of reasons, and ensure that participants have sufficient information on which to proceed (Abelson et al. 2003).One of the attractive features of mini-publics is that, in principle, they make it possible to create enclaves of deliberation genuinely insulated from all force except that of reaching agreement on the basis of validity claims. However, mini-publics are an incomplete solution to the problem of strategic uses of language for two reasons. First, as discussed above in the section on psychological constraints on deliberation, framing, heuristics, and language choices, among other features of discourse, can themselves exert force on reasoning that exceeds the substance of claims. There is no reason to think that participants, even in the relatively controlled circumstances of a mini-public, will overcome these psychological tendencies?and indeed, group dynamic may exacerbate them and introduce others (Perrin 2006). Nor, as a general matter, are mini-publics likely to remain insulated from powerful discourses that pervade the public sphere, which can affect the way that reasons are articulated and received within more sheltered processes (Fraser 1990). Second, as Chambers points out, the trend towards thinking that mini-publics are the only sites at which deliberation can take place is troubling because this emphasis comes at the expense of the democratic quality of political systems more generally (2009, 330). The framework I develop in this dissertation opens up new possibilities to respond to Chambers?s concern. I argue that democratic quality at the system level does not require proliferating deliberative venues or rendering established venues more deliberative, desirable as these may be. Rather, democratic theorists need to identify the ways that communication exerts influence that may either underwrite or compromise autonomous judgment. Instead of 26thinking of mini-publics as isolated enclaves of deliberation, then, one should consider them as supplements to the deliberative quality of political systems. Mini-publics create space for types of speech that, while not necessarily deliberative, correct for the autonomy-compromising tendencies of public discourse around certain fraught issues. They may inject new reasons or interpretations of problems or give people space to engage in dialogue that allows them to interrogate the sources of their own beliefs. Chapters 7-9 argue for an expanded set of functions for mini-publics based on these possibilities. A moral approach: Klemp?s typology of deliberative persuasion, strategic persuasion, and manipulationKlemp identifies and responds to the problem of strategic speech by developing a set of conceptual tools for distinguishing among various forms of non-ideal speech (2012, 5)?in other words, better and worse versions of the rhetoric that pervades public discourse. According to Klemp, it is possible to distinguish three kinds of speech: manipulation, strategic persuasion, and deliberative persuasion. The typology is based on two criteria: first, whether the speech bypasses the listener?s rational capacities and, second, whether the speaker intends it to do so. Rhetoric is manipulative when it deliberately bypasses the listener?s rational capacities?with an intention to use this avoidance of rational capacities to secure some outcome. Strategic persuasion takes place when the listener?s rational capacities are engaged openly, but with an intention to win. Deliberative persuasion, finally, is pursued in good faith and on the merits of the argument (2012, 57). Klemp?s typology remains grounded in Habermas?s distinction between communicative and strategic action, and merely introduces an additional distinction within strategic action. In contrast, by using Brandom?s basic framework instead of Habermas?s, I 27am able to focus on the dynamics of language use rather than its intent. Both my approach and Klemp?s highlight speech?s power to induce agreement, and ask when this capacity compromises autonomy. However, I proceed on the assumption that the best way to study speech as a feature of political life is to look at language more deeply, and ask how words and arguments can be said to exert any force at all. Following from his use of Habermas, Klemp keeps intentions at the heart of his argument, whereas my purpose is to try to understand the normative potentials and hazards of speech, including strategic speech, apart from intentions.3 An intentions-based account like Klemp?s asks if people are being treated as ends or means. A framework based on the dynamics of language use asks what it means for a choice or judgment to be considered autonomous in the face of exposure to a range of often competing influences, each deploying the resources of information and context. My model of communicative influence takes this second approach. While intentions may dictate the moral quality of speech, democratic theorists should be more concerned with the effects of speech, measured in terms of their ability to secure or undermine the goods associated with talk-based politics.Situating the model of communicative influenceThe model I develop in this dissertation overcomes many of the limitations of the approaches discussed above. It uses pragmatic theories to explain communication on its own 283 Klemp draws on Philip Pettit to support his use of intentions, arguing that theorists need to be concerned, not with protecting people from chance, but with protecting them from one another (Pettit, 1999 p. 52-53 qtd. in Klemp, 2012, p. 70). However, as I discuss in Chapter 5, the domination with which republicanism is concerned is a feature of interpersonal relationships. I argue these should be the subject of a different type of normative evaluation from the dynamics of speech, which should focus on the autonomous quality of judgment rather than parallel questions of whether the choice that follows is empowered. Manipulation may be a form of domination, but it must also be subject to evaluation as part of a broader complex of communicative types. terms, apart from speakers? intentions or stipulations about the conditions that actors must fulfill in order to engage in deliberation. I explain, in Chapter 3, that the inferential structures inherent in language use?structures that parallel the webs of inferences that confer meaning and enable reasoning?can exert force by directing reasoning through inferences that actors are compelled to accept, insofar as they are competent users of language. Like many of the approaches discussed above, I develop a typology; however, where other typologies focus on how communication is connected with decisions, I ask about the different ways speech operates as a form of influence. While the approaches discussed above do include belief change as a mechanism of reaching agreement or coordinating action, they do not specify how speech can change beliefs, except by appeal and agreement. Assigning analytic priority to good intentions has contributed to de-politicizing deliberative democratic theory. I intend to offer a theoretical means of reversing this tendency. Focusing on intentions neglects the possibility that even well-intentioned speech can compromise autonomous judgment to some extent?for instance, by perpetuating a strong frame or claiming undeserved status. At the same time, speech intended only to win might actually enhance autonomous judgment by placing claims in direct competition. To grasp these possibilities, one needs an account of speech that does not rely on speakers? intentions or motivations, since these only matter for the force of language by virtue of the structures that motivated speakers are able to project to exert influence.The model of communicative influence developed in the first half of the dissertation generates a series of arguments presented in the second half about the kinds of institutions that can best respond to the hazards strategic speech can pose to autonomous judgment. Rather than introducing mini-publics into chains of political decision-making in order to tap 29into the benefits of genuine deliberation, I ask how democratic goods can be secured at both the site and the system level through mixed types of communicative influence. This yields a much wider range of suggestions, while maintaining a role for mini-publics as environments in which institutional variables can be most directly controlled to draw out specific types of speech.The model of communicative influence and the idea of strategic speechDespite the problems with Habermas?s formulation of strategic action, I do use the term strategic speech in the dissertation. Without the assumptions about egocentrism, deception, illocutions and perlocutions that Habermas loads onto the concept, however, ?strategic speech? is more of a useful shorthand than a meaningful conceptual category. I use the term strategic to describe speech that directs its force towards some end beyond merely achieving understanding. What is theoretically more interesting than this orientation, I argue, is the variable magnitude and nature of this force. The model of communicative influence draws on Brandom?s pragmatics to build an alternative account of speech, including strategic speech. In the model, I distinguish among a number of ways speech can exert influence, each with different implications for autonomous judgment and for democratic systems more broadly. Rather than treat strategic speech as one among these types, I use the term to suggest that speakers aim towards some broader purpose. The important variation is not between strategic and non-strategic forms of speech, but in how speech is producing outcomes. Communicative Influence and AutonomyI have argued for abandoning Habermas?s concept of communicative action because of its untenable distinction between ?good? communicative action and ?bad??strategic action. 30Habermas?s conceptualization, however, does have the virtue of establishing the positive democratic potentials of talk and to possibility of shifting conflict onto the resources of language. The normative goal of the theory I develop is to establish that it is possible for speech to secure these same democratic goods without requiring communicative intent?that is, without excluding speech governed by the strategic incentives that go hand in hand with politics.I begin by providing an alternative account of the force of speech. On this account, speakers may still borrow force from hearers? agreeing to or endorsing claims. But speakers may also use this endorsement strategically to leverage further influence without necessarily treating people as means. In Chapter 5, I reintroduce into this framework normative considerations derived from democratic theory. I argue that the primary basis for the normative evaluation of communicative influence should be the autonomy of the judgments speakers try to influence through speech. Here, autonomy refers to the reflexive character of these judgments, and their authenticity?their grounding in reasons actors can recognize as their own. Building an account of autonomy into the model of communicative influence sets aside a subset of democratic theory?s most important questions?notably, the empowerment of those affected by an issue to have a say in decisions. Instead, the framework I develop in this dissertation focuses directly on the speech-based influence that precedes those decisions, perhaps even targeting them most forcefully when a decision is most empowered. Revisiting the idea of the force of language creates opportunities to establish which kinds of speech and conditions can underwrite autonomous judgment. This reconceptualization opens new 31possibilities for institutions to respond specifically and directly to threats to autonomy, arguments I develop in Chapters 7 to 9.The concept of autonomy is generally associated with intentionalist ethics, wherein autonomous persons are to be treated as ends, not means. Ultimately, while noting that intentions may determine the ethical standing of speakers?whether they are indeed treating people as means or ends?I focus on the consequences of speech. There are ethical dimensions to my argument that follow from the importance I place on autonomy. This said, it is not a work in ethics. The model of communicative influence locates these consequences at the level of the effects of speech on judgments, and that in turn drive changes in individuals? commitments. Theories of autonomy help determine whether these consequences are good or bad, but democratic theorists should be concerned primarily with the conditions that bring these consequences about. Kant recognized the value of such a two-track approach in Perpetual Peace. On the one hand, a theory of right serves to justify normative imperatives like the requirement to treat others as ends, rather than means. On the other hand, Kant also tells us that ?as hard as it may sound, the problem of setting up a state can be solved even by a nation of devils (so long as they possess understanding)? (Kant 1795 (1970), 112). In this remark, the task that interests him is one of bringing about certain forms of conduct, rather than consequences, by setting up neutralizing oppositions. The same impulse can be applied to the problem of setting up institutions that instantiate behaviors consistent with autonomy in practice, even against a background of self-interested incentives. 32Chapter OutlineEach chapter of this dissertation builds on the theory developed in the preceding one and, as a result, introduces greater complexity to the model and to its implications. It begins at the micro-level of pragmatics: first, to understand how assumptions about intentions are built into Habermas?s account of language, and how this can be avoided (Chapter 2); and second, to account for the force of language, and the variation in the nature and extent of that force (Chapter 3). It then introduces considerations of context to communicative influence, based on the idea that the character and extent of influence is not solely a property of claims and arguments, but of their interaction with a discursive environment (Chapter 4). I then reintroduce the democratic goods currently associated with speech-based politics. I use the lens of autonomy to ask how communicative influence affects these goods (Chapter 5). These three pieces?inferential structures, the discursive ecosystem and the democratic goods of speech-based politics?provide the conceptual foundation for the dissertation. I also discuss certain recognizable, generalizable patterns of communicative influence that we might recognize from everyday political communication: metaphors, analogies, narratives, hedging, deceit, and the like (Chapter 6). In the remaining chapters, I consider the implications of the model for existing democratic institutions as well as emerging democratic innovations. I argue that institutions can create the conditions that render communicative influence consistent with autonomous judgment: first, by controlling and correcting for contextual features that excessively limit or compromise reasoning processes (Chapter 7); second, by ensuring that when speakers? statuses increase their influence, the grounds for this magnification can withstand critical reflection (Chapter 8); and, third, by ensuring that institutions can select for an appropriate 33mix of types of communicative influence (Chapter 9). Finally, I illustrate the potential of the model to provide tailored, effective responses to threats to autonomous judgment that are particular to a given set of institutional arrangements. I suggest that political systems favouring different mixes of representative behaviors create different hazards of communicative influences. These, in turn, require different correctives to ensure that people have an adequate basis for autonomous judgment (Chapter 10). I conclude by identifying some further implications of the analytic framework the model of communicative influence offers (Chapter 11). Chapter 2. Habermas and Brandom: From illocutionary force to deontic scorekeepingChapter 2 begins with a closer examination of Habermas?s pragmatics. I focus on the concept of illocutionary force, highlighting the way it builds assumptions about speaker intent into the model of communication that ultimately underwrites deliberative democracy. I also explain how this formulation necessarily leads to the conclusion that strategic speech treats people as means, not ends. I then introduce and explain Brandom?s pragmatics, which provides the basis for an alternative account of the force of speech. Before proceeding to the model of communicative influence, I identify the necessity of moving beyond a baseline justificatory model that accounts for meaning and understanding, to an active, other-regarding model that accounts for the influence of speech on the commitments?the beliefs and actions?of others.Chapter 3. The model of communicative influence I: Inferential structuresIn Chapter 3 I introduce an alternative conceptual framework for understanding speech, using the notion of inferential structures. The starting point is the claim that 34articulating single assertional commitments, the fundamental units of Brandom?s model of deontic scorekeeping, is the most basic thing people can do with speech. When making an argument, for instance, speakers offer a series of inferentially linked propositions or commitments, leveraging accepted rules of inference to direct the reasoning processes of others, often towards certain conclusions. Chapter 3 develops the idea of communicative influence?influence that works directly on the commitments of others on the basis of the uptake of claims that are arranged in these inferential structures. I argue that these arrangements can vary in three ways. First, they can be more or less rigid, exerting pressure on reasoning and making the hearer take on certain commitments by virtue of the inferential connections embedded in the structure. Second, they can be more or less reciprocal, in the sense of engaging or appealing to the hearer?s existing commitments, and borrowing force from those commitments in order for influence to be successful. Third, the elements that make up the structure?the propositions and inferences among them?can be more or less transparent. Together, these three dimensions generate an eight-part typology of communicative influence, enabling useful analytic distinctions among types of speech that we intuitively believe are different: proof, persuasion, rhetoric, manipulation, explanation, discussion, assertion, and disclosure. Chapter 4. The model of communicative influence II: Discursive ecosystemsInferential structures represent only part of the model of communicative influence. In Chapter 4, I argue that the character and extent of communicative influence depends on the interaction between inferential structures and the context in which communication is embedded. I use the metaphor of an ecosystem to capture the ways that the parts of a complex system interact with speech to produce outcomes. The parts of this ecosystem 35comprise, first, the existing commitments speakers must engage with (or avoid) and the frames that establish connections and salience among them; second, subject-based features that establish credibility through trust or expertise, on the speaker side, and limit potential reciprocity by the absence of commitments, on the hearer side; and third, the institutional arrangements that incentivize certain types of communicative influence over others. I group the ecological effects of these features into three types. First, modification effects alter inferential structures directly, rendering them more or less reciprocal, transparent and rigid depending on whether or not they involve commitments that are part of a pre-existing structure like an issue frame. Second, magnification effects occur when inferential structures or their components are more likely to secure uptake because of something about the speaker,as opposed to something about the structure. Finally, selection effects capture the incentive structures institutions generate, and the costs or consequences they impose on certain communicative behaviors.Chapter 5. Communicative influence and democratic goodsChapter 5 reintroduces normative content to the model of communicative influence. It begins by demonstrating that the goods democratic theorists associate with discourse?the motivating force of reasons and freedom in will formation?are grounded in considerations of autonomy. Autonomy can, in turn, be disaggregated into three components: what is autonomous, in this case, a judgment, choice, or action; reflexivity, or the awareness of one?s own reasons for judgment, choice or action; and authenticity, or the requirement that those reasons be in some meaningful sense the autonomous actor?s own. The three elements of autonomy map onto the three dimensions of communicative influence. Rigidity limits choice and constrains judgment; transparency enables reflexivity; and reciprocity enables 36authenticity, in that the commitments exerting influence can be endorsed on one?s own terms. This leads to the normative claim that for communicative influence to be consistent with autonomy, increases in rigidity must be accompanied by increases in reciprocity and transparency. Following from this claim, I argue that features of the discursive ecosystem can be subject to normative evaluation in terms of autonomy because of their modification, magnification, and selection effects on communicative influence, the medium through which autonomy may be either supported or damaged.Chapter 6. Patterns of communicative influenceIn Chapter 6, I apply the model of communicative influence to commonplace features of political communication. Patterns of communicative influence are recognizable, repeated, generic characteristics of inferential structures. The patterns I group together have certain pragmatic features and functions in common. From these, I theorize their relationship to the three dimensions of communicative influence, to the immediate context of interaction, and to the discursive ecosystem more broadly. The five patterns are, first, narrative patterns, including storytelling and testimony; second, comparative patterns, including metaphors, analogies and examples; third, evocative patterns, including loaded language and persuasive definitions; fourth, deceptive patterns, including lies and exaggeration; and fifth, avoidance patterns, including hedging, equivocation, and non-answers to questions. In each case, I make claims about how these may alter inferential structures. For example, analogies may render them more rigid by establishing necessary inferences. Narratives may, in some cases, promote reciprocity by putting arguments in easily relatable terms. Deceptive patterns necessarily compromise transparency. In addition, the existence of these tendencies from each type of pattern means that their use has implications for the discursive ecosystem. For 37example, narrative patterns may establish credibility, while being caught in deception may undermine it. Chapter 7. Institutional responses to the anti-democratic hazards of modification effectsIn Chapters 7 to 9, I develop the idea that the way to ensure communicative influence is consistent with, or even enhances, autonomy is to direct institutional interventions at the discursive ecosystem to modify the conditions under which communicative influence takes place. The framework for these chapters is based on two premises. First, interventions can serve two purposes. On the one hand, direct interventions shape communicative behaviour directly by generating incentives or imposing limits on speech. On the other hand, through corrective interventions, to the extent that influence might threaten autonomy, arrangements are in place to counter its effects. Second, institutional interventions might be applied at the level of a specific site or venue of discourse, or at the level of what theorists have begun to call the deliberative system, which captures the interaction among sites of talk-based politics (Parkinson and Mansbridge 2012). Within this framework, I argue that institutions should respond to the specific anti-democratic hazards each type of ecological effect poses. In Chapter 7 specifically, I examine modification effects that derive from the structure of pre-existing commitments and audience characteristics. The presence of dominant frames in the discursive ecosystem, for example, enables rigidity by establishing pre-given connections among commitments. At the same time, as I argue in Chapter 4, they have the potential to compromise both transparency and reciprocity by rendering these connections implicit, and by excluding alternative considerations. The second set of hazards concerns the modification effects of audience-based ecological features?notably the reciprocity-compromising effects of lack of knowledge. In both cases, I suggest ways that rules, procedures, tasks, and other 38features of institutions can control and correct for these effects, and how sites might balance one another. Chapter 8. Institutional responses to the anti-democratic hazards of magnification effectsChapter 8 follows the same logic as Chapter 7, identifying direct and corrective institutional interventions at the site and system levels that respond to the specific anti-democratic hazards of magnification effects. Magnification effects occur when speaker-based ecological features?particularly the status or credibility conferred by expertise, first-hand experience, shared identity or trust?enhance the uptake of individual commitments, with or without an accompanying inferential structure. This status enhances the uptake of inferential structures, thus magnifying whatever kind of influence they would otherwise exert. I argue that the anti-democratic hazards specific to magnification effects occur when magnification effects are poorly connected to the grounds for credibility. Misplaced trust or misleading, mistaken claims to expertise, or selective appeals to a record enable influence that lacks the potential for authenticity or transparency. Direct interventions might include, for example, keeping a record of speakers? words and actions, or empowering sites to track the behavior of others. Correctives would ensure moments of reflexivity focused on the sources of credibility, and make sure actors are not forced to depend on speaker status for their judgments by providing accessible justifications as well.Chapter 9. Institutional responses to the democratic hazards of selection effectsAs with the chapters on modification and magnification effects, Chapter 9 begins with the anti-democratic hazards of selection effects. The most direct form of selection hazard is that institutions might select for manipulative influence?which is to say, they incentivize 39rigidity, along with non-transparency and non-reciprocity. Assuming that a site is meant to produce a certain type of influence, I suggest a set of direct institutional interventions that might, hypothetically, favour more or less of each of the dimensions of communicative influence. For example, raising and lowering the stakes may prompt greater or lesser rigidity in argument. The second hazard concerns the appropriate mix of influence types in a discursive site or system. I argue that autonomy is best served when communicative influence types balance one another. I explain why an excess of any one type of communicative influence is itself a hazard to the conditions of autonomous judgment, and suggest a set of corrective balancing relationships that can help determine the purpose of a discursive process, and its component parts. Chapter 10. Communicative influence and representationIn Chapter 10, I apply the model of communicative influence to a central feature of modern democratic systems, namely, representation. I take up Mansbridge?s theory of types of representation that exceed a simple principal-agent based theory. I argue that three types of representation?anticipatory, gyroscopic and surrogate?each generate their own particular hazards, which are ultimately hazards of the institutions that promote these types of representation. For example, in anticipatory representation, actors may attempt to change voter preferences ahead of the next election. Competition in such circumstances produces hazards that map onto those of selection effects; namely, that the pressures of competition might select for manipulation, or select for too much rhetoric and not enough of its counterpart, explanation. Institutional arrangements that are likely to channel electoral competition into anticipatory representation, then, should include correctives such as institutionalizing spaces for low-stakes discussion, to ensure that people can develop and 40reflect on their own commitments so that the basis for autonomy remains intact (although I do not address the link between institutions and representation types in depth). Similarly, institutions favouring gyroscopic representation generate the hazards of modification effects, in that actors may try to render certain sources of credibility or status claims more salient and establish, often implicitly, connections between their abilities and the demands of the role. Similarly, surrogate representation is likely to yield hazards of magnification. I use the correctives developed in Chapters 7 to 9 to suggest institutional responses to the hazards inherent in competition for representative status under a given set of institutional conditions. Chapter 11. ConclusionIn the final chapter, I elaborate some further implications of the model of communicative influence for democratic theory and for further empirical research. I summarize the argument of the dissertation, and I identify its five principle contributions: a platform for connecting democratic theory with empirical research in several key fields; a research agenda in its own right; an intervention into democratic institutional innovations; a strategy for thinking about deliberative systems; and a politicized theory of deliberative democracy.41Chapter 2. Habermas and Brandom: From Illocutionary Force to Deontic ScorekeepingThe first half of this dissertation develops a conceptual framework for differentiating among the ways that speech can exert force, whether or not actors are leveraging this force for strategic purposes. This framework will be grounded in pragmatics?a theory of what it is to use language. This chapter establishes the background in philosophy of language that sets the conceptual foundations for understanding communicative influence, beginning with an explanation of Habermas?s concept of communicative action. I then identify an important and fatal misstep in Habermas?s theory: he derives the persuasive force of speech from the illocutionary motives of speakers, rather than from the pragmatic consequences of speech acts. I show that the central, if implicit, status he assigns to intentions renders the problem of strategic speech insurmountable within Habermas?s framework. In the second section, I suggest that democratic theory move towards a better account of the pragmatics of language use. I introduce Brandom?s model of deontic scorekeeping as such an alternative. I explain the concepts most central to Brandom?s model of deontic scorekeeping in preparation for Chapter 3, in which I construct the model of communicative influence. Brandom?s approach displaces assumptions about intent in favour of rules of inference (logics of antecedents and consequences). In doing so, he sets out a pragmatic framework that can preserve the democratic value of reason-giving, while making room for the reasonable assumption that speakers? intentions may exceed the immediate end of reaching understanding. In the third section, I set up the model of communicative influence by suggesting that Brandom?s ?thin? model of language use needs to be expanded to provide analytic purchase on the ways that actors can use speech to exert influence, rather than 42merely to justify their own commitments and to understand those of others. Specifically, I suggest that when actors articulate claims, they may do more than merely authorize others to take them up. Rather, they can advance groups of interconnected commitments that leverage the force of linked propositions to produce commitments in others. The Pragmatics of Communicative Rationality and Communicative ActionHabermas begins The Theory of Communicative Action with an argument for communicative rationality, a form of rationality proper to discursive practice and fundamental to social coordination achieved through language. He develops this notion out of a concern with preserving the integrity of the intersubjective basis of social life in the face of the rising dominance of instrumental rationality. Communicative action both perpetuates and relies on language as the ultimate shared resource for social integration. Language operates on the rational basis of systematically interconnected universal validity claims. This internal connection between meaning and validity ensures that reasons can have ?rationally motivating? force, which in turn provides for autonomous will formation (Habermas 1996, 1987). The conditions of possibility for communicative rationality are the unavoidable presuppositions of argumentative practice (Habermas 1984, 17)?implicit in how people communicate even when they are not arguing?that can tap into the system of context-transcending validity claims. The perspective of someone engaging in argumentation, in which the only force is ?the force of the better argument,? is the only perspective from which the validity of claims can ultimately be established. To establish the validity of action norms, Habermas adds that this idealized rational discourse must include all those affected (Habermas 1996, 107).43Communicative action taps into this rationality simply by using the resources of the linguistic medium according to their proper end: that is, reaching understanding. However, communicative action requires certain performative presuppositions on the part of actors: assumptions that participants in communication have to make in order for it to make sense to be persuaded by each other?s claims. These presuppositions constitute what Habermas refers to as universal pragmatics. In communicative action, speakers give warrants that their claims are well-grounded, that is, that the claims? validity conditions are met. The validity basis of a claim can be its truth, grounded in the objective world of facts; its rightness, grounded in the social world of norms; or its truthfulness (or sincerity), grounded in the subjective world of inner states. Warrants to truth claims can be redeemed in theoretical discourse, and to rightness, in practical discourse. One of the presuppositions speakers must make when acting communicatively is that their validity claims would be convincing under idealized conditions of theoretical or practical discourse. Communicative action refers to the dynamic process of reaching agreement on the basis of the validity of an utterance?agreement that is motivated by the persuasive force of valid claims, and achieved rather than imposed. The acceptance of a speech act offer on the basis of its connection to a criticizable validity claim renders communicative action consistent with autonomous opinion and will formation (Habermas 1979, 1984). Illocutionary force, communicative intentCommunicative action achieves coordination through rationally-motivated agreement. As I described briefly in Chapter 1, Habermas explains the persuasive force of communication with reference to the distinction, borrowed from speech act theory, between locutionary, illocutionary, and perlocutionary components of an utterance. The locutionary 44component refers to the propositional content of an utterance, or what is said. The illocutionary component refers to what the actor is doing in saying something: for example, asserting, stating, affirming, denying, or commanding (Habermas 1984; Austin 1975). The illocutionary component is the only part of a speech act offer that can motivate acceptance (Habermas 1987, 68). That is, when an actor accepts the illocutionary component of an utterance, he or she understands or accepts it as an assertion, a warning, an affirmation, or a command. The nature of the illocutionary offer determines the type of validity claims that could motivate its acceptance (Habermas 1984, 308). In accepting a speech act?s illocutionary offer, a hearer accepts that the validity conditions appropriate to that type of speech act are met?that it asserts a true fact or commands in accordance with a valid norm. A speech act?s illocutionary force derives from this acceptance, along with whatever obligations are incurred as a result. The illocutionary roles that, for Habermas, can establish the validity basis of speech include: first, constatives, which assert or represent a state of affairs, and can thus be contested on grounds of truth; second, regulatives, which refer to something in the shared social world, establishing an interpersonal relationship that can be contested in terms of its rightness; and, third, expressives, which refer to something in the subjective world of inner states (Habermas 1984, 325-26).4  In accepting the illocutionary component of a speech act 454 Habermas?s typology of illocutions also includes imperatives, which drop out of the analysis because they are only expressions of will; communicatives, which organize speech according to whether something is an objection, answer, question, prediction, etc; and operatives, which merely apply rules, for instance by classifying or identifying.offer on the basis of these criticizable validity claims, interlocutors take on commitments that are relevant to the ongoing interaction.5By relying on speech act theory in this way, Habermas makes the meaning of an utterance dependent, in part, on the speaker?s intent. The illocutionary intent, or what the speaker is trying to do with the speech act, is what renders communicative utterances self-sufficient or intelligible. Meaning is internally connected the intention of the speech act offer, because the illocutionary intent (whether something is meant to represent a state of affairs, establish a relationship, or express a subjective state) determines the validity basis on which hearers should accept (or reject) the speech act offer. Understanding a speech act requires understanding what it would take to validate or falsify it, and what it would take (that is, its validity basis) is determined by the illocutionary intent. Action is communicative when speakers ?unreservedly pursue illocutionary aims? (Habermas 1984, 305). In order for illocutionary force to hold?that is, in order for actors to agree to a speech act?s validity and incur rational obligations as a result?actors must adopt the performative stance that they and others are pursuing only the ends implied by the illocutionary component of their claims, and not ends tied to the perlocutionary consequences of illocutionary force being successfully applied. Following Habermas?s logic, if the intention reaches beyond the stated illocutionary content, interlocutors would need to understand that further intention in order to understand the meaning of the claim? which is difficult because it is only ?contingently related? (1984, 289) to the content of the utterance. In Habermas?s framework, this type of language use constitutes strategic action, in which 465 Illocutionary types do not themselves determine which action category a speech act belongs in, although certain types of illocutions tend to go with certain action types?for instance, expressive with dramaturgical action. Communicative action is distinguished, in that actors seek to reach understanding on the basis of validity claims, the types of which are determined by the illocutionary component of an utterance.speakers instrumentalize speech acts in order to produce perlocutionary consequences. For such a claim to have force, however, the hearer would still need to accept the illocutionary component of the speech act offer. Habermas takes a further logical step in stating that in order for such speech to produce its consequences, the hearer must be unaware that intentions exceed the illocutionary component of the speech act.It is this move that excludes strategic speech from the realm of normative acceptability. Because strategic speech requires an element of concealment, in which hearers? naive acceptance of illocutionary offers can harness their wills to the speaker?s ends, acceptance of the speech act cannot claim the basis of rational agreement. Following Habermas?s logic, treating speech acts as means rather than ends entails treating subjects as means rather than ends. Moreover, political circumstances often yield cases where language is being used with an orientation to success or to producing a particular outcome, and generally both parties to the exchange as well as their audiences know it. And yet even in such situations, speech is nevertheless expected to have some impact on judgment. The idea of illocutionary force as it is currently formulated offers no way of accounting for the force or influence of speech in such circumstances.What is the alternative to this intentions-based conception of communicative action? I argue that it is possible to avoid this dependence by turning to an alternative pragmatics of language use: Robert Brandom?s model of deontic scorekeeping. In a 2008 interview, Brandom directly suggests extracting political consequences for political theory by a method similar to that of Habermas, but starting with his own very different understanding of language (Brandom in Pritzlaff 2008, 365). The goal is to displace intentions from their central, if implicit, status as determinants of the democratic quality of speech. 47According to Brandom, a philosophy of language like Searle?s, on which Habermas builds his account of illocutionary force, is based on agent semantics (1998, 147).6  Agent semantics ground meaning in how the speaker intends to be understood. The conceptual content derives from the speaker?s attitude that the utterance is an expression of a belief, combined with the intention that it be understood as an expression of that belief (Brandom 1998, 147). Thus, agent semantics depends upon the prior content intentional states to explain how speech acts can have meaning (Brandom 1998, 151). The alternative to agent semantics is a functional semantics. A functional semantics ascribes content in the first instance to linguistic acts by virtue of the role they play in a system. The system in question is a system of discursive or linguistic practice, where the function of a unit depends on the ?proprieties of input and output, antecedents and consequences? (Brandom 1998, 147). Brandom argues that we should understand intentional states in terms of linguistic practice, and not the other way around. We cannot understand what it is to have and to articulate a belief without also understanding what it would mean to use that belief in linguistic, discursive practice (1998, 152). Assigning priority to linguistic practice over intentionality at the level of meaning opens the door to the possibility of a force of language independent of intentions. In fact, Brandom achieves what Habermas sets out to do: namely, loading the work of social interaction onto the resources of language, rather than intent. However, Brandom?s account of linguistic practice is thin. It is geared towards merely 486 Habermas does acknowledge intentionalist or agent-based semantics within his own framework, but in the context of indirectly producing understanding by way of context. When speakers do not yet have sufficient shared understanding of a situation to achieve coordination on the basis of agreement, they can produce understanding through perlocutionary effects. Intentionalist semantics treat perlocutionary effects as ?giving someone to understand? something (Habermas 1984, 331). Yet Habermas fails to see how intentionalist semantics turn up in the very idea of the illocutionary component of any speech act offer, rather than simply its perlocutionary forms.providing an account of the fundamental characteristics of what it is to use language?which, he argues, is conceptually prior to rational agency, to having beliefs, and to undertaking actions. A thicker account of linguistic practice might seek to capture a greater range of what competent speakers can use language to achieve?including to influence one another?s commitments and actions. I suggest replacing Habermas?s concept of communicative action?which is tied to the intentional component of illocutionary force?with Brandom?s model of deontic scorekeeping as the basis for determining the democratic value of communication.Robert Brandom?s Model of Deontic ScorekeepingIn Making it Explicit (1998), Brandom introduces a pragmatic model of discursive practice as deontic scorekeeping. The purpose of this section is to explain this model, which provides many of the concepts I use throughout the dissertation. The basic idea of the model is that in discursive practice, actors have a constantly evolving set of deontic statuses, having to do with the set of assertions and actions to which they are committed and entitled. Because commitments are related to eachother inferentially, deontic statuses are not limited to what the speaker has explicitly articulated; they also include those to which he or she is further committed by virtue of the commitment-preserving (must) and entitlement-preserving (can) inferential relationships among claims (1998, 169). When people communicate and reason within one another, they do so by keeping track of their own and one another?s commitments, hence Brandom?s term ?deontic scorekeeping.? Disagreements occur when there is an inconsistency in the ?score,? that is, when the commitments an actor attributes to himself or herself are not the same as the ones that others attribute. When this happens, actors must make explicit the relevant commitments and relationships among them (1998, 178). Figure 492.1 provides a visual representation of Brandom?s model of deontic scorekeeping, with each concept explained in detail below.Figure 2.1. The Model of Deontic ScorekeepingDeontic statuses: Commitment and entitlementA deontic status refers to either a commitment or an entitlement. A commitment is something that is, or could be, asserted. Commitments can be beliefs about a state of affairs that obtains (which Brandom calls ?doxastic commitments?) or intentions to bring about a state of affairs (which Brandom calls ?practical commitments?). Doxastic and practical commitments?what one takes to be true, or intends to make true, respectively?are the 50fundamental units of the model of deontic scorekeeping. To undertake a commitment is to make a claim, typically in the form of an assertion, and this act is at the heart of discursive practice. Speech acts that perform different functions?functions Searle and Habermas, among others, seek to categorize?are ultimately derivative of the constitutive practice of claim making (Brandom in Pritzlaff 2008, 379). While Brandom acknowledges the distinction between practical and doxastic commitments, and while practical speech acts and practical commitments may be especially salient in political institutions (Brandom in Pritzlaff 2008, 379), Brandom?s baseline model concerns discursive practice based on declarative or assertional (doxastic) commitments. The deontic status of being committed or entitled to a belief or intention determines the ?score? that participants in ?the game of giving and asking for reasons? are tracking. The score is determined by the previous score?previous commitments and entitlements?and the acceptability of the inferences that led to the current commitments. An actor is entitled to a claim if he or she has reasons for it, has undertaken other commitments that lead to it by accepted patterns of inference, and has not taken on any incompatible commitments.Deontic attitudes: Undertaking and attributingDiscursive practice is essentially social, and thus requires a social perspective. The deontic score of the players is a property of their attitudes towards themselves and others. Deontic attitudes involve both undertaking and attributing deontic statuses?that is, commitments and entitlements. To undertake a commitment involves acquiring and acknowledging new commitments, taking responsibility for them, and keeping track of the way they change the score. Actors then must treat themselves as having those commitments, and draw further inferences consistently, according to whether or not having that 51commitment necessarily entails, generates entitlement to, or precludes another. Actors also attribute commitments and entitlements to others (1998, 166). When there is a disconnect between commitments undertaken and commitments attributed?when people view the score differently?participants in discursive practice are forced to make their commitments and their inferences explicit. In undertaking a commitment, actors also take on a responsibility to demonstrate entitlement to that commitment. Moreover, by asserting a claim, they authorize others to undertake that claim themselves. Patterns of inferenceDoxastic commitments are connected by patterns of inference, logical if/then relationships of three types. First, committives are commitment preserving: if someone is committed to X, he or she is necessarily also committed to Y. Second, permissives are entitlement preserving: if someone claims X, he or she may also claim Y. Third, incompatibility relationships establish that someone who is committed to X cannot commit to Y. Together, the inferential relationships among commitments generate a justificatory structure or web that confers meaning on individual commitments by establishing which claims can be given as a reason for which further claims or actions (Brandom 1998, 90, 188; Fultner 2002, 123). This idea represents a more intuitive understanding of how dialogue unfolds than Habermas?s metaphor of grounding. Interlocutors can challenge not just the commitments and inferences that constitute the sources of entitlement to a claim, but also those that necessarily follow from it, as conversations spread to cover related issues or consequences that might follow from the commitment, as well as the reasons for the commitment. The idea that commitments have both entailments and justifications (that is, both inferential antecedents and consequences) also points towards the possibility of speakers 52?racking up? commitments that ultimately might compel them to undertake other specific commitments.People reason from states of the world?doxastic commitments?to practical commitments by patterns of practical inference. These types of inference follow linguistic patterns of ?should? reasoning, sometimes crystallized in normative concepts and evaluative categories. Patterns of practical reasoning can be of three types. First, prudential oughts follow the logic of preferences or desires. An actor, committed to a certain state of affairs, reasons to an action based on his or her preferences. Second, institutional oughts use norms to reason from states of affairs to action. Some kinds of institutional roles or statuses come with widely-shared norms attached that dictate action, given a certain state of affairs. Third, unconditional oughts refer to moral imperatives that apply universally, regardless of preference (245-47). This aspect of Brandom?s model offers two advantages over Habermas?s theory of communicative action. First, it incorporates norm-based and instrumental patterns of reasoning into the same baseline model of reasoning, seeing them as a different form of the same type of discursive practice rather than fundamentally different orientations to the world and to language. Second, Brandom?s recognition that people reason from states of affairs, via patterns of ?ought? reasoning, to action opens the possibility that interlocutors can challenge practical commitments either at the level of preferences, norms, or morals, or at the level of beliefs about states of affairs. Habermas acknowledges a similar possibility when he suggests that prior to being able to engage in communicative action, actors may have to negotiate a common situation definition if one is not already sufficiently shared (Habermas 1984, 331). However, where Habermas relegates this possibility to (in his view) the lesser domain of 53strategic action and perlocutionary consequences, the possibility that disagreement can operate at either or both of these levels fits easily within Brandom?s overall model. At a more practical level, Brandom?s framework reflects the intuition that, in political life, disputes are often not over principles, but rather over the states of affairs to which they are said to apply?and that competing understandings about the world can lead to quite different outcomes, albeit through application of the same principle.Inferential articulation: Communicative and justificatoryFor Brandom, there are two types of moves in the game of giving and asking for reasons?two ways in which a person can acquire new commitments. The first type of move is the intrapersonal, intercontent (1998, 186) acquisition of commitments. This move occurs within a single actor?s inferential web, as when the same speaker demonstrates entitlement through reference to other content within his or her set of commitments. Within a single actor?s reasoning (intrapersonal), he or she can make inferences, acquire new commitments and entitlements, and grasp the connections among his or her commitments (intercontent) in terms of the proprieties of inference that connect the universe of possible beliefs and actions. This internal reasoning type of move need not be explicit; rather it refers to the process of keeping track of inferential consequences and adding and subtracting commitments. One commitment may necessitate some further commitments and entitle an actor to others (169). This type of inferential articulation is justificatory: it involves recognizing how one commitment can serve as a reason for or necessitate another.The second type of move involves the interpersonal, intracontent acquisition of a commitment: in short, acquiring a commitment from someone else through communication. Making an assertion is also the fundamental move in interpersonal communication. ?Putting 54a sentence forward in the public arena as true is something one interlocutor can do to make that sentence available for others to use in making further assertions? (170). This type of communicative move works through authorization and responsibility, which enable interlocutors to inherit entitlements to commitments from one another. In communicative inferential articulation, the speaker authorizes the claims that follow from his or her assertion. For Brandom, this type of move in the game of giving and asking for reasons is interpersonal, intracontent because when actors undertake commitments as a result of another speaker?s authorization, the commitment travels ?in one piece,? and only subsequently alters the hearer?s own set of commitments through intrapersonal, intercontent (justificatory) articulation (Brandom 1998, 175).55Table 2.1. Summary of Brandom?s Deontic Scorekeeping ModelDeontic Scorekeeping ConceptExplanations and Sub-typesDeontic StatusesCommitment and entitlementDeontic AttitudesUndertake and attributeWhen actors make commitments?They undertake responsibility to demonstrate entitlement and authorize further assertions for themselves and others.Commitment TypesDoxastic (beliefs) and practical (intentions)Patterns of InferenceConditional if/then language: Committive (if x, then necessarily y) and permissive (if x, then possibly y) patterns of inference, which are ?entitlement preserving?; and incompatibility (if x, then not y). Patterns of Practical InferenceNormative should/shall language (where should is an accepted pattern of reasoning from a state of the world to an intentional state, and shall is the intentional state or practical commitment). Patterns of practical inference can be of three types: (a) prudential ought (model of preferences or desires); (b institutional ought (model of norms for which there is shared endorsement that they apply to someone with a given status or role);(c) unconditional ought (entitlement preserving for everyone, blind to preferences. This is a moral ought).Discursive Practice is Communicative when...Commitments are inherited ?interpersonal, intracontent.? This is to say, when one person undertakes a claim put forward by another, the content of that claim does not transform. Discursive Practice is Justificatory when...Commitments are related ?intrapersonal, intercontent.? This is to say, claims are related to one another in terms of proprieties (or accepted rules) of inference. From Thin to Thick Models of Language UseThe model of communicative influence I introduce in the next chapter builds on these core concepts. However, I argue that Brandom?s is a minimalist theory, including only those concepts and assumptions necessary to account for reasoning and meaning in terms of discursive practice. I use this as a foundation for a thicker account of language use, focusing on the things actors can do using the articulation of commitments, including argument, 56explanation, and rhetoric. The idea of authorizing inferences is my starting point. One speaker?s assertion has an effect on the commitment structures of others, and alters the score because it implicitly authorizes further claims. I argue that the idea of authorizing and inheriting claims in an ?interpersonal, intracontent? way is insufficient to understand the force of speech, and the many things that strategic actors can do to leverage this force. Among other things, Brandom?s thin model does not help us to understand the force of argument  as opposed to assertion. The question of what it would take to adapt deontic scorekeeping to the understanding of how speakers use language in argument points towards a thicker model of language use. Both Habermas and Brandom are interested in a basic account of language use?a theory of the conditions of possible understanding (Habermas) and grounding of the normativity of social practice (Brandom). In both cases, justification and the possibility of justification are fundamental to language. I argue that a thin account of language is a good foundation, but is insufficient for understanding influence. Beyond the practice of justification, how do Habermas and Brandom envision the force of language? As I discussed above, for Habermas, the coordinating effects of speech rest with illocutionary force. Speakers make speech act offers of various types, which hearers can either accept or reject on the basis of their validity, and in doing so they incur commitments. Similarly, for Brandom, in interpersonal communication, inferential articulation between commitments?which can generate changes for an actor?s score?temporarily ceases in favour of interpersonal, intracontent inheritance of claims. Speakers authorize others to undertake the commitments they articulate, and for which they take responsibility. I argue below that the idea of inheriting commitments is insufficient for understanding language use, and i particular strategic language use in politics. In some cases, particularly where strategic 57motives guide action, speakers do not merely authorize others to undertake commitments, but in some way compel them to do so.Both Habermas and Brandom also theorize a distinction between everyday communicative practice and discourse, a reflexive stance towards the generative structures of speech. In Brandom?s terms, when we make a commitment, we undertake responsibility to demonstrate entitlement for that commitment, and authorize others to take it up and make inferences from it. These aspects of communicative practice?responsibility and authorization?remain implicit unless actors encounter a discrepancy in the score. Commitments and proprieties of inference must be made explicit when people encounter a discrepancy in the score. Justificatory practice makes the previous scores and proprieties of inference that lead to a commitment (or set of commitments) explicit, much like argumentation serves as a court of appeal for testing the validity of claims in Habermas?s theoretical framework. Missing from Brandom?s argument is an account of the opposite end of the spectrum from justification?a reflexive, deliberate, explicit form of communication that lends force to authorization, allowing it not just to justify new commitments, but to compel or induce them. Figure 2.2 illustrates these claims.Figure 2.2. Implicit and Explicit CommunicationUnderstanding this dynamic requires abandoning the premise that the inheritance of commitments in communication is solely interpersonal, intracontent. That is to say, what is 58communicated is not necessarily a single, discrete claim from which the hearer then draws his or her own inferences. Rather, we also communicate larger inferential structures. These function to ?rig?? antecedents and consequences so that specific reasoning pathways follow from putting forward a commitment, along with a surrounding inferential structure. Just as, for Habermas, argumentation serves as a court of appeal that enables us to speak meaningfully about the validity of claims, at the explicit end of the communication spectrum that is oriented towards producing rather than grounding commitments, argumentation can be a source of influence?and this is particularly the domain of politics, and that of strategic speech. Exerting communicative influence is not the same as demonstrating entitlement. The concept of communicative influence is similar to justification, in that actors take an active stance towards inferential structures and prior commitments, but it produces new commitments in others rather than grounding a speaker?s own commitments. The practice of argumentation drives the idea of communicative influence, although I show in Chapter 3 that communicative influence does not have to be argumentative. New commitments are produced, not as a consequence of the speaker?s intent, but as a consequence of what must and can follow from the inferential structure that a speaker articulates. Chapter 3 takes up these ideas to develop a detailed model of communicative influence, which operates through these inferential structures, and on the commitments of others. By producing certain commitments, I argue, these inferential structures are what enable language to exert force beyond the simple consequences of undertaking an authorized claim; however, the nature of these inferential structures?how they exert force and produce consequences on the commitments of others?can vary in important ways. Moreover, just as 59within Brandom?s model the exact consequences of authorizing a commitment through assertion depend on the existing score, I argue in Chapter 4 that communicative influence ultimately depends on the interplay between these inferential structures and the discursive context in which interlocutors are embedded. Ultimately, rather than targeting speakers? behaviors and the inferential structures they produce, this contextual variation might enable interventions to ensure that communicative influence is consistent with autonomous judgment. 60Chapter 3. The Model of Communicative Influence I: Inferential StructuresThe model of communicative influence developed in this chapter offers a way of explaining the force of language. By focusing on the dynamic structures through which speech produces new commitments by imposing inferential constraints and leveraging existing commitments, the model avoids having to rely on speakers? motives to determine the democratic value of reason giving, and allows us to focus instead on the pragmatic consequences of speech for democratic goods. The concept of communicative influence is intended to capture the ways speakers can use words and arguments to shape the commitments of others?both to beliefs and to actions?by drawing on and directing the pragmatic, inferential consequences that follow from getting someone to make a commitment, even if that starts with the simple acceptance of a claim.In Chapter 5, I will specify the democratic goods normative theorists attribute to talk-based politics. Here and in Chapter 4, I identify how features of speech itself, on the one hand, and the discursive context in which it is embedded, on the other, generate force. This framework will allow me to articulate a theory of how speech produces its pragmatic effects, and to connect the issue of how utterances are effective (as opposed to the intentional question of why they are made) to normative accounts of the democratic goods of speech. Though the strength of incentives under the conditions of politics often preclude directly altering speakers? behaviour and the inferential structures they produce, it is possible to envision ways of arranging institutional environments to get the most out of communicative influence, while correcting for its more harmful tendencies. The model of communicative influence opens up a range of possibilities that extends beyond current approaches in 61deliberative democratic theory. This chapter provides the conceptual foundation for the subsequent development of such a theory.The force of speech operates through inferential structures: packages of assertions or commitments and the inferential links that connect them. In explaining inferential structures, and communicative influence generally, I use the terms ?speaker? and ?influence target? or ?target? throughout the chapter and the dissertation to refer to the source of the inferential structure (the actor trying to exert influence) and its target (the actor who is subject to the influence). These inferential structures vary in how they secure influence. Notably, the amount of force a speaker builds into the structure can vary, as can the level of engagement of the influence target?s own commitments. These three dimensions generate eight ideal types of communicative influence. Unlike communicative action (which asks how people can achieve sufficient understanding to provide a rational basis for social coordination) and deontic scorekeeping (which asks what it is to have a belief or practical commitment), communicative influence is a generic concept that gives analytic purchase on the question as to how people use speech to modify the commitments of others. I show how the model of communicative influence pays off later in the dissertation. In this chapter, I explain the conceptual framework in detail, first by identifying the three dimensions of variation in communicative influence?rigidity, reciprocity, and transparency; second, by characterizing the eight types of communicative influence; and finally, by highlighting some of the theoretically significant ways these types relate to one another. 62Communicative InfluenceCommunicative influence refers to the capacity of speech to alter the commitment structures of others?to induce them to adopt certain beliefs or take certain actions. This type of influence is communicative because it operates through speakers putting forward claims that others take up and from which they reason. As in Brandom?s thin model, consequences in targets? webs of commitments may be the product of taking up a single assertion. However, I argue that assertions are only the simplest, most basic source of influence. Particularly when actors have a strategic interest in using language to secure an outcome, they may use inferential structures?packages of inferentially linked commitments?to extend their influence. A syllogistic argument, for instance, is an inferential structure comprised of premises and a conclusion. An inferential structure can exert communicative influence by leveraging agreement with some or all of its components into an arrangement that directs an influence target to adopt certain commitments, potentially producing far ranging consequences for the target?s overall commitment set. While their directive force may derive in part from their formal logical properties, I demonstrate below that there are other ways inferential structures can generate and direct force, even where their properties may actually violate criteria of formal logic. These inferential structures may be more or less explicit but, once taken up, they modify commitments from within. Their capacity to do so, and the way in which they achieve this effect, depends on the both the nature of the inferential structure?addressed in this chapter?and features of the discursive context in which it is embedded, including traits of speakers, hearers, and institutional environments?addressed in the next. 63Communicative influence is a more general concept than strategic speech, in that it does not involve assumptions about the speaker?s orientation to others, or to certain ends. Rather, actors with any kind of intent may use inferential structures to secure those ends by producing a given effect on the reasoning processes of others. Their strategic intent may start and end at changing the commitments of others, and in that case the possibility that they can tailor inferential structures to suit that end does not necessarily exclude the influence itself from democratic merit. Such behavior is consistent with the full range of influence types identified below. The concept of communicative influence recognizes that careful, deliberate, even strategic planning?in the sense of anticipating the beliefs, preferences, and likely choices of others (Schiemann 2000)?may go into choosing the right inferential structure for the task. It is variation in how inferential structures produce their effects that determines the democratic quality of influence; that is, its consistency with the exercise of autonomous judgment, rather than the speaker?s attitude towards the hearer. For example, choosing examples to promote the uptake of a given commitment could be consistent with either thoughtful persuasion or outright manipulation, depending on how those examples are situated within inferential structures. The sources of variation in communicative influence depend on differences in inferential structures. These sources of variation serve to establish a set of conceptual categories more fine-grained than Habermas?s communicative/strategic action dichotomy. The analytic leverage these categories provide will give democratic theory the tools to make normative differentiations not only among particular instances of 64communicative influence, but also among the conditions that promote better or worse types of influence and condition their effects.7Dimensions of Communicative InfluenceThe notion of ?the force of the better argument? (Habermas 1984, 25) implies that some arguments are stronger and others weaker, and that the former have the capacity to override the latter?but that the consequences of this overpowering effect are the result of understanding and acceptance, rather than inducement (Pellizzoni 2001). The purpose of articulating a concept of communicative influence is to capture what this force actually entails. The key idea is that when influence targets buy into part or all of an inferential structure, their undertaking of its commitments has the capacity to produce further commitments and entitlements, and invalidate others, with potentially far ranging consequences.The features of the inferential structure that exert this influence vary along three dimensions that will be important for distinguishing better and worse effects from the standpoint of democratic goods. The first dimension, rigidity, describes the logical structure of speech. A rigid inferential structure, such as a syllogism, derives force from the arrangement of the commitments it contains, and the relationships among them. Communicative influence influence rests on the uptake of these commitments. The second dimension is reciprocity. Reciprocity concerns the extent to which inferential structures rely on, require, or draw in interlocutors? existing commitments. The force of a reciprocal 657 Klemp also offers a typology for understanding speech?s capacity to secure agreement. However, he takes rhetoric as a general category that is roughly equivalent to persuasive speech, of which ?deliberative persuasion,? ?strategic persuasion,? and ?manipulation? are subcategories. I address the differences between Klemp?s approach and my own in the introduction.inferential structure derives from its resonance with these commitments. Arguments are never put forward in a vacuum and, as scholars of rhetoric note, speech is often most compelling when it resonates with existing beliefs (Perelman and Olbrechts-Tyteca 1969, 19; Delia 1970). The third dimension is transparency. Transparency captures the extent to which elements of the inferential structure?its premises, its conclusions, and the proposed logical relations among them?are rendered explicit and thus made available for evaluation.Taken together, rigidity, reciprocity and transparency?and their opposites?characterize communicative influence in a way that opens it up to normative evaluation and, ultimately, institutional interventions. The effects of each dimension are as follows. Rigidity concerns the magnitude of influence: how much force the inferential structure exerts on the reasoning that follows from it. Reciprocity concerns engagement and buy-in, but also control over reasoning. With increased reciprocity, the speaker may give up some measure of control as the reasoning process becomes more of a joint intervention, with each step subject to external validation or endorsement. However, the control the speaker gives up is compensated for by the force derived from an increased buy-in. Transparency is the property of inferential structures that determines the extent to which the commitments and inference upon which communicative influence depends?whether its force derives from logical connections, external buy-in, or both?are made explicit or left implicit. The remainder of this section explains each of these dimensions, laying the groundwork for identifying eight ?ideal types? of communicative influence. These three dimensions of judgment have complex implications for questions of autonomy, which I examine in detail in Chapter 5. 66RigidityRigidity describes the extent to which the inferential structure, should the commitments that make it up be accepted, necessarily entails a given conclusion or directs reasoning towards a limited set of beliefs or actions. In Brandom?s terms, rigid inferential structures take advantage of committive proprieties of inference (if you accept X, then you necessarily accept Y) and incompatibility relations (if you accept A, then you cannot also hold B).To illustrate the idea of a rigid argument, consider Lewis Carroll?s ?What the Tortoise Said to Achilles?:?Now that you accept A and B and C and D, of course you accept Z.	  ?Do I?? said the Tortoise innocently. ?Let?s make that quite clear. I accept A and B and C and D. Suppose I still refuse to accept Z??	 ?Then logic would take you by the throat and force you to do it!? Achilles triumphantly replied. ?Logic would tell you ?You can?t help yourself, you must accept Z.? So you?ve no choice, you see?? (Carroll qtd. in Winch 1958, 56).Although the Tortoise points out that, in fact, he still must not accept Z, the idea that logic could ?take someone by the throat? and force them to acceptance effectively captures the concept of rigidity. Actors are compelled to either take on a new commitment or risk incoherence. No such compulsion exists with non-rigid inferential structures. In such cases, speakers make assertions or sets of assertions in ways that leave the hearer free to take them up themselves, or not, and alter their own commitment structures as they see fit. The two 67dynamics here are, on the one hand, a constraining effect of rigidity on the inferences that can be drawn from speech and, on the other hand, the enabling effect of non-rigid speech, which works by opening up new inferential pathways. Non-rigid influence most closely resembles Brandom?s baseline model, where people articulate commitments and sometimes demonstrate entitlement. In making an assertion, for example, speakers authorize the uptake of others, but the commitment itself does not compel or demand uptake. By contrast, as Achilles and the Tortoise illustrate, the paradigm case of rigid communicative influence is an argument. Arguments work by establishing logical or inferential connections among commitments such that by accepting some of them, targets are induced, by virtue of those connections, to accept further commitments.Does the dimension of rigidity merely smuggle intentions back into the framework under a guise different from that of illocutionary force? A speaker?s success in building a rigid inferential structure may require the intent to produce a certain commitment. For instance, it is hard to imagine someone unintentionally constructing a correct syllogism. However, a speaker can intend that the target make a commitment?adopt a belief or take some action?without necessarily employing rigidity to achieve this outcome. The intention to produce an outcome is consistent with a variety of ways speech can exert force. The normative evaluation of this influence should depend on how it secures uptake, and the model of communicative influence gives conceptual traction to the nature of this variation. Having an intention to produce or alter a commitment may be a condition for producing a rigid structure, but rigidity itself is not defined by this intention. Moreover, intention may not even be a necessary condition for rigidity. As I argue in the next chapter, following certain 68common patterns of inference?for instance, analogies or loaded definitions?can add rigidity to an inferential structure without the speaker necessarily having intended it. 	 To further illustrate the concept of rigidity, consider a hypothetical discussion about policy responses to climate change. It might contain a single assertion, for instance, that climate change is caused by greenhouse gas emissions. There is an implicit inferential structure behind this assertion, namely, evidence, or a theoretical explanation of the connection. It accounts for entitlement to the claim, but?particularly when it is implicit?exerts no force, except insofar as the target uses those same grounds to attribute entitlement, and thus becomes more likely to undertake the claim himself or herself. Undertaking the claim that climate change is caused by emission could lead to a range of other commitments; that, for instance, governments need to do something about it, and what in particular they ought to do. But by itself, the assertion and the implicit structure that renders it intelligible is not exerting force. However, if the speaker combines the original assertion with the claims that climate change will have severe and irreversible consequences, and that intervention could make a difference, then a structure begins to emerge around the claim that drives reasoning towards a particular conclusion?namely that something needs to be done. Rigidity does not guarantee the force of an inferential structure across contexts. Any of the commitments and inferences that make up a given inferential structure, while perhaps consistent within the context of the inferential structure itself, may prove incorrect when an interlocutor reveals a new piece of information. For instance, an interlocutor might challenge the argument that ?something needs to be done about climate change? with the claim that any available measure would cause short term harm. In doing so, he or she would neutralize that piece of the inferential chain. Depending on which commitment a target rejects, more or less 69of the inferential structure may fall apart. For instance, if, instead of focusing on the possible harms of action, an interlocutor were to offer data that conclusively denied the existence of climate change?the starting premise of the above example?he or she would have neutralized any of the force deriving from subsequent inferential connections. Although articulating a set of claims necessarily opens them up to challenge, whether in terms of their internal coherence or external validity, rigidity is a useful way of describing the structural arrangements of the commitments and inferences that make up an argument. An inferential structure is rigid if each claim that is successful?that garners agreement or uptake?is able to direct that success to some further inference. The possibility that one of the commitments that makes up an inferential structure may fail to secure uptake does not make the original structure less rigid, in the sense of its design leveraging potential uptake into influence. By contrast, an assertion without such a rigid structure surrounding it can generally do little to pre-determine the inferential consequences that follow from its uptake. To the extent that even a simple assertion can or does work to shape the consequences uptake has for reasoning?for example, a simple comparison?it begins to take on rigid qualities. Returning to the example of climate change, a speaker might even argue that some specific action needs to be taken. He or she might, for instance, argue for controlling carbon emissions through a cap and trade system, for standards and regulation, or for individual behaviors. While the argument is still rigid, in that it directs the target to a conclusion?we should do something, therefore we should do x?the final piece of the structure is not solid enough to generate force for the final step. The inference from doing something to doing one specific thing requires endorsement on a number of fronts if the target is to accept it: that the measure would be effective, feasible, or consistent with other values and priorities. These 70considerations suggest the importance of the next dimension of communicative influence: reciprocity.ReciprocityReciprocity refers to the potential for an inferential structure to be redirected at a given stage as a result of a weak, incomplete or porous links that requires buy in on the basis of the target?s other commitments, external to the inferential structure. Reciprocity conveys the extent to which an inferential structure makes room for considerations the audience finds pertinent to the issue at hand. A reciprocal inferential structure still exerts influence, but the target of the influence can claim some authorship or agency with respect to where and how that influence manifests itself in changes to the target?s own commitment set. In essence, the nature of the inferential structure allows the target to participate in the construction of the argument by ensuring that it is consistent with his or her prior beliefs.8 Inferential structures are reciprocal to the extent that they depend, for their force, more on these prior commitments, and less on commitments, inferences, and assumptions supplied by the speaker. To the extent that a speaker uses unfamiliar or inconsistent commitments to impose order on a target?s reasoning?in the extreme case, allowing for little or no engagement of the target?s prior commitments?an inferential structure is non-reciprocal. I address the normative trade-offs along this dimension in Chapter 5.The existing commitments on which reciprocal inferential structures can draw belong to what Bohman calls perspectives?a set of practical stances towards the world, informed by experience, that give reasons their cogency (2006, 179-80). In a similar vein, Minozzi, Neblo and Siegel (2010) develop a formal model of interactive reasoning on the assumption that 718 I am grateful to Chris Kam for this formulation. participants in discourse start with cognitive structures?webs of belief, modeled as networks of connections between simple, inferentially-linked statements?on which opinions and discursive priorities are grounded. If one thinks of rigid arguments as having force and direction, the idea of reciprocity introduces the possibility that new additions to the inferential structure can alter the outcomes of this force. Reciprocity in an inferential structure does not guarantee that such a shift will necessarily take place. If a speaker has constructed an argument entirely on the basis of commitments he or she knows the listener would agree to, we should not expect any major changes in direction; however, to the extent that an argument draws in these commitments, it is distinct from a non-reciprocal situation, in which the conclusion depends only on the commitments the speaker advances. The idea of reciprocity also draws on Aristotle?s concept of enthymeme, a form of incomplete or degenerate syllogism in which the speaker leaves the audience to fill in missing assumptions or premises (Tindale 2011, 389; Dyck 2002; Rodden 2008).9 As Klemp points out, Rawls uses the principle of reciprocity to determine the acceptability of public speech (Klemp 2012, 28). The key to public reason, for Rawls, is to address people in terms of reasons they can accept, which is similar to the idea that inferential structures can borrow force from the commitments undertaken. However, Rawls operates on a different level of analysis than the model of communicative influence I am developing here. He uses the principle of reciprocity to ground an account of political legitimacy. According to Rawls, ?Our exercise of political power is proper only when we 729 Whether the substance of claims compromises the autonomy of certain subjects by compromising their interests is a separate question. The substance of such claims must be judged, just as Habermas suggests, in terms of rightness or moral validity?the discourse over which might be subject to communicative influence, insofar as the ideal rational dialogue can only be approximated. Here, reciprocity concerns the means by which speech exerts force, and specifically the extent to which that force must rely on commitments the target has already undertaken.sincerely believe that the reasons we would offer for our political actions?were we to state them as government officials?are sufficient, and we also reasonably think that other citizens might reasonably accept those reasons? (1997, 771). While Rawls appeals to the same intuition?that people should be able to accept or reject reasons on their own terms?at the micro-level of communicative influence, reciprocity is about the linguistic processes by which someone ?reasonably accepts? an argument put to them.When an argument is non-reciprocal, the inferential structure excludes the target?s commitments, ensuring that reasoning proceeds on the original basis of the argument. A target is free to accept or reject a non-reciprocal inferential structure in its entirety, whether by rejecting a key premise or denying its internal consistency. Although the force of such a structure depends on the uptake of commitments, the consequences that follow if they do so depend on properties of the inferential structure, rather than on targets? prior commitments. For instance, a false dichotomy, a strong analogy, or a dominant frame in argument can forestall or devalue alternative considerations, and so impose a conclusion by virtue of a combination of (a) the internal logic of the argument and (b) its structural exclusion of alternative considerations. For example, a mathematical proof, which would also be rigid, is judged only on the basis of the premises stipulated in the problem?a student cannot just make up a new value and change the outcome. In a more political example, an argument equating abortion with murder is not very reciprocal; the very definition of murder works to exclude any form of reasoning besides the universal moral claim that it is necessarily wrong. Just as rigidity in inferential structures limits the range of possible conclusions that a target can draw from an inferential structure, non-reciprocity insulates inferential structures from external modification. Non-reciprocal inferential structures, whether elaborate 73arguments or relatively straightforward claims, deny targets access points for their own knowledge and facts. Non-reciprocal inferential structures can be rejected as false, but they must be accepted or rejected as stated by the speaker. Consider, again, the example of the climate change deliberation. A relatively non-reciprocal inferential structure might look like the following set of statements. If greenhouse gas emissions cause climate change, and we want to slow or halt climate change, we must reduce greenhouse gas emissions. Of course, outside of math?or a mathematical representation of logical arguments?it is hard to come up with a perfectly non-reciprocal set of claims, because to even be intelligible one needs to accept them. To accept them we need to understand the reasons for them, and they quickly root into our own commitment structures. However, it is possible for an inferential structure to be more or less reciprocal, as I illustrate below. With a reciprocal inferential structure, from the premise that greenhouse gas emissions cause climate change and that something must be done to prevent this, advancing a more specific course of action might depend on the target?s endorsement of a broader principle. For example, an actor committed to market-based measures might suggest a cap and trade system, whereas another might envision a greater role for government regulation. The pathway from cause (greenhouse gas emissions) to action, while promoting the uptake of some commitment, also leaves scope for the hearer?s own commitments to redirect the conclusion. If this deviates from what the speaker had envisioned, he or she can respond, but the original argument, insofar as it needed to borrow force from the endorsement of a premise or inference to secure the conclusion, opened itself to revision on the basis of engagement with the influence target?s own commitments. A reciprocal inferential structure also need not be rigid. Rather, it might be more exploratory, offering 74commitments to elicit a response in order to scope out the reasons others hold for their commitments. Actors can establish the sources of entitlement to each others? claims without necessarily having to undertake those claims themselves.A reciprocal inferential structure create space for the target?s commitments, external to the inferential structure, to shape and direct its force. In some cases, reciprocity may simply reinforce the existing inferential structure by multiplying points of agreement. While actors give up some measure of control in order to tap into this force, they can be strategic by anticipating the commitments of others and building inferential structures accordingly, if their aim is to produce a specific commitment. However, speakers are limited by the need to achieve resonance with targets? existing commitments. For instance, Brewer and Gross (2005) show that in the school vouchers debate, the same value, equality, can send people in different directions if they disagree over its meaning in a given situation. Both rigidity and non-reciprocity can be used to exert control over an argument?s intended effect. But whereas rigidity protects arguments from challenge?creating conditions whereby targets have no choice but to accept the speakers inferences and conclusions?non-reciprocity works by insulating them from modification or redirection as a result of target-suppled premises or facts.10TransparencyThe third dimension, transparency, concerns the extent to which the different elements of the inferential structure?the commitments that make it up and the inferential relationships among them?are made explicit. To what extent are implicit premises working to secure uptake, and would they still be able to do so if they were made explicit? The force 7510 I thank Chris Kam for this insightful formulation. of argument will sometimes, even often, depend on omitted assumptions. Where this is the case, articulating those assumptions can show them to be either false in terms of the targets? prior commitments (implicating reciprocity) or logically incoherent and inconclusive (implicating rigidity). Revealing potentially faulty aspects of an inferential structure can thereby change or neutralize the force of communicative influence. Non-transparent arguments bury premises or obscure logical connections. When arguments are transparent, hearers can accept or reject pieces of the inferential structure by evaluating proprieties of inference internal to the argument (in non-reciprocal inferential structures), or in terms of their own commitments (in reciprocal inferential structures). Non-transparent inferential structures put forward sets of claims whose logical foundations and entailments evade scrutiny, perhaps by using rhetorical tropes, by diverting attention, or by relying on assumptions targets typically take for granted. Transparency also raises a question about whether or not these dimensions vary independently of one another, particularly reciprocity and transparency. As an inferential structure becomes more transparent, each inference can become a point of potential challenge or endorsement by the target. However, as the case of a math proof demonstrates, the possibility of each transparent commitment being potential grounds for acceptance or rejection does not thereby render the inferential structure reciprocal. Transparent commitments and inference do not necessarily constitute opportunities for participation or authorship, as reciprocity would imply. Rather, they ensure that targets can be aware of the commitments and inferences that are shaping their reasoning processes. Analytically, the two dimensions address different issues in communicative influence. In a mathematical proof, for example, all the assumptions are explicit, and yet it must be judged on criteria internal to the 76structure itself?notably its coherence or consistency. By contrast, a reciprocal, transparent inferential structure enables contributions that can modify its shape by reinforcing loose inferences, reordering the commitments, or identifying points of divergence. For example, one might argue from the premise that greenhouse gas emissions cause climate change to a conclusion defending a cap and trade system, as I suggested above. A non-transparent version of this argument can leave any or all of the in-between steps in reasoning implicit. And yet, the very tenuousness of the connections invites response?or relies on the hearer?s ability to make the missing assumptions. Merely asserting the need for a given policy measure likewise lacks transparency. As Brandom remarks, any assertion, just to be intelligible, must come with an implicit set of sources of entitlement and the entailments that follow. With a bare assertion, the structure remains implicit. Figure 3.1 illustrates the three dimensional space constituted by rigidity, reciprocity, and transparency that allows us to characterize moves in ?the game of giving and asking for reasons? (Sellars qtd. in Brandom 1998, 167).Figure 3.1. The Three Dimensional Structure of Communicative Influence77Types of Communicative InfluenceThe model of communicative influence enables one to describe influence in terms of its inferential structure and, by extension, its effect on ongoing reasoning processes. The three dimensions yield eight ideal types of communicative influence.Table 3.1. Types of Communicative InfluenceCommunicative influence Type Rigidity Reciprocity TransparencyNon-Rigid Types of Communicative InfluenceDiscussion - + +Assertion - + -Explanation - - +Disclosure - - -Rigid Types of Communicative InfluencePersuasion + + +Rhetoric + + -Proof + - +Manipulation + - -Although the table represents each mode of communicative influence as a binary?rigid or not, reciprocal or not, transparent or not?in fact, they are matters of degree. Moreover, in practice, one is likely to find mixes of these in a given discursive interaction.11  In what follows, I explain how each of these modes combine rigidity, reciprocity, and transparency to exert a particular type of communicative influence. The concluding discussion will demonstrate how the model helps with some of the pressing questions around the strategic 7811 Can the scale of an inferential structure can vary, from sentences to paragraphs, or entire texts or speeches? The answer to this problem, for the time being, is that the inferential structure stands as the key unit for communicative influence, and it can vary in scale?from the overarching thread of a campaign, to a speech, to a sentence. A great deal of text could serve to articulate a very simple structure of commitments. What matters is how the speaker constructs the relationships among commitments. Inferential structures may be nested within one another. For instance, a speaker may use one non-reciprocal claim?equating abortion with murder, for instance?as a piece in a broader argument justifying a political position. On the other hand, the overarching theme of a political campaign, manifested in or supported by individual instances of communicative influence of any type, may have a given structure that resembles one of these types.nature of speech, including how one ought to distinguish between manipulation and persuasion, or between more or less acceptable forms of rhetoric. Non-rigid types of communicative influenceNon-rigid types of communicative influence more closely resemble Habermas?s and Brandom?s pragmatics. In non-rigid influence, inferential structures merely represent offers, rather than representing force. The offers may be rendered more or less appealing in terms of their validity, as Habermas would suggest, but the agency involved in the uptake and subsequent modification of the targets? commitments is their own. Indeed, communicative influence that is non-rigid and non-transparent, as I explain in Table 3.2 and the following discussion, is characterized by the lack of explicit structure built around a commitment. Non-rigid communicative influence involves inferential structures that do not exert directive force or control over the reasoning of others. It does not leverage agreement with one commitment into uptake of others. The speaker?s influence?his or her ability to secure desired consequences?is less than with rigid types of communicative influence, although the speaker also knows that whatever uptake he or she does get is secured by the target?s broader commitment structure. Non-rigid communicative influence types include discussion, assertion, explanation, and disclosure. Table 3.2. Types of Non-rigid Communicative InfluenceNon-Reciprocal ReciprocalTransparentNon-TransparentExplanation DiscussionDisclosure Assertion79DiscussionDiscussion is non-rigid, reciprocal and transparent. Rather than seeking to change one another?s commitments, actors reveal the reasoning behind one another?s claims. In many ways, discussion most closely resembles the ideal of deliberation conveyed in Habermas?s account of discourse, and subsequent theories of deliberative democracy. For Habermas, although communication?in which actors are not explicitly oriented to validity claims?serves an action coordination function, discourse does not (Habermas 1979, 19). In discourse, participants demonstrate the validity of their claims or work to achieve a common definition of their situation, including identifying relevant norms and facts. Neither implies the forcefulness one intuitively associates with a persuasive argument. Rather, in discussion, actors explain to one another ?where they are coming from.?  Discussion is characterized by the capacity to permeate boundaries, the requirement to give justification, and the knowledge that assertions can and will be challenged, but it does not necessarily serve directive functions in interaction (Landwehr 2010b). Discussion is reciprocal in the sense that Speaker 1 reveals his commitments and Speaker 2 does the same and, in doing so, they can each explore points of convergence and divergence. Discussion does not necessarily require that one attribute cooperative intentions to actors?each may just want to know where the other is coming from so that one can influence the other more effectively. The key is that the inferential structure in this case, which might follow from encountering a disagreement, is a partially shared or overlapping structure of entitlements. To that end, discussion is transparent, in that its purpose is to clarify aspects of commitment structures already present in discourse, including probing implicit premises (Goodin 2006, 255). Such an inferential structure can be visualized as emanating 80from two points, perhaps of agreement or disagreement. It lacks force because it lacks direction, but the reciprocity derives from producing a claim that elicits a response that connects it to the target?s commitments. As I show with disclosure, there is not always a potential for such connections. This shared inferential structure helps social actors navigate one another?s commitments. AssertionIf discussion most closely resembles Habermas?s search for mutual understanding, assertion belongs to Brandom?s model. Assertions are non-rigid, reciprocal, and non-transparent. They are non-rigid in that they involve single commitments rather than structures. They do not impose their effects, but rely on hearers to take them up and use them as reasons?with little immediate control over what the target treats them as reasons for. This is the sense in which assertions are reciprocal. In most cases, the influence of assertions depends on uptake, on the target?s independent evaluation of the speaker?s entitlement, and on whatever inferences the target thinks follow. Assertions are non-transparent in the sense that these connections?both in terms of the grounds for the assertion, and its potential entailments?are not spelled out. Assertions are the basic building block of all the forms of influence, but what distinguishes their non-rigid, non-transparent reciprocal mode is the fact that they are  typically not articulated as part of inferential arrangements that pre-determine the consequences of accepting them.Although they generally lack a defined structure, sometimes assertions can take on the properties of more rigid forms. To the extent that the way an assertion is articulated predetermines the consequences of uptake, even a  relatively simple assertion can take on rigid qualities. For example, the assertion that abortion is murder, if taken up, immediately 81prompts the inference that abortion is wrong or should be forbidden in a similar manner to murder. And to the extent that this predetermination is a function of the exclusion or prohibition of alternative way of thinking about the claim, it is less reciprocal. I return to this issue in Chapter 6 in a discussion of the pragmatic consequences of persuasive definitions and loaded words. ExplanationExplanation is closest to Brandom?s notion of ?making it explicit??demonstrating entitlement to an assertion. It is non-rigid, non-reciprocal and transparent. Explanation demonstrates the speaker?s sources of entitlement to a claim?what the speaker takes as reasons for it. To the extent that these patterns of commitments ought to compel universal agreement, explanation may take on a more rigid character. However, demonstrating entitlement need not produce the compelling force of a rigid inferential structure. For instance, a speaker may explain how he or she came to have a particular commitment on the basis of another speaker?s authority, a personal experience, or a broader set of consistent commitments that would have no relevance to the target. A speaker might explain why he or she does not believe in evolution by referring, in a systematic, logical way, to religious premises. But the target would have no need to accept the speaker?s conclusion if his or her own commitments were to scientific evidence rather than religious belief. Explanations establish a speaker?s justification and do not necessarily invite response. Explanations are non-reciprocal, because there is little room for engagement and redirection. An explanation articulates a set of entitlements particular to the speaker, although in principle there is no reason it could not be a shared entitlement. As with non-reciprocal inferential structures more 82generally, accepting or rejecting a piece of an explanation applies to the entire structure of entitlement as articulated.DisclosureDisclosure is non-rigid, non-reciprocal and non-transparent. Like assertions, disclosures consist of unitary claims, rather than whole inferential structures. Unlike assertions, they are non-reciprocal. Non-reciprocity in disclosure means that the (implicit) sources of entitlement are particular to the speaker?they represent sources of entitlement only for the speaker. These implicit justifications might be sufficient for a target to attribute to to the speaker a commitment and an entitlement, but not sufficient for the target to undertake that commitment himself or herself. The exemplary case of disclosure is articulating a subjective state. People are free to keep their reasons largely to themselves and, as with subjective states, disclosures are immune from challenge on the basis of someone else?s commitments. Truth and reconciliation commissions, for example, may prompt disclosures, in the sense that the nature of the communication does not invite listeners to introduce their own commitments or challenge sources of entitlement on their own terms. As a mode of communicative influence, disclosure has little force except what it contributes to context; for instance, building empathy or credibility. In disclosure, speakers may reveal preferences or subjective impressions.The difference between assertion and disclosure is like that between a question and a rhetorical question. The latter discourages contradiction by assuming agreement. It tends to be a conversation ender, not a conversation starter. But explicitly valuing disclosure as a mode of communicative influence also creates space for important aspects of identity politics. Self-representations and other ways that people seek to establish a social identity are 83valuable in themselves, and should not necessarily be judged on whatever justifications could be given for them. This is particularly the case for marginalized identity groups, who may feel that any attempt at reciprocal engagement will necessarily take place on the terms of the dominant group, and thus lead to distortion and potential further harm. There is the risk that to the extent disclosures verge towards rigidity?towards directing someone to adopt a conclusion or take action on the basis of a non-transparent, non-reciprocal form of communicative influence?they begin to resemble manipulation. For example, a speaker might make a claim about some experience of victimization in a way designed to elicit some effect?perhaps avoiding a responsibility or receiving some benefit. Had the connection been made transparent but non-rigid, the speaker would have articulated an explanation for their entitlement to the outcome. Had it been rigid and transparent, it might have sought to prove that entitlement. However, to the extent that the disclosure itself generates rigid influence towards some outcome or conclusion, without being transparent about that influence, manipulation should be a concern.A hypothetical discussion of humanitarian intervention helps illustrate these different types of communicative influence. Beginning with assertion, a speaker may simply claim that the UN should authorize an intervention in Country A. Whether this is an assertion or a disclosure would depend on what the potential, but implicit, sources of entitlement to the claim are, and whether they are somehow attached to the speaker?s subjectivity. A disclosure might be differentiated by signals within the articulation that mark it as subjective; by its content, for instance, with a declaration of preferences; or by a contextual cue. The distinguishing feature is that the entitlement cannot be inherited. In this example, it might look like ?I cannot tolerate non-intervention.? The transparent types of influence would 84reveal the structures behind such claims, perhaps revealing or contesting a commitment to a norm or moral principle, or exploring whether or not the conditions for its application are met. A discussion along these lines would yield a shared inferential structure that would enable speakers to understand each other?s stance, what commitments they share, and where they disagree.Rigid types of communicative influenceI concluded Chapter 2 by suggesting an expansion on Brandom?s model of deontic scorekeeping to account for how speech could have force. The inferential structures articulated in speech, I suggested, have the potential to generate their own momentum, rigging antecedents and consequences of reason-giving, and leveraging agreement into force that extends into the commitment structure of the influence target. The paradigm case of a rigid inferential structure is argument. One actor seeks to convince another to accept a conclusion on the basis of a series of premises or commitments and logical relations among them. Rigidity can range from an iron-clad syllogism with undeniable premises to a probabilist enthymeme that requires hearers to endorse implicit premises based on common knowledge (Gross and Dascal 2001, 277). They share, however, the common feature of directing changes to commitments on the basis of the logical arrangement of claims, rather than merely putting something forward and leaving the hearer free to alter or maintain commitments accordingly.Table 3.3. Types of Rigid Communicative InfluenceNon-Reciprocal ReciprocalTransparentNon-TransparentProof PersuasionManipulation Rhetoric85PersuasionPersuasion is distinguished by the fact that the speaker must use reasons that a hearer finds acceptable on his or her own terms?which is to say, it requires reciprocity. This understanding of persuasion is consistent with Wrong?s definition of persuasion as a form of power, ?when A presents arguments, appeals, or exhortations to B, and B, after independently evaluating the content in light of his own values and goals, accepts A?s communication as the basis for his own behaviour? (Wrong 1979, 32). More than any of the other types, persuasion tends towards leveraging explicit agreement into further commitments. Persuasion makes use of the active judgments of the targets of influence. Reciprocity means that people?s own commitments lend an argument force, and this, in addition to the logical structure implied by rigidity, is what makes persuasion compelling. For example, if the US committed to intervening in Syria if there were evidence of chemical weapons being used (and someone produced such evidence), an inferential structure promoting intervention could borrow compulsive force from the target?s explicit commitment. Persuasion, on this model, also requires transparency. Targets being persuaded can only actively engage their own commitments if they are aware of opportunities to do so. To the extent that premises are hidden but continue to exert force, they circumvent endorsement or rejection. The more of the argument that is hidden in this way, the less an inferential structure is a case of persuasion. This is not to say that transparency and reciprocity are identical. As I demonstrate, inferential structures can be reciprocal without being transparent. Rhetoricians can harness people?s values to give their arguments force without listeners necessarily being aware that they are doing so. Likewise, speakers can construct arguments that are perfectly transparent without being reciprocal, because they create no space for 86evaluation or modification on the basis of commitments external to the logic of the inferential structure. Persuasion, however, requires both transparency and reciprocity, and the two work together to ensure that people are influenced on the basis of commitments they would reflexively endorse. Inferential structures that are transparent and reciprocal still leave speakers with plenty of scope to make strategic use of communicative influence to secure some end. For example, speakers can activate a favourable latent consideration that listeners had not previously taken into account (Dickson, Hafer and Landa 2008; Hafer and Landa 2007). Thus, persuasion as I am describing it here represents an important point of departure from the initial Habermasian framework explained in Chapter 2.The measure of control strategic actors can maintain over the reasoning process depends on the arrangement of the inferential structure contained in their argument. For instance, how many junctures are there at which the introduction of an external commitment might derail the speaker?s intended outcome? To the extent that a single commitment is sufficient to entrap the target in a chain of reasoning, influence is less reciprocal and tends towards proof. Likewise, if the inferential structure can somehow predetermine which external commitments enter the picture?for instance, by invoking an issue frame?persuasion also becomes less reciprocal.RhetoricRhetoric is rigid, reciprocal, and non-transparent. Its structure pushes towards certain inferences and conclusions, and it draws force from resonance with the commitments targets already hold. But it does not necessarily engage or invoke this agreement explicitly. In Saving Persuasion, Garsten attempts to re-evaluate rhetoric for modern political discourse by introducing a neutral definition of rhetoric as ?speech designed to persuade? (2006, 5). While 87avoiding what he sees as a typical move of equating rhetoric with the manipulative or the superficial, Garsten instead undermines a distinction that I argue is worth maintaining between persuasion and rhetoric. Like persuasion, rhetoric is a rigid form?it drives towards conclusions by way of inferences. Also, like persuasion, it does so by engaging pre-existing commitments?values, beliefs, and dispositions (Danblon 2009; Dryzek 2010; Perelman and Olbrechts-Tyteca 1969). However, unlike persuasion, it does not necessarily do this in a transparent manner. Rhetoric may even obfuscate rather than clarify the (perhaps flawed) logical relationships speech seeks to establish,12 or it may hide necessary premises?often the most vulnerable ones (Rodden 2008, 163). Rhetorical influence may also work through style, arrangements, or proofs of character (Dascal and Gross 1999), not all of which can or must be fully transparent. This understanding of rhetoric as a communicative influence type is consistent with Goodin?s notion of rhetoric as ?implicit assertion,? in which premises necessary for the argument to hold together may be kept in the background, and would be rejected were they presented in explicit form (1980, 96). Rhetoric becomes less reciprocal when inferential structures are arranged in such a way that premises are supplied and evaluated from within the scope of the argument, rather than from the hearer?s existing commitments. Consider the famous ?47 percent? video of republican presidential candidate Mitt Romney speaking to donors, leaked during the 2012  campaign. The rhetorical structure of the message equated ?not paying income tax? with an attitude of victimhood, and with people being unwilling to shake dependency on the government and take responsibility for their own lives. Much as this statement may have resonated with audience?s values of self-help and independence, the structure of the 8812 The understanding of rhetoric as obfuscating problematic logical relationships was suggested by Chris Kam in discussions about a separate paper at UBC in Spring 2011. statement worked to exclude a competing commitment that members of the audience also likely shared: namely, that many of the people included in the ?47 percent??particularly seniors and vulnerable groups who receive credits and exemptions that dropped them below the federal income tax threshold?actually should pay lower taxes (Lowrie 2012). Instead, Romney?s speech implied that all ?47 percent? were freeloaders who would, in the absence of this leg up, be forced and expected to make it on their own. By masking the true referent domain of the ?47 percent,? the talking point worked to exclude many of the relevant considerations that would arise, and subsequently did arise, when that number was articulated in terms of the demographics it actually describes.Rhetoric may engage commitments implicitly by directing targets? attention to some considerations over others (Johnston et al. in Iyengar and Simon 2000, 158). Such practices, while not necessarily transparent, are also not necessarily that harmful. Rhetoric is similar to persuasion in that it derives part of its force from the target buying-in to the claims being made. The difference is that in rhetoric, this agreement remains implicit, and thus less subject to critical reflection. Examining the extent to which communicative influence is relatively reciprocal, rigid, or transparent pinpoints more precisely the normative potentials and pitfalls of rhetoric. When rhetorical influence, which is non-transparent, becomes non-reciprocal as well?that is, when it derives its force from imposed commitments rather than from the target?s own commitments?it turns into manipulation. ?True? rhetoric, which is non-transparent, is less cause for concern?although, as I show in Chapter 9, there are problems that accompany over-reliance on this type of communicative influence as well. 89ProofsProofs are rigid, non-reciprocal and transparent. Their structure exerts logically compelling force so that they only have to borrow minimal force from buy-in to secure influence. Proofs operate as coherent units. Because rejecting one piece means rejecting the proof as a proof, targets cannot redirect force by introducing new commitments. Proofs most resemble pure syllogisms, or even mathematical representations of arguments. In such cases, while the premises and conclusions are transparent, as are the rules of inference, judgments on the basis of considerations external to the argument are neither relevant nor invited. Proofs exert influence insofar as the hearer buys into the inferential structure?perhaps by endorsing a single premise from which everything follows, or perhaps by accepting the whole inferential structure as an independently compelling unit. The exchange does not require audiences to accept intermediate premises or temporary conclusions in order to build upon their commitments. Rather, the structures are beyond challenge, whether because a challenge would make little sense, as with a mathematical problem; because the structures exhaust all possible situations; or because the argument is structured in such a way that alternative considerations are not relevant. Hobbes?s Leviathan is an example of the proof mode of influence of this third type. His premises are transparent, but he does not put them forward in a way that invites commentary or evaluation in terms other than those already contained in the scope of the theory; rather, Hobbes sought to develop a ?geometry? of politics, structured tightly and syllogistically so as to propel his readers towards a necessary conclusion (Wolin 1960, 247). Readers are free to reject the premises, and with them, the entire argument, but to do so would be beside the point of a proof. 90ManipulationManipulation is rigid, non-reciprocal, and non-transparent, and for these reasons is the most normatively problematic form of communicative influence. Elements of manipulative inferential structures cannot be exposed to judgment because they are not made available (non-transparency). And because they are non-reciprocal, they are removed from potential external grounds of judgment because inferential structures supply premises, conclusions, and their own validity conditions. For example, in relation to a 2012 bill giving authorities increased power to track internet activity, Canadian Minister of Public Safety Vic Toews claimed that critics of the bill could either support it, or stand with child pornographers (National Post 2012). This claim approaches the status of manipulative influence. If these were truly the only two options, the situation would generate a very forceful inferential structure. It would leave the target with a negligible choice if, as the speaker reasonably assumes, being against child pornography is a nearly universal commitment. Ultimately, the Canadian public rejected Toews?s argument because the assumption that these two alternatives exhausted the possible positions on the bill was obviously untenable. But it still demonstrates the potential force of non-transparent, non-reciprocal claims within an inferential structure. This theory of manipulation diverges from others found in the literature. In his account of manipulation, Wrong defines manipulation in terms concealed intent, in that the desired response on the part of the recipient has not been communicated (1979, 27). It is closer to, Riker?s emphasis on structure in his account of manipulation as heresthetics, ?the art of setting up situations?composing alternatives among which policy actors must choose?in such a way that even those who do not wish to do so are compelled by the structure of 91the situation to support the heresthetician?s purpose? (Riker 1996, ix). Riker uses the term heresthetic to refer to changing the features of an issue space, but the concept also sheds light on what individual arguments can do to channel, direct, and exclude certain forms of judgment. Manipulative inferential structures can combine non-transparency and non-reciprocity to generate influence without the target being aware that his or her capacity to engage his or her own commitments is being sidestepped. Goodin?s (1980) criteria for manipulatory politics resembles these dimensions of non-transparency and non-reciprocity. He examines the range of ways manipulators can lead rational agents to act against their putative will by taking advantage of imperfect information, laying linguistic traps, deceptively playing on emotion, and ?rigging the obvious,? among other techniques. Because the combination of non-transparency (concealment or deceit) and the structural exclusion of external judgments makes manipulation so difficult to resist, it represents the ugliest way of exercising power (Goodin 1980, 19; Wrong 1979, 30). This account of manipulation is similar to Klemp?s, in that it reflects the idea that manipulative speech bypasses rational capacities (2012, 54) and exerts hidden force (2012, 47). However, Klemp?s distinction between manipulation and persuasion rests on the intention to bypass rational capacities, rather than asking how speech manages to bypass these capacities. It may be that this intention is a necessary condition for engaging in manipulation?it would be hard to accidentally construct an argument that closes out other considerations and masks connections that might be subject to challenge?but it is not essential to manipulation, nor is it necessary for understanding what manipulation entails.To illustrate the differences among rigid influence types, I turn once again to a hypothetical argument about humanitarian intervention. A persuasive inferential structure 92might look something like this: ?If you accept the validity of the norm that the international community of states has a responsibility to protect civilians, and you accept that the conditions for the norm to justify that intervention are met, then you must accept intervention as the right course of action.? While the argument is made up of committive patterns of inference, the force of its inferential structure depends on the target undertaking, or having undertaken in the past, the commitments that serve as the argument?s premises. Both premises present potential points of departure, challenge, or revision of the conclusion. A rhetorical inferential structure, by contrast, might look like this: ?states must intervene in Country A because there is a genocide? or  ?states must intervene or face another Rwanda.? Here, the inferential structure is less transparent, as the premise ?if it is a genocide, states must intervene? is now implicit rather than explicit. Nevertheless, the argument remains reciprocal because to have force, the implicit principle must resonate with the targets? own commitments. How is this different from Toews?s argument, which I used as an example of manipulation? There, the inferential structure depended on the implicit claim that people would not stand with child pornographers if there were an alternative. In that case, the hidden premise was nearly a guarantee in that rejecting the proposal would entail also abdicating many other commitments. In other words, the claim was both non-transparent and non-reciprocal. Moreover, in the Toews quote, the assumption that the two positions?supporting the government or supporting child pornographers?are exhaustive generates more of the force towards supporting the bill than the simple commitment to reject child pornography. The premise ?states must intervene in a genocide? or ?states must intervene to avoid another Rwanda? can be the object of active agreement or disagreement, while maintaining 93the integrity of the targets? broader commitment set. The latter formulation is a less transparent, and thus more rhetorical, version of the same logic?that a certain condition generates a universal, moral obligation to respond. It buries the premise that establishes the conditions for intervention under another layer of interpretation. The Rwanda analogy also embeds the assumption that actors need to avoid another Rwanda at all costs into the inferential structure as self-evident. It is thus less subject to external endorsement than the claim ?states must intervene because it is a genocide,? and thus perhaps slightly more manipulative. A proof, finally, might look like this: ?large-scale loss of life perpetrated by the state or which the state is unable to prevent, or large-scale ethnic cleansing establish the criteria for military intervention. Those conditions are met, so the conflict meets the criteria for military intervention.? Targets can contest the validity of either premise?that the conditions are met, or that they establish a threshold for intervention?but to reject one or the other is to reject the entire proof. It will not be a matter of engaging, redirecting, or changing the conclusion that can be drawn from a set of facts by introducing some new consideration.13 Summary and Comparison of Communicative Influence TypesTable 3.4 provides an overview of the eight communicative influence types, with diagrams to illustrate the nature and direction of the force of different types of inferential structures, and brief examples. 9413 The premises of these example arguments are grounded in the principles of the Responsibility to Protect norm articulated in the ICISS (2001) Report on the Responsibility to Protect.Table 3.4. Summary of Communicative Influence Types (Letters are commitments or claims. Subscripts - S: speaker commitment; T: target commitment; S/T: potentially shared commitment; None: property of the inferential structure)Type Inferential Structure ExamplesDiscussion The speaker thinks there should be intervention (XT) because he or she endorses the norm of responsibility to protect (BS/T). The target disagrees (YT). They determine that while he or she may also endorse the norm (BS/T), but he or she thinks the conditions have not been met (CT), whereas the speaker does (AS). Assertion The speaker asserts that the international community should intervene (X). Implicit are sources of entitlement that are potentially shared with the target (AS/T-CS/T).Explanation The speaker justifies intervention (X) on the basis of of the principles of the norm, and their applicability (AS/T-CS/T). They may be shared, but do not need to be to constitute an explanation.Disclosure The speaker asserts that he or she can or cannot support an intervention, on the basis of sources of entitlement that are not potentially shared, though they can be attributed. Persuasion The speaker argues that the target must accept intervention (X) if he or she accepts the validity of the principle of responsibility to protect (AT), that it?s conditions hold (BT), and that there are no other overriding considerations (CT).Rhetoric The speaker argues for intervention (X) by claiming that ?we must not have another Rwanda??that implicitly invokes the targets values or commitments: that it was a tragedy and that it could have been prevented, and that case X is sufficiently similar?though the latter may be a property of the structure.Proof The speaker demonstrates that the conditions (A-C) for the norm (X) to apply are met. Force depends on assuming the premises?for instance, the existence of the norm (AS/T). Manipulation The speaker claims that if the target values human life (AS/T) he or she must support intervention (X) because to do otherwise would amount to sanctioning genocide. Implicit is that it is a genocide, and to do nothing is to sanction.95A few important pairwise comparisons begin to demonstrate the usefulness of this model of communicative influence. Several influence types parallel one another across the rigid/non-rigid divide. For example, persuasion and discussion are analogous, but where the former uses inferential structures to drive towards new commitments, the latter is more exploratory, working backwards through chains of reasoning rather than forward. Because rigidity is a matter of degree, the boundaries between persuasion and discussion are fuzzy. Discussion begins to look like persuasion to the extent that the speaker introduces arguments that exert influence on the targets? commitments. Similarly, explanation and proof are analogous. Where explanation demonstrates the speaker?s entitlement to a claim, proofs show that anyone committed to a certain set of claims would also necessarily be committed to a certain conclusion. To the extent that inferential relationships in the inferential structure are commitment-preserving (if you accept x, you must accept y), they tend towards proof. Entitlement-preserving inferences are more characteristic of explanations.Some of the types are opposites. For instance, within rigid modes, rhetoric is the opposite of proof. Rhetoric draws people?s commitments in so that they are more inclined to accept a conclusion or inference, without necessarily realizing that their existing commitments are generating this force. A pure proof would treat these commitments as irrelevant and yet, precisely because such forms of influence do not rely on shared knowledge, all premises must be made available and obvious in order to exert leverage. Both rhetoric and proof can transition into other types. As rhetorical premises become more transparent, influence begins to resemble persuasion. As the logical structures of a proof become less transparent, such that only partial buy-in is required for inferential structures to secure outcomes without targets? knowledge or agency, it moves towards manipulation. The 96fact that in real situations, speakers will mix the eight ideal types further complicates the process of disentangling how speech secures influence.Consider the example of climate science and the role of experts in the deliberative system. Well-positioned experts explain, injecting connected facts into public discourse that other actors may take up as reasons and from which they may even make their own arguments. In practice, however, information is rarely presented in neutral ways. Findings are often presented in a way that urges change. Insofar as this reasoning is transparent, it looks like proof. To the extent that such arguments try to compel action by, for example, articulating the implications of climate change in terms of values people hold or their long term interests, communicative influence is persuasive. Such interventions from experts are unlikely to yield discussions between experts and the public. The presentation of scientific facts does not invite hearers to respond with alternative considerations, except perhaps among expert peers. The positions of speakers and audiences relative to one another, as well as the venues within which speech takes place, can alter the nature of communicative influence. The interaction between inferential structures and contextual conditions that condition communicative influence is the topic of the next chapter.ConclusionUnlike communicative action, which describes a process of interpersonal coordination through the potential for mutual justification inherent in language, I offer communicative influence as a more generic concept of what people are doing when they use language to try to modify the beliefs and actions of others. Although all types of communicative influence work only when targets take up new claims and use them in their 97own reasoning processes, there are important sources of variation in how this occurs. This variation is captured by three theoretical dimensions: rigidity determines whether speech constrains reasoning; reciprocity, whether people can bring their own commitments into a reasoning process and, potentially, alter its direction; and transparency, whether speech clarifies or obfuscates the commitments that targets are being asked to make. Taken together, these three dimensions generate eight ideal types of communicative influence. How different are the accounts I offer here?particularly of persuasion and discussion?from the original model of rational discourse envisioned by Habermas and early theorists in the deliberative democratic tradition? The criteria of transparency and reciprocity seem just as rigorous (and perhaps just as difficult to replicate in reality) as the conditions of ideal speech. However, the model of communicative influence offers several advantages. First, it focuses on the features of speech itself rather than on the features of motives or institutional arrangements (although, as I will explain in later chapters, it has implications for institutional arrangements). Second, the dimensions of communicative influence capture the fact that both reciprocity and transparency are matters of degree, and speech should not be dismissed as strategic the moment one or the other is compromised. Third, the added dimension of rigidity enables us to examine the ways speech can leave room for strategy?understood as trying to secure specific outcomes in light of the influence target?s existing commitments?without compromising its normative value. By contrast, understanding rational discourse as the testing of validity claims does not take account of strategic dimension present even in friendly, cooperative exchanges?for instance, when speakers casually overlook examples that may hurt their case, or try to argue one another into a corner. In Chapter 5, I explain the connection between the dimensions of 98communicative influence identified here and an autonomy-based account of the democratic goods of speech. Fourth, the model of communicative influence allows us to situate speech on three dimensions rather than one. For instance, looking at rhetorical speech, one can ask: was it more like manipulation, in that the implicit premises were supplied rather than engaged? Or was it more like assertion, in that it did not forward a logical structure that drove the influence target to accept some further conclusion? Fifth, by reconstructing speech and argument in terms of the pragmatics of discursive exchanges, the model of communicative influence presented in this chapter creates a common ground for bridging a number of theoretical and empirical approaches to the problem of strategic speech. Because it is abstract, the model enables us to pull out the generic features of communication, which provide a translation point between findings in psychology and rational choice, on the one hand, and normative democratic theory, on the other. The conceptual framework will enable democratic theory to reflect the real constraints and incentives speakers and audiences face when using words under conditions of political conflict. This will include, in the latter part of the dissertation, making concrete suggestions for change that will harness the democratic potentials of strategic speech, while correcting its most pathological effects. In the next chapter, I theorize the contextual limitations on communicative influence, as well as the incentives and opportunities actors have to deploy different types. 99Chapter 4. The Model of Communicative Influence II: Discursive EcosystemsIn this chapter, I introduce and develop a second component of the model of communicative influence, based on the idea that the context in which inferential structures are embedded partially accounts for the nature of their influence. For example, knowledge imbalances between speakers and targets on a given issue?like those between scientific experts and lay publics?may make reciprocal types of communicative influence unlikely in that area, thereby favoring explanation over discussion, and proof over persuasion. Credibility deriving from expertise, trust, first-hand knowledge or shared interest may allow some speakers to exercise greater influence than others, regardless of how convincing or unconvincing their arguments are. In addition, the prior commitments into which inferential structures intervene may be organized according to what political psychologists call frames, and these can affect the way targets weigh the considerations prompted by inferential structures. And, of most importance for designing democratic institutions, institutional environments create incentives to engage in different types of communicative influence. For instance, institutions that enable people to challenge implicit premises may induce speakers to simply make their premises explicit in the first place. Or speakers may instead engage in reciprocal argumentation that incorporates the targets? own commitments in ways that forestall potential criticisms. In institutions that record explicit commitments and make it costly to violate them, speakers may leave more of their inferential structures implicit. The aim of this chapter is to systematize these contextual considerations in order to build them into the model of communicative influence.100Contextual constraints can render well-intentioned speech harmful to autonomous judgment, as when someone unconsciously invokes a dominant frame that crowds out competing considerations. But context also has the potential to draw out the democratic value, even in speech that lacks deliberative intent. For instance, institutions can draw benefits out of strategic behaviour by allowing actors to challenge one another, rendering implicit aspects of inferential structures explicit?benefits that could not be identified using intention based theories of deliberation. Examining the context dependency of communicative influence provides a more complete picture of the force of speech, and is consistent with a pragmatic approach. Indeed, in the study of language, pragmatics is distinguished from semantics by virtue of its focus ?on the users and context of language rather than on reference, truth, or grammar? (Fotion 1995, 709). Attention to context reveals how strategic actors leverage opportunities to expand their communicative influence. For instance, they may seek to build up their own credibility and undermine that of their opponents. They may invoke frames that support their position in order to take advantage of familiar patterns of reasoning among influence targets. And  they may use institutional rules and procedures to hedge and equivocate, while perhaps maneuvering to force opponents to take unpopular positions. The conceptualizations I develop here lay the foundations for Chapters 7 through 9, in which I theorize the ways interventions at the level of context?particularly altering institutional environments?can secure better forms of communicative influence, even harnessing actors? self-interested behaviour to do so. To help develop this piece of the model of communicative influence, I conceptualize the discursive context in which speakers and audiences are embedded as an ecosystem. An 101ecosystem is ?a biological system composed of all the organisms found in a particular physical environment, interacting with it and with each other,? although the term can also describe complex non-biological systems that comprise both internal and environmental influences (Oxford English Dictionary). The analogy between discursive context and an ecosystem will enable me to talk about the structure of prior commitments, the statuses of speakers and subjects, and the institutional arrangements that govern communication, all under the headings of ecological features and ecological effects. Much like physical environments, these discursive ecological features interact with the primary objects of concern: speakers, targets, and inferential structures. Moreover, the ecological analogy draws attention to the idea that when people speak and listen, they both respond to and alter their discursive environments, and even minor modifications to those environments can have far reaching effects. The idea of a discursive ecosystem resembles the phenomenological concept of the lifeworld (Schutz 1970), which Habermas uses to work out the notion of communicative action by positioning it as ?a reservoir of taken-for-granteds, of unshaken convictions that participants in conversation draw upon in cooperative processes of interpretation? (Habermas 1989, 170). The notion of a discursive ecosystem differs from that of the lifeworld in that it grants special status to subjects and institutions, as well as the structural relationships among pre-existing commitments. Beyond merely labeling them as shared meanings necessary for communication, ecological features can have real force, altering incentives, directing reasoning, empowering some subjects, and disempowering others.The first section of this chapter establishes and explains the ecological features that make up the discursive ecosystem. I describe these features in pragmatic terms, but the ideas 102are drawn from other fields, primarily political psychology and rational choice. First, structural/substantive features frame issues and establish their dimensionality. Second, subject-based features like expertise, first-hand knowledge, or trusted status generate credibility, on the speaker side. On the target side, knowledge, engagement, and psychological tendencies set limits on the ways inferential structures can engage targets? commitments. Third, institutional features determine whether or not explicit commitments are costly to back out of, as well as who speaks to whom, when, and with what kind of audience. In the second section, I explain three types of ecological effects associated with these features: modification effects, magnification effects, and selection effects. In order to illustrate some specific propositions about selection effects, I also include a theoretical exercise in which I trace the consequences of institutional characteristics of some typical discursive venues, given a set of assumptions about the behaviour of political adversaries. Finally, I reflect on how strategic actors might target discursive ecosystems directly, in order to generate advantages and opportunities to exert greater communicative influence. Chapters 7 through 9 take each of these ecological effects in turn, explaining the kinds of generic anti-democratic hazards they generate for autonomous judgments, and suggesting institutional interventions that might respond to these hazards.Conceptualizing the Discursive EcosystemsRecall from the last chapter that communicative influence operates on the reasoning processes of others, leveraging uptake and agreement into further commitments. Speakers advance inferential structures: sets of inferentially connected commitments with the capacity 103to drive logical and practical reasoning towards conclusions. These inferential structures rely for their force on explicit or implicit commitments and inferences, which are supplied either by the speaker or from the targets? own prior commitments. But their influence is also conditioned by the context in which speech takes place, which I have described as a discursive ecosystem. My aim in this section is to describe the three sets of ecological features introduced above?structural/substantive, subject-based, and institutional?in terms of pragmatics, setting up the second section in which I explain the implications of these features for communicative influence. Admittedly, theorizing context as I am adds additional complexity to my model. But I hope to make that complexity that arises from theorizing context tractable by ideal-typing both types of ecological features and types of ecological effects. Table 4.1 summarizes the types of ecological features explained below.Table 4.1. Ecological Features that Condition Communicative InfluenceEcological Feature Object of interest Types and Sub-typesI. Structural/SubstantiveConcern the arrangement of pre-existing commitments, both in terms of salience and common connectionsFramesII. Subject-BasedConcern subject characteristics that condition communicative influenceTarget CharacteristicsSpeaker CharacteristicsIII. InstitutionalConcern the rules, procedures and practices within which speakers, and thus communicative influence, operateGenerative InstitutionsTransactional InstitutionsSubstantive/structural features of the discursive ecosystem: FramesThe uptake of inferential structures, and thus the influence of speech, depends at least in part on their resonance with existing beliefs (Delia 1970; Benford and Snow 2000; Iyengar and Simon 2000). What people already take to be true matters for the inferential 104consequences of incorporating a new proposition. This ecosystem feature is closely connected to reciprocity, one of the three dimensions on which the inferential structure of utterances varies. Reciprocal inferential structures are open to external commitments or shut them down. When an inferential structure successfully alters a commitment, this produces consequences for other commitments that are inferentially linked to the one that was changed. In this section, I add complexity to this account by considering the role of contexts in determining what happens when an inferential structure activates a target commitment with strong links to further commitments constituted by the substantive structures that, I argue, are superimposed onto the basic inferential connections that make up an actor?s web of commitments. I conceptualize this ?superimposed? structure in terms of the concept of frames, drawn from political psychology. To the extent a frame is pervasive in public discourse around an issue, it structures existing commitments in ways that modify communicative influence. In practice, framing occurs when ?a speaker?s emphasis on a subset of potentially relevant considerations causes individuals to focus on these considerations when constructing their opinions? (Druckman 2001, 1042). It establishes the dimensions of evaluation and render beliefs available, accessible, and applicable (Chong and Druckman 2007b, 105, 11). I am assuming that framing, in addition to being something speakers can do, also exists as part of the political landscape that constitutes the discursive ecosystem around an issue. To some extent, frames are an inevitable part of people?s ability to think and talk about politics (Nelson and Kinder 1996; Lau and Schlesinger 2005); however, they also have the potential to allow strategic speakers to leverage the existing inferential pathways constituted 105by the structure to exert influence that extends far beyond the substance of what they claim. Frames may pit tax hikes against welfare programs, individualism against compassion, or jobs against the environment. In pragmatic terms, frames work by packaging commitments together, and by creating tight links among beliefs, between beliefs and values, or between values and actions. These bundles may also work through processes of exclusion. By establishing, for instance, that an issue is of a certain type (e.g. a moral issue or a national security issue), a frame renders some commitments more relevant than others. On this account, frames work not by changing beliefs, but by causing some considerations to be weighed more heavily than others in making a judgment. That means that by invoking a frame, a speaker can activate inferential processes, without necessarily having to spell the inferences out.14 Moreover, frames may mask instances in which some commitments, which an actor would not endorse, may follow implicitly from another aspect of the frame that they have endorsed (A. Calvert and Warren 2012). Frames can also limit the range of considerations an actor is likely to bring to bear when reasoning in response to a speaker?s effort to influence them. There are two important ways in which the frame-based discursive ecologies can vary. The first concerns frame strength. Although it is not apparent what makes a frame strong, it is clear that certain frames are more powerful than others, and more powerful frames can trump the effects of frame repetition (Chong and Druckman 2007b) in making beliefs more available, accessible, and applicable (Arceneaux 2012, 273). If frames have the pragmatic effect of linking and excluding inferences, these effects are likely to be more 10614 In much of the literature, framing is distinguished from persuasion in terms of ?the weight parameter.? Whereas persuasion involves actual belief change, framing merely alters the weight given to different considerations (Nelson, Oxley and Clawson 1997).pronounced with stronger frames. Second, competition among frames over a given issue also has implications for communicative influence. Empirical research has shown that conversations that cut across frames mitigate their effects, increasing the accessibility of a broader range of underlying considerations (Chong and Druckman 2007a, 652). Others qualify this finding by pointing out that although the magnitude and direction of shifts in opinion may be less when subjects are exposed to competing frames, the frames nevertheless continue to shape the range and substance of considerations subjects access when articulating their opinions (Brewer and Gross 2005, 943). Moreover, competing frames may have the pragmatic effect of what Calvert and Warren (2012) call ?epistemic discounting,?  in which audiences give less weight to the substance of a speaker?s commitments if those commitments are associated with a competing frame, thus hindering the potential for dialogue to move people beyond the limiting effects of frames. For example, communication between Republicans and Democrats may be more challenging if, as Barker (2005) suggests, Republicans respond to individualist frames and Democrats to egalitarian ones.According to Riker, strategic political actors can seek to modify the dimensions along which actors situate their preferences when evaluating an issue (1990, 46-47). Adding dimensions alters the shape of the issue space, breaking down majorities and moving voters around in relation to set choices, without actually changing their preferences (Riker 1986). While Riker?s account of dimensionality is different from the political psychologists? account of frames, both work by determining which considerations or preferences inform judgment on a particular issue. The powerful pragmatic effects of this practice, which Riker dubs heresthetics, can also be explained at the level of practical reasoning. When people engage in practical reasoning, they move from an account of the state of the world to intentional states 107through patterns of practical reasoning (types of oughts, including preferences). By defining the situation as consisting of a particular set of dimensions, strategic actors can modify outcomes without actually speaking directly to preferences, institutional roles, or moral obligations. Taken together, frames and heresthetics offer both a psychological and rationalist take on the idea that the pre-given structures of the target commitment sets into which inferential structures intervene matter for communicative influence. Subject-based features of the discursive ecosystem: Speaker and target characteristics The discursive ecosystem is populated by speaking and listening subjects whose characteristics and tendencies favour or undermine the uptake of claims and yield more or less careful evaluation of inferential structures. Table 4.2 lists such subject-based ecological features according to whether they belong primarily to the speaker or to the target. Table 4.2.Subject-Based Features of Discursive EcosystemsTarget-Based Features Speaker-based Features- Knowledge (including engagement and personal experience)- Stake (including interests and level of commitment to the beliefs or actions)Credibility, deriving from?- Expertise,- Records and issue ownership- Trust (interest and identity-based)Target-based ecological featuresTarget characteristics and tendencies can alter the way the inferential structures speakers articulate affect their reasoning. Are targets knowledgeable on an issue? Are they politically sophisticated or engaged enough to situate that issue in a broader context? Are they motivated to engage their cognitive capabilities? A variety of empirical findings suggest that audience traits matter in how people receive and process arguments. A first set of traits concerns knowledge and engagement, which tells us something about whether an influence 108target is likely to have prior commitments on an issue, and how well grounded those are in his or her broader web of commitments. Several noteworthy empirical tendencies suggest the importance of target characteristics. Those who are less engaged in politics are more likely to rely on metaphors to construct their opinions (Barry et al. 2009) or rely on cue-based processing, in general (Bullock 2011). Moreover, when people are given little information, plausible explanations will trump evidence in argument (Brem and Rips 2000, 584; Rips, Brem and Bailenson 1999, 117). Finally, the reference group with which a target identifies on a given issue may determine the substantive basis on which he or she responds to arguments (Chong 2000, 74) or, in pragmatic terms, which commitments they access.Second, the effects of communicative influence on reasoning may also depend on a target?s stake in a given problem. Knowing that one is likely to be affected promotes conscious, thoughtful, or rational processing over reliance on cues or heuristics (Petty and Cacioppo 1986, 126; Marcus, Neuman and Mackuen 2000, 129). Stake can also be conceptualized in terms of a target?s web of commitments. If an inferential structure puts pressure on a core commitment?for instance, a belief that is integral to someone?s comprehensive worldview?the pragmatic consequences of giving it up will be more profound than for a relatively peripheral belief like which movie is best. Because of the magnitude of these pragmatic consequences, the force required to change a core commitment or a commitment in which the target has a strong stake, may be higher. Thus, for influence to be successful in such cases, inferential structures would need to leverage greater reciprocity, greater rigidity, or both. To further complicate matters, a targets? perceived stake also depends on his or her perception of the risks involved. Prospect theory suggests that people have a bias towards avoiding loss over making equivalent gains (Mercer 2005). If the target is in a 109domain of potential loss, then arguments promising to avoid loss are more compelling than those promising gains (Arceneaux 2012, 281).In pragmatic terms, we can think of target characteristics as altering the conditions under which targets undertake commitments. Some characteristics raise the bar for the uptake of commitments, requiring that inferential structures be more transparent, reciprocal, or rigid in order for a speaker to generate communicative influence through uptake. For example, if targets have a higher stake or greater knowledge, speakers may need to use more transparent inferential structures to secure influence. By contrast, targets with lower stakes, lower anxiety, or less knowledge, will be more susceptible to non-transparent inferential structures. Similarly, low knowledge situations may make it harder for speakers to construct reciprocal inferential structures because there are fewer pre-existing beliefs to engage. Finally, if targets determine, consciously or unconsciously, that the relevant basis for judgment on a given issue is tied to some aspect of their identity, speakers who fail to engage that identity may not be able to draw target commitments into an inferential structure, weakening their capacity to generate communicative influence Speaker-based ecological features Intuitively, some speakers wield greater influence than others. Rightly or wrongly, they may be perceived as more credible, knowledgeable, or genuine. These kinds of statuses, however, usually exist only in relation to a given issue. The person to whom one implicitly attributes entitlement to make military decisions is not necessarily the same person one would trust to make decisions about environmental policy. Thus, the enhanced communicative influence generated by speakers? status within the discursive context must be defined in relation to issues. 110At this point, I wish to introduce a pragmatic distinction between credibility and authority. Credibility is a feature of the discursive ecosystem adhering to subject statuses. Authority, on the other hand, can be a source of appeals, and can be built into inferential structures. Both credibility and authority establish entitlement; however, with authority, the source of authority functions as an implicit or explicit premise?accept X because I, the speaker, have some source of authority pertaining to X. By contrast, with credibility, its source is not actually part of the argument. In pragmatic terms, credibility concerns the deontic attitude of attributing entitlement. Put simply, on some issues, some speakers are more credible than others, and this status enhances their influence, even when inferential structures do not make appeals on the basis of this status. This theoretical claim resonates with empirical findings. Persuasion appears to be most effective when it comes from a credible source (Iyengar and Simon 2000). The effects of credibility also raise normative concerns, to which I return in Chapter 8. Targets may also sometimes attribute unwarranted credibility in the form of misplaced trust, misunderstood expertise, power or popularity (Landwehr 2010a), or by basing their judgments on misleading accounts of a speaker?s record. Actors sometimes have good reason to take people at their word?in pragmatic terms, to attribute entitlement to a claim. The most obvious basis of credibility?and perhaps the most difficult to distinguish in practice from an implicit claim to authority?is when targets value the speaker?s expertise or firsthand knowledge. Credibility might also derive from consistency of a speaker?s record on a given subject. The attention politicians pay to building up, defending, and promoting voting records suggests the importance of this source of credibility (Ansolabehere and Iyengar 1996, 111-12). In addition, the need to maintain this 111credibility by protecting reputations constrains the kinds of commitments speakers can articulate. Targets might also attribute entitlement to claims on the shakier ground of party affiliations, and targets? cognitive associations between speakers, parties, and issues (Iyengar and Valentino 2000, 128), which may also preclude credibility or entitlement on issues seen as ?belonging? to another party (Feldman and Conover 1983). As I shall explain in detail below, credibility enhances communicative influence. When a target attributes to a speaker entitlement to the claims that make up an inferential structure, the likelihood that the target will undertake the new commitments, and the conclusions that follow, increases.Institutional featuresThe most concrete of the ecological features are institutional. Institutional features determine whether speakers can reach a wider or narrower audience; they establish rules concerning questioning and cross examination; and they reward or penalize different types of communicative influence?favouring, for example, explanation over persuasion, or rhetoric over discussion. Institutions may also create incentives that affect when and how people choose to speak and what they reveal. For instance, if a speaker knows he or she will be challenged for deceitful speech, he or she may be less likely to be deceitful in the first place. Institutions constitute much of the terrain on which political communication plays out. In this section, I present a description of these institutional features. Defined in relation to pragmatics, institutions can be grouped according to two types: transactional and generative. Transactional institutional features determine how speakers are positioned in relation to targets, including whether or not the targets also become speakers and whether or not parties to an exchange aim influence towards a non-participating audience. Generative institutional features embed commitments in the discursive context, making it easier for 112actors to keep track of one another?s commitments and build them into inferential structures. They generate solid reference points around which subsequent discursive interactions may revolve.Transactional institutional featuresTransactional institutional features determine the direction of communication, the number of parties, and the likelihood of exchanges.15  They also concern whether or not speakers have the opportunity to verify one another?s claims and impose penalties for lying (Lupia and McCubbins 2000, 50). These features establish the structure of the relationship between speaker(s) and influence target(s) and, in dialogic situations, enable these positions to be reversed?or permit both speakers to try to influence third parties, as in televised leaders? debates. A courtroom?s transactional features, to take another example, are adversarial,16 but they also establish strict rules of conduct, evidence, and burdens of proof. In representative assemblies, transactional features establish rules for who gets to speak when, and whether others have an opportunity to respond.In pragmatic terms, transactional features of institutions alter the extent to which inferential structures can be made explicit, either voluntarily because of speaking rules, or forcefully because speakers have been challenged. Moreover, as I explore in the next section, knowing that one may be challenged may prompt certain behaviors?for instance, a speaker 11315 The term ?transactional? is drawn from Pinto (2010, p. 232) but expanded considerably. I use it to capture whether communication is one-directional?from speaker to audience?or two-directional, allowing for exchanges between parties.16 Ideas about the importance of adversarial relationships and the status of court-room style exchanges were developed in conversation with Bernard Manin at a conference in June 2011. Manin also discusses the importance of adversaries in Democratic Deliberation: Why We Should Promote Debate Rather than Discussion (Manin 2005).may anticipate the interlocutors? commitments in order to co-opt claims that might otherwise enable challenges.Generative institutional featuresGenerative institutional features arise when the commitments speakers articulate become embedded in institutional contexts, and thus alter the discursive ecosystem. Actors can then leverage these explicit commitments in future interactions. For example, when representatives vote on bills and amendments they enter positions in the public record, generating explicit commitments to which they can later be held. Similarly, governing parties in minority parliaments can extract positive votes from opposition parties with the threat that the opposition will be ?responsible? for triggering an election if they do not make those commitments. Further examples of generative institutions include provisions to agree to stipulative definitions, or reliance on definitions already in place in legislation (Walton 2001). Rules that enable agenda-setters to manipulate the order of votes can generate commitments by forcing actors to take position against one alternative before proceeding to the next pair in order to eliminate otherwise popular alternatives (Riker 1986). Institutions might also be generative in slightly less concrete ways. For instance, a gesture on behalf of a state?a formal apology for past injustices, for instance?can provide a permanent, if symbolic, point of reference for ongoing discursive interaction. As Mihaela Mihai (2013) argues, even strategically articulated official state apologies generate symbolic recognitions that have profound consequences for relationships between victims and the demos. She argues that official apologies can create positive shifts in political culture by virtue of the capacity of official apologies to set an example that is taken up in the form of judgments about the past (Mihai 2013, 218).114By exposing and punishing inconsistency to varying degrees, generative institutions can raise or lower the costs of going back on commitments, and, to the extent that reneging is costly, speakers can build into inferential structures targets? past commitments as dependable points of leverage. Institutions that raise the costs of violating expressed commitments thus also make these commitments more meaningful, both as subsequent pressure points and also because actors may think twice about making them in the first place. Hobbes takes this notion to an extreme in claiming that ?covenants, without the sword, are but words, and of no strength to secure man at all? (Chapter XVII,  93). For Hobbes, without explicit costs to reneging, actors cannot ever depend upon the word of others.Even without going to Hobbes?s extreme, any argument for the force of language must assume that commitments have some staying power internal to the communicative process. But particularly where past commitments conflict with current incentives, the extent to which speakers can count on one another?s commitments may vary with the structure of the institutions in which they are embedded. Finally, merely by participating in institutions actors can incur certain general commitments?for instance, a commitment to the legitimacy of the institution itself, a commitment to following the rule of civility, or a commitment to abiding by the outcome of collective decision-making, even if it does not go one?s way. Indeed, most institutions work by generating norms to which people who occupy a position, such as an elected office, are held. Institutions impose normative commitments?to represent, to be accountable, or to hold the public trust?simply by recognizing someone as holding a position. Generative institutions give speech a sticky quality: commitments are recorded and formalized, providing solid reference points within their web of commitments.115Features of discursive ecosystems: SummaryMy aim in this section has been to lay the groundwork for a discussion of how these ecological features affect communicative influence. Re-constructing research findings around reasoning and behaviour in terms of pragmatics makes it possible to relate these findings to the idea of communicative influence?and, in later chapters, to their democratic implications. By incorporating the notion of a discursive ecosystem, the model of communicative influence provides a pivot point between these empirical findings and a normative account of the democratic functions of speech. Table 4.3 presents a summary of the descriptive ecological features and their pragmatic effects.116Table 4.3. Summary of the Features of Discursive EcosystemsFeature Type Pragmatic Effects Sources of VariationStructural-Substantive:Framing- The presence of frames, particularly strong frames, in the discursive ecosystem imposes a structure on top of the baseline inferential links among commitments, establishing stronger and weaker connections or rendering some more salient than others. Establishing issue dimensionality can change the grounds for judgment and decision. - Frame strength, Frame reliance- Issue dimensionalitySubject-Based:Target- Audience characteristics may raise or lower the threshold of how much reciprocity/rigidity/transparency is necessary for influence to be successful. Lack of knowledge may set a ceiling on reciprocity. - Knowledge- Stake in an issue (both interests and level of attachment to the commitment speakers seek to dislodge)Subject-Based:Speaker- Credibility promotes attribution of entitlement, increasing the likelihood of uptake of the commitments that make up an inferential structure.- Expertise, first-hand knowledge, individual or party records on a given issue, trust (identity or interest based)Institutional:Transactional- Structure discourse, establishing who gets to speak to whom, when, the audience size, and whether target/speaker positions get reversed.- One-way, two-way, or two-way directed at a non-participating third party- Adversarial, cooperative, cross-examination, questions, turn-taking, time allotmentsInstitutional:Generative- Embed articulated commitments into the discursive ecosystem, making them available as reference or leverage points in future interaction. They make speech ?sticky,?- Do institutions record commitments and make it costly for people to break them?- Does participation involve making a commitment to shared rules or definitions?- Can they establish non-inferential links between commitments (i.e. omnibus bills, amendments)?Effects of Discursive Ecosystems on Communicative InfluenceIn addition to identifying these features of discursive ecosystem, it is possible to think systematically about their effects. Ultimately, understanding these effects will help to guide 117institutional remedies against forms of influence that undermine autonomous judgment. There are three ways in which ecological features can affect communicative influence. The first is modification. Here, the interplay between inferential structures and the discursive ecosystem moves communicative influence along its three dimensions, increasing or decreasing rigidity, reciprocity, and transparency. The second type of effect is magnification, in which ecological features promote the uptake of the commitments that make up an inferential structure, thus magnifying its force. The third is a selection effect, in which ecological features?especially institutions?incentivize certain types of communicative influence over others. Generally speaking, I find that structural/substantive and target-based features of the discursive ecosystem exert primarily modification effects; speaker-based features exert primarily magnification effects; and institutional features exert primarily selection effects.Table 4.4. Types of Ecological EffectsType of Effect Ecological Features Ecological effect...Modification - Structural/Substantive: Frames- Subject-based: Target characteristicsInteraction between inferential structures and the discursive ecosystem modifies communicative influence by rendering it more or less rigid, reciprocal, or transparentMagnification - Subject-based: Speaker characteristicsInteraction between inferential structures and the discursive ecosystem magnifies influence by promoting uptake of commitments within inferential structures. Selection - Institutional: Generative and transactionalInstitutions incentivize more or less rigidity, reciprocity and transparency, selecting for different types of communicative influence.118Modification effectsModification effects occur when inferential structures combine with ecologies to render communicative influence more or less rigid, more or less reciprocal, or more or less transparent. Inferential structures do not operate in a vacuum, but interact with pre-existing commitments, from which they can also draw force. These commitments are often subject to frames, which organize beliefs, values, and preferences around an issue. Frames work to strengthen some inferential connections and weaken others, and to alter the salience of commitments. When an inferential structure touches on a frame, whether deliberately or not, it can trigger frame-based reasoning processes that modify communicative influence.First, frames can modify by compromising reciprocity. They do this by establishing that an issue is of a certain type, which blocks considerations people might have brought to bear in the absence of the frame. For example, a dominant frame in Canadian discourse equates discussions of health care reform with privatization and Americanization. The dominance of this frame precludes meaningful consideration of alternative pathways to reform and deters useful comparisons with other, better performing systems (A. Calvert and Warren 2012). Frames are part of context, but to the extent that inferential structures implicate frames, the frames themselves can work to favour some commitments over others that might have steered the reasoning process in a different direction. Second, frames can modify the transparency of communicative influence, enabling inferential structures to leave some of their components implicit. For example, in Canada the Harper government framed the F-35 fighter jet purchase as a military necessity, suppressing considerations of the actual cost by casting any questions about cost as military weakness or as neglect of the needs of military personnel. In pragmatic terms, because frames ?bundle? 119commitments and pre-determine the inferential relations among them, familiar frames can prompt people to skip steps and endorse implicit assumptions. Because frames trigger a script instead of constructing an explicit argument (Mutz, Sniderman and Brody 1996, 5), hearers will not necessarily recognize the grounds on which they are agreeing to something. Third, some frames may also modify communicative influence by rendering it more rigid. Frames can embed assumptions that render certain inferences more likely or even necessary. For instance, framing a social program as a tax hike can leverage people?s preference not to pay more taxes. Depending on the strength of the frame, this may allow an inferential structure to simply bypass considerations of the merits of the policy. Understanding framing effects in pragmatic terms suggests possibilities for correcting for them. For example, institutions that place actors in adversarial positions?as in a courtroom exchange?may artificially induce transparency and reciprocity by prompting them to challenge one another?s inferential structures, thus revealing implicit assumptions and alternatives. Empirical evidence suggests that such competition mitigates the effects of frames (Chong and Druckman 2010). I examine these possibilities in greater depth in Chapter 7. Like frames, target-based ecological features can also work to modify communicative influence. For instance, an inferential structure open to reciprocal engagement cannot by itself account for reciprocity in communicative influence. Targets must actually have the commitments to engage, and sufficient understanding to do so. Similarly, if an actor is not motivated to engage in active processing because of low stakes or low anxiety (Marcus 2002), influence is also unlikely to be reciprocal. By contrast, knowledgeable, engaged, or motivated targets are more likely to engage in reciprocal, active processing. They may 120perhaps even impose transparency and reciprocity on inferential structures by challenging commitments the inferential structure treats as given. In terms of potential correctives, examined in Chapter 7, these potentials for positive modifications reinforce the importance of learning phases in deliberative processes and additionally suggest that achieving autonomy-preserving exchanges might require helping people identify their own stake in an issue. Magnification effectsMagnification effects also result from subject-based ecological features, but from speaker rather than target characteristics. I associate magnification effects primarily with statuses, traits, and relationships that make speakers credible. Credible speakers, by virtue of their relationships with an issue or with an audience, are more likely to be granted entitlement to their commitments, increasing the chances that audiences will undertake those commitments themselves, without actually altering the inferential structures they articulate. By magnifying influence, expertise or trusted status may make targets more likely to be influenced even by assertions, although the speaker still cannot control for what it serves as a reason. Conversely, lack of credibility can also undermine influence, potentially even the influence of otherwise compelling inferential structures. In recognition of this possibility, strategic speakers may seek to prevent their words from being dismissed our of hand as either self-serving or otherwise not credible in a given situation. One such familiar tactic is to preface an argument with ?I?m not a Democrat/Republican/environmentalist/conservative/local resident, but??.The notion that a speaker?s communicative influence derives in part from his or her credibility may create an environment where speakers face unequal starting points for influence on certain issues. Some have to overcome a ?credibility gap? that forces to offer 121more elaborate inferential structures to secure the influence someone else could achieve with an assertion. While there may be good reasons for credibility to be unevenly distributed, deliberative systems need to be attentive to this form of inequality for two reasons. First, in order for this aspect of the discursive ecosystem to be consistent with autonomy, the sources of credibility on a given issue should be able to withstand reflexive, critical scrutiny?and , moreover, should be subject to periodic reconsideration to ensure that they do. Second, speakers should have opportunities to use inferential structures to exert influence in ways that can overcome this credibility gap. I return to the problems magnification creates for autonomous judgment, and ways of responding to those problems, in Chapter 8. Selection effectsSelection effects occur when institutional ecological features incentivize some types of communicative influence over others. Institutions do this particularly well by structuring the consequences of speech. For example, transactional contexts that favor exchanges, rather than one-directional communication, may promote greater transparency. Not only can adversaries (or colleagues) make explicit the implicit components of a speaker?s original inferential structure, but the knowledge that their implicit claims may be rendered explicit on someone else?s (unfriendly) terms may make speakers more forthright in the first place?or at the very least more careful in what they articulate (Lupia and McCubbins 2000, 50). When the target of influence is part of the exchange, knowing that he or she will have a chance to introduce his or her own arguments may lead the speaker to attempt to engage those commitments in the first place, perhaps in the form of appeals to common interests (Dryzek and List 2003). In some cases, speakers may engage in reciprocal argument in highly strategic ways; for instance, in an effort to foreclose avenues of challenge, a speaker might 122co-opt an opponent?s values into an inferential structure that favours the speaker?s conclusion. In other instances, institutional structures will select, not by indirectly incentivizing behaviours, but by directly forcing speech to conform to certain rules. For example, the forced transparency of scientific publishing?citing sources, providing data for replication, and subjecting work to challenge in organized forums?produces communicative influence of the ?proof? and ?explanation? types. At the same time, insofar as competing theories are at work, the context may also select for persuasion and discussion. People are forced to work out their differences in reference to a shared set of commitments?a body of previous findings and work?that provide common ground and allow actors to ground their arguments reciprocally in commitments that others presumably share. In generative institutions, where talk may be decision-oriented and binding, as opposed to everyday talk, actors know that commitments represent potential future force, and may be more evasive and less responsive to the reasons of others as a result.17Selection effects of generative and transactional institutional featuresIn order to expand on the idea of selection, I briefly illustrate how institutional arrangements might incentivize different types of communicative influence, using a thought experiment that assumes a set of incentives, and asks how these play out across different institutional conditions. For the purposes of the argument, I specify a set of assumptions that stylize political contestation. The starting assumption is that speakers are motivated to influence others?to convince them to undertake a new commitment or abandon an old one. I 12317 For a discussion of the differences between everyday and decision-oriented talk in the deliberative system, see (Mansbridge 1999).also assume that strategic actors want to minimize the number of commitments they articulate when institutions are generative, because each commitment limits potential future claims by virtue of relationships of incompatibility that threaten incoherence. Additionally, each represents a potential point of pressure that can be used against them by an adversary. This assumption is a simplification, since in any given situation, there would be many harmless commitments speakers would willingly acknowledge for every one they might strategically wish to avoid. However, I maintain the assumption in order to simplify this exercise. I further assume that strategic actors want others to have more commitments on the record so that they can increase the leverage they gain through concrete attributions of explicit commitments. In short, this set of assumptions stipulates that actors typically want others to rack up explicit commitments, without racking up commitments themselves. Transactional features of institutions can be one-way, in which only one actor gets to speak; two-way and direct, in which two speakers direct influence at each other; or two-way and indirect, in which two speakers direct influence at a non-participating audience. I assume that institutions can be generative for neither speaker, for one, or for both. For the sake of argument, I treat a venue or political practice as either generative or not, when it may actually be a function of how relatively costly the institution makes it to renege on explicit commitments. Table 4.5 lists some typical political practices that reflect the eight possible combinations of generative and transactional conditions.124Table 4.5. Institutional Ecological Characteristics of Common Discursive PracticesOne-Way Two-Way, Direct Two-Way, IndirectGenerative for neither actorGenerative for one actorGenerative for both actorsAnonymous comments Anonymous comment threads (i.e. internet message boards)Anonymous comment threads (i.e. internet message boards) with a perceived audiencePublic statement, position paperCross-Examination,Town hall Q&A (I assume institutions are generative only for the answerer)Press Conference,Parliamentary Question Period(I assume institutions are generative only for the answerer)?- Deliberation where actors are trying to influence one another, rather than an audienceDebate (i.e. televised leaders? debates during campaigns; public representative assembly debates)These categorizations highlight a number of possible selection effects. First, institutions that are generative for whomever articulates an inferential structure lead to less transparency because each commitment a speaker makes explicit represents a potential point of leverage against them. Second, generative institutions lead to greater rigidity, because actors want to produce commitments in others, and can derive force from inferential structures in order to do so. Finally, two-way transactional contexts, particularly those that are generative for both actors, may favour some form of reciprocity, because actors may want to take advantage of the commitments of others, rather than articulating their own, to construct an inferential structure. In the first row in Table 4.5, we might expect to see any of the influence types. A non-generative venue like an online forum does not select against transparency. Because actors cannot be held to their commitments, they do not mind articulating them. Likewise, a non-generative venue would, under the specified conditions, fail to select for rigidity because 125there is no particular incentive to produce commitments in others to which they cannot later be held. Because there is no incentive to construct rigid inferential structures while avoiding articulating commitments of one?s own, such practices also do not select for or against reciprocity. If we relax the above assumptions to allow that people may be willing to abide by their commitments (and others can attribute that willingness to them) one might expect to see increases in rigidity as people actually try to convince one another. On the whole, though, in institutional venues featuring these non-generative conditions?think, again, of online, anonymous forums?all three transactional contexts are consistent with the full range of communicative influence types and, speaking in broad strokes, this is what one sees in anonymous, uncontrolled exchanges like internet comment threads.The second row of Table 4.5 comprises venues and arrangements that are generative for one of the two actors. They include public position statements, town hall Q&As, and press conferences. Their generative character may work against transparency on the part of the speaker for whom they are generative. That is, a speaker may not to spell out all of his or her commitments?for instance, in a public position paper?lest these cause trouble later on, particularly now that records of communication are so easily archived, retrieved and disseminated. However, two-way contexts like town halls and courtrooms also position interlocutors to prompt or compel speakers to spell out their commitments ?on record.? While these two-way transactional features do not necessarily select for transparency by altering incentives, this arrangement may end up producing greater transparency because the questioners force the speakers to render their commitments explicit. By contrast, the third column describes institutional circumstances in which influence is aimed at an audience that is not part of the exchange, with one of the two participants?the answerer?having the 126additional concern of avoiding generating commitments on record. In the press conference example, while questioners may attempt to impose transparency?asking questions in a rigid way designed to prevent equivocation, or invoking the speaker?s prior commitments in an effort to get ?a straight answer??speakers nevertheless have incentives to avoid transparency. Ruling out transparent types, then, circumstances that are two-way, indirect, and generative for one actor select for assertion, disclosure, manipulation and rhetoric.The final row describes venues that generate commitments for both actors. This is where we might locate deliberative venues?dialogic exchanges involving reasons, preferences, beliefs, and values. Transactional features will be two-way, and either direct or indirect. In two-way, direct, and generative circumstances?that is, when parties to an exchange direct influence at one another, and both incur commitments that stick?one might expect to see rigid types of communicative influence as participants try to get one another to take on commitments. Because I assumed that actors do not want to articulate their own commitments, generative features continue to select against transparency. As such, this row leaves three options: manipulation, rhetoric, and persuasion. Manipulation and rhetoric each apply because they are both rigid and non-transparent. Proof is excluded because it would require actors to articulate commitments of their own. Persuasion is included because it is reciprocal. As such, it makes it possible for speakers, using persuasion, to build compelling inferential structures from commitments their opponents already have on the record, rather than having to articulate their own. The possibility of constructing compelling inferential structures out of one another?s commitments opens up possibilities for moving from manipulation towards persuasion. If both parties can begin to articulate claims in terms of the commitments of their opponents, 127they can gradually expand the pool of commitments to which speakers can appeal, increasing the scope for persuasion and resisting the pull towards manipulation. This theoretical result suggests a possible insight for how two-way, direct exchanges in adversarial situations?assuming actors actually have some reason to attempt to influence one another at all?can move from manipulation, through rhetoric, to persuasion. For instance, using an intermediary to have actors sign on to a set of beliefs, an agreed upon account of events, or a set of values can enable actors to engage in reciprocal rhetoric or persuasion and over time expand the range of commitments to which speakers can appeal.This means of escaping pressures towards manipulation by gradually building up shared resources for persuasion may not be possible in more public exchanges?described in the third column?where influence is indirectly aimed at a non-participating audience. In cases like televised campaign debates and parliamentary question period, speakers are typically not trying to influence one another. To the extent that speakers do incorporate opponents? commitments into inferential structures, it is most likely for the strategic purpose of undermining the opponent?s credibility in the eyes of the audience. Rather than a source of reciprocal influence, opponents? commitments are offered as premises from which audiences are meant to reason: ?my opponent said X, therefore he or she cannot share your commitment to value Y.? Because speakers know that in articulating a commitment they may leave themselves exposed to such tactics, generative, two-way indirect contexts select against transparency. Because influence is directed at an audience, it is engagement with audience commitments that has the potential to render communicative influence reciprocal. Arrangements such as leaders? debates select for rhetoric or manipulation?rhetoric to the 128extent that they seek to resonate or engage with audience?s beliefs and values, and manipulation to the extent that they seek to include in inferential structures necessary assumptions that would likely not receive uptake. Elaborating the theoretical consequences of transactional and generative institutional features under a given set of assumptions about political actors? incentives demonstrates the way the discursive ecosystem concept helps to draw out the implications of everyday political practices for communicative influence.Summary: Effects of discursive ecosystems on communicative influenceIn the preceding section, I identified three types of ecological effect. Modification of communicative influence occurs when inferential structures encounter either entrenched frames, or subject tendencies that produce variation?and gaps?in target commitments through knowledge, engagement, and stakes. Magnification effects occur when some speakers, for reasons of expertise, trust, or perceived shared interest, wield greater communicative influence than others, independently of the inferential structures they articulate. Selection effects occur when institutional incentives favour some types of communicative influence over others. Institutional features of discursive ecosystems, however, have implications for all three types of effects. In addition to being designed to select for specific types of communicative influence, institutions can correct for modification effects of frames by prompting speakers to unpack the ways others are using frames to render certain inferences implicit, inducing transparency and potentially reciprocity. They can facilitate learning, engagement, and awareness of stakes. Institutions can confer or revoke markers credibility, and render explicit the sources of speakers? credibility so that it can become objects of critical reflection, and enable concrete demonstrations of credibility. Institutions are versatile in their ability to produce and condition ecological effects. I explore 129this versatility in Chapters 7 to 9, in which I explore institutional responses the three types of ecological effect?modification, magnification, and selection?each of which poses its own particular anti-democratic hazards.Strategic Actors and Discursive EcosystemsIt follows from this analysis that strategic actors can also take advantage of the opportunities generated by credibility, strong frames, or institutions. It is therefore unsurprising that actors often orient themselves not directly to modifying the commitments of others through rigid influence, but rather to modifying the discursive context, thereby securing influence indirectly. Strategic considerations may lead speakers to try to introduce or to perpetuate a frame that is favourable to their position; to try to enhance their credibility by touting a record, reporting personal experience, or invoking partisan cues; or to force opponents into taking unpopular positions on legislation. These strategies can certainly have a profound effect on actors? ability to exert or resist communicative influence in subsequent interactions. For example, in seeking to perpetuate or take advantage of strong frames, speakers may merely ?state facts? rather than making arguments; however, carefully chosen facts, stories, or words can entrench a frame, creating the conditions under which only certain appeals can achieve purchase with a given audience.A speaker might also use a simple assertion for heresthetic purposes, increasing the dimensionality of issues in order to move people?s choices into a more favourable range (1990, 1986). A simple assertion can raise a dimension of judgment that splits the opposition, bringing an evaluative dimension to the forefront without actually constructing an argument. Taking advantage of modification effects that pre-determine the commitments on which a 130target draws, and their level of attention to those commitments, opens up a range of strategic possibilities for even non-rigid forms of speech: assertion, explanation, disclosure, and discussion. For example, asserting X may not persuade someone that X is true, but the assertion may suggest that something about X is worth considering, thus modifying the dimensionality of some issue. Likewise, a disclosure, articulated without background and with no expectation of response, may be sufficient to enhance a speaker?s credibility on some issue, or to undermine someone else?s. Strategic speakers can also alter the ecological conditions under which communicative influence takes place by raising awareness of an issue?increasing knowledge, sophistication, or even perception of having a stake in order to prompt more deliberate processing. A rich ecological account allows us to understand how statements that do not, on their surface, appear to seek influence over a target nonetheless modify the discursive ecosystem in ways that can have a significant indirect impact on communicative influence. This may be particularly true of several of the common patterns of communicative influence addressed in the next chapter. For example, narratives typically involve non-rigid types of influence rather than those trying to secure agreement on a commitment. And yet they might have even greater long term effects by virtue of their potential to perpetuate a frame or establish credibility through first-hand experience or shared identity.ConclusionThe concept of a discursive ecosystem is critical to the model of communicative influence, and thus to understanding the democratic implications of strategic speech, because it allows us to theorize how context affects communicative influence. This chapter 131highlighted the ecological features?structural/substantive, subject-based, and institutional?that condition communicative influence. I argued that these ecological features interact with the inferential structures of speech to produce three generic types of effect?modification, magnification and selection?and that we cannot understand communicative influence or its democratic implications without grasping this context-dependency.Each of these types of effect provides the foundation for the analysis I undertake in Chapters 7 through 9 to show how institutional interventions into discursive ecosystems can generate democratically desirable consequences from strategic speech. This kind of analysis has been largely excluded from democratic theorists? understanding of communication. In order to pursue this line of inquiry, however, democratic goods need to be systematically reintroduced into the conceptual framework of the model of communicative influence. Unlike Habermas?s distinction between communicative and strategic action, the multidimensionality of communicative influence will allow the identification of some of the trade-offs among the multiple democratic functions of political communication. 132Chapter 5: Communicative Influence and Democratic GoodsHabermas developed the notion of communicative action as a means of understanding how human interaction based in language proceeded on the basis of rational agreement. This idea holds powerful normative appeal. Having established, in the model of communicative influence, an alternative way of thinking about the force of language in politics, I now return to the normative principles behind democratic theory?s interest in speech and ask: how do different types of communicative influence and the discursive ecosystems that produce them enhance or undermine the quality of democracy? The search for mutual understanding on the basis of validity claims both depends upon and reinforces the democratic norms of mutual respect, reciprocity, equality, and consent. The notion of an exchange of reasons that provides a rational, consensus basis for political decisions that includes the agreement of all affected provides one normative justification for deliberation. This is one justification; equally important is the justification grounded in the value of autonomy. Genuine reason-giving treats others as self-authenticating sources of validity claims (Rawls 1993, 72). Communicative action thus both requires and enables political autonomy (Habermas 1996, 127). Yet, as the first chapter demonstrated, by establishing these norms as requirements and setting themselves apart from strategy at their very pragmatic foundations, the proponents of theories of deliberative democracy effectively removed many political situations? characterized as these are by disrespect, ill intent, and strategic imperatives governing the use of language?from the scope of autonomy. The model of communicative influence starts from the premise that, in situations where there are incentives to do so, actors will use language in an effort to secure favorable judgments, belief changes, and action 133among targets of influence. It allows us to understand, quite precisely, the potential and risks of a variety of forms of strategic speech, including, for example, the much discussed question of the role of rhetoric in political deliberation (Chambers 2009; Dryzek 2010; O'Neill 2002; Young 2002).The aim of this chapter is to determine what becomes of the normative arguments for communicative action?the democratic goods it is said to secure, and the supposed benefits of deliberation in meeting its strict standards?when one replaces the pragmatics of communicative action with the pragmatics of communicative influence. Strategic orientations are excluded from models based on communicative action precisely because they are taken to violate autonomy (O'Neill 2002). However, I argue that the same normative principles as underwrite communicative action can be integrated into the model I develop in this dissertation in order to distinguish better and worse uses of communicative influence. The key difference is that according to the model I develop here, in many cases, strategic discourse can still support autonomy. This chapter proceeds as follows. In the first section, I identify the key normative criteria at stake in evaluating communicative influence by looking at existing normative arguments for deliberation, which I reconstruct in terms of autonomy. The second section examines how what I will call ?discursive democratic goods??understood in terms of autonomous judgment?fit into the framework of communicative influence. I argue that they do so by linking the three dimensions of communicative influence (rigidity, reciprocity, and transparency) with the key elements of autonomy. By connecting discursive democratic goods to the dimensions of communicative influence, we can have a clear sense of why, for example, persuasion is superior to manipulation, despite both having strategic elements. In 134the third section, I show how the normative concepts discussed in the first two sections link with the idea of the discursive ecosystem introduced in the previous chapter. This discussion will offer an account of when contextual conditions?understood in terms of structural/substantive, subject-based, and institutional ecologies?uphold autonomy.To give a preview of the arguments to come, this chapter lays the groundwork for subsequent chapters by bringing discursive democratic goods into the model of communicative influence alongside the concepts of inferential structures and the discursive ecosystem. In Chapter 6 I apply the model to a series of pervasive features of political communication including, among others, story-telling, analogies, loaded words, exaggeration, and hedging. In Chapters 7 to 9 I consider how institutional interventions into the discursive ecosystem can be used to secure discursive democratic goods.Discursive Democratic Goods and AutonomyThe concept of communicative action captures two features of discourse that further normative ends, which I call discursive democratic goods. They are, first, the motivating force of reasons and, second, freedom in discursive opinion and will formation, both of which are ?democratic goods,? I argue, because they instantiate the demands of autonomy at the level of discursive practice. For the purposes of this dissertation, I leave aside the democratic and universalization principles, which specify the validity of norms and the moral character of rules in terms of whether they would gain rational acceptance in idealized discourse (Habermas, 1996). While undoubtedly an important component of communicative action as part of a moral and ethical system, they pertain only to idealized discourses that serve as the testing grounds for norms and as sources of (hypothetical) justification. 135The purpose of this section is to explain Habermas?s ideas of the motivating force of reasons and freedom in discursive opinion and will formation in order to establish the conceptual connection between the practice of reason-giving and the principle of autonomy. According to Habermas, intersubjectively shared and discursively produced beliefs have their own kind of motivating force. Accepting or rejecting a speech act offer implies a tacit acceptance or rejection of obligations that follow from the speech act (Habermas 1996, 147). The very act of taking this yes or no position is a manifestation of the second key feature of discourse, which assures the democratic desirability of speech: freedom in discursive opinion and will formation. The motivating force of reasons can only derive from ?structures of undamaged intersubjectivity? (Habermas 1996, 151)?that is to say, the ?force? of reasons cannot be imposed, but by its very nature must be accepted.In a Habermasian framework, the obligations that result from the acceptance of speech act offers are consistent with democratic autonomy insofar as discourse works only through the discursive democratic goods of the motivating force of reasons and freedom in discursive opinion and will formation. This normative argument for discourse is separate from other claims that provide important normative bases for democracy?for example, the all-affected principle of democracy?which suggests that people should have a say in the decisions that affect them. The questions of how, when, and whose choices should be empowered is central to democratic theory, and are those to which the all-affected principle is a suitable answer; however, the normative question of what makes a choice autonomous, and thus worth empowering, is distinct from questions of empowerment. While there are many definitions of autonomy, current definitions of political autonomy converge on the notion that 136a person is autonomous if he or she acts intentionally on the basis of reasons (Forst 2005, 230).I argue that autonomy is the appropriate normative principle with which to judge communicative influence for two reasons. The first has to do with its internal connection to the discursive democratic goods identified above. The ideas of both the motivating force of reasons and of freedom in discursive opinion and will formation represent what Habermas sought to develop: a social, discursive account of Kant?s imperative to treat others as autonomous beings (as means, rather than ends) and one that loads this requirement onto the linguistic resources that enable social action to begin with. The second reason has to do with autonomy?s position as a conceptual counterpoint to influence. Specifically, I am able to frame both autonomy and influence at the same pragmatic level of analysis, in which influence imposes constraints on reasoning, and autonomy provides the principle with which it is possible to judge the democratic acceptability of those constraints.Before proceeding to a deeper analysis of autonomy and communicative influence, I briefly consider two alternative approaches to a normative analysis of communication. The first of these two approaches is also based on the concept of autonomy, but instead comes at autonomy from the the speaker?s intent?in particular, the speaker?s intent to bypass the listener?s rational capacities (Klemp 2012, 54) and his or her intent to engage in a cooperative search for the truth (Dowding 2011). In both cases, the moral quality of the action rests with the speaker?s intent. As I argued in the introduction, however, such an approach cannot get traction on questions of how speech actually exerts force, nor on the question of how different ways of leveraging speech into influence have different effects on autonomous judgment. Questions of the morality of the speaker?s behaviour aside, what 137should be at issue for discursive democratic goods?and thus for the deliberative quality of a democracy?is a target?s ability to grasp, and potentially challenge, the grounds a speaker articulates for adopting a commitment, both in terms of the structure of claims and the conditions under which they are articulated. In other words, it is not concealed intent that makes manipulation problematic for democratic theory, but concealment of the assumptions and inferences that are exerting force on targets? choices, judgments, and actions. Focusing on these effects means that, rather than ruling out the strategic forms of speech that accompany political incentives?even the openly strategic behaviours Klemp labels as morally neutral?the model of communicative influence can identify and correct for harms of the specific effects of strategic speech in discursive ecosystems.A second plausible alternative approach to normatively evaluating communicative influence lies in theories of domination. According to Pettit, domination describes the capacity, actual or potential, for an actor to interfere arbitrarily with the actions of another. Interference refers to actions that reduce the range of options available to another, or which alter the profile of those options?for instance, by altering their payoffs and so making some options more salient than others. Interferences are ?arbitrary? insofar as they as discretionary?initiated and controlled only by the will of the interferer (Pettit 1999, 52). The idea of domination offers some useful insights. For example, if inferential structures are conceptualized as potential forms of interference at the level of reasoning and judgment, manipulation can certainly be described as arbitrary. It is not subject to the will of the person being interfered with. If it were transparent, the target would be free to reject it. If it were reciprocal, his or her will would have been attended to in the first place. Domination as a concept is best suited to describing interpersonal relationships, and particularly the positions, 138statuses, or capacities that put one actor in a position to interfere with another. Thus, manipulation may enable domination. But, for the purposes of the model of communicative influence, domination is not the opposite of autonomy. Rather, autonomy is a property of the will actors can bring to bear on the inferential structures that interfere with their reasoning processes. Communicative Influence and AutonomyRecall from Chapter 2 that the impetus behind the move from communicative action to communicative influence is the the insight that speech is often used to secure outcomes?to alter the beliefs, judgments, and actions of others. The key reason to develop the model of communicative influence is that it allows us to consider how instances of speech that do not meet the model of communicative action?for instance, disclosure or rhetoric?can nevertheless engage people?s capacity for judgment and reflection. But at the same time, we still need to be able to distinguish between normatively desirable and undesirable types of communicative influence. The discursive democratic goods that enable theorists to make normative arguments in favour of talk-based politics are the right ones, but they need to be matched with an alternative account of communication itself. Whereas the previous section worked upwards from the specific normatively desirable features of discourse to the general principle of autonomy, this section works from autonomy to communicative influence. Here, I break the concept of autonomy into its component parts, and then situate these parts in relation to the three dimensions of communicative influence. I then use this framework to characterize types of communicative influence?rigidity, reciprocity and transparency?in terms of autonomy. 139A person is autonomous if he or she acts intentionally on the basis of reasons (Forst 2005, 230). The intentionality component of Forst?s definition reflects one core elements of autonomy?reflexivity, meaning that people are aware of their reasons for action. This definition should be qualified to include the requirement that an autonomous person have the ?capacity to judge, decide, and act on the basis of her own attitudes and reasoning? (Mele qtd in Bl?ser, Sch?pf and Wilaschek 2010, 240). This refinement specifies that autonomy involves her own attitudes and reasoning, which reflects an important component of autonomy: what Christman and Anderson call authenticity (Christman and Anderson 2005). An autonomous person is in a position to give voice to his or her reasons (Benson 2005), and these reasons must be his or her own. While there are persistent academic disagreements about what autonomy entails, it is possible to extract several themes that constitute its core elements. The first element of autonomy is choice.18 Where we are concerned with the impact of specific instances of speech on reasoning, it makes more sense to talk about autonomous judgments. Judgment is the focus here over choice, because at the level of reasoning, judgment precedes the choice of whether to undertake a commitment. The vocabulary of autonomous judgment captures something important about the moment at which an inferential structure begins to generate pragmatic consequences for a target?s commitments. The second element of autonomy is reflexivity. Reflexivity refers to how consciously or intentionally arguments or assertions are processed. If autonomy means acting for reasons, an actor needs to be aware of those reasons and able to identify them. Theorists of autonomy suggest a more fine-grained understanding of reflexivity by distinguishing between the 14018 A person may be described as autonomous, but autonomous personhood is the domain of theories of domination, because it depends on the social relationships in which that person is embedded.capacity for and the exercise of reflection. I take up this distinction with respect to the transparency of inferential structures in terms of whether speech enables or disables capacity for reflexivity, understood as thinking through the propositions and entailments that make up an inferential structure. Autonomy assumes the capacity for critical reflection, but the question for communicative influence is whether or not speech has the effect of engaging critical reflection. The third element of autonomy is authenticity. Authenticity concerns the extent to which a target can recognize himself or herself in the reasons for action. The key here is not just that people are moved by reasons, but that these reasons are, in some way, meaningful for them. That reasons are meaningful for the target is part of what accounts for their robustness to reflexive scrutiny. Like transparency and reciprocity (as discussed in Chapter 3), authenticity and reflexivity are interrelated. Non-authentic reasons?those that would not resonate with an actor?s other beliefs or values?would not withstand reflexive scrutiny. On the other hand, without some substantial commitments with which to compare a reason, scrutiny would be meaningless. Authenticity captures the extent to which uptake of an inferential structure depends on reasons one can own, and identify with, or at least that one would not feel alienated from upon taking a more reflexive stance (Christman 2005, 333).The relationships between these elements and communicative influence should begin to become apparent from this analysis. Choice pertains to rigidity. By virtue of the logical entailments that necessitate certain propositions and foreclose others, rigid structures constrain choice. If speech is to achieve anything, at a certain point it must drive towards decision, which demands some amount of limiting, ruling out, and directing judgment. The deliberative democratic commitment that rests on communicative action is that these 141decisions should be made on the basis of reasons. Interpreted through the model of communicative influence, however, the deliberative democratic commitment should be that decisions should follow from chains of commitments linked by reasons that those being influenced could endorse. Admittedly, in their directive function, the chains of commitments constituted by inferential structures do reduce choice. The extent to which this is acceptable depends on the other two dimensions of speech. Before proceeding to this question, it is worth noting that in situations of non-rigid speech, choice is constrained only by ecological factors, and not by the logical structure of inferences. Hence, one must evaluate non-rigid speech in terms of its effect on context, and how it affects subsequent instances of communicative influence. I return to this issue in the following section.The first criteria that determines whether the constraints that accompany rigidity in communicative influence are democratically acceptable is reflexivity. In terms of the model of communicative influence, whether communicative influence will engage or repel a target?s capacity for reflection depends on the transparency of its inferential structures. When inferential structures are transparent, people reflect on their sources of entitlement and their implications, whether on the grounds of their own pre-existing commitments, or the grounds supplied by the speaker. Reciprocity describes whether an inferential structure prompts or enables targets to bring their own commitments to bear on the matter. Reciprocal inferential structures promote authenticity, the third core element of autonomy. To be autonomous, it is not enough to be moved by reasons. Those reasons must be grounded in considerations the hearer can accept on the basis of his or her own web of commitments. Authenticity that renders the discursive democratic good of the motivating force of reasons is normatively 142desirable. To be motivated by reasons does nothing for autonomy if those reasons are not one?s own. Consider again the example, discussed in Chapter 3, of a hypothetical discussion of humanitarian intervention. If a target agrees that the criteria he or she has endorsed?for instance, the requirements for the principle of ?Responsibility to Protect? to justify intervention?are met in a given case, and that these are the only considerations relevant to the conclusion, the target is inferentially compelled to agree with an intervention because of her commitment to the principle. The speaker may have provided information that show the criteria are met. But agreeing to those facts compels further inferences only as a result of the target?s commitment to the principle of Responsibility to Protect. To the extent that the argument is transparent and thus subject to reflexive endorsement (as opposed to implied by analogy, for instance), the ?force? of communicative influence is consistent with autonomous judgment. Even argument by analogy can draw out active and reflexive engagement, so it does not necessarily preclude autonomous judgment. On the other hand, if agreement were compelled by reasons a target could neither recognize nor endorse, speech would compromise his or her autonomy.The discursive democratic goods of the motivating force of reasons and freedom in discursive opinion and will formation require both reflexivity and authenticity?which is to say, speech must be both transparent and reciprocal. To take a yes/no position on a validity claim?one that generates further obligations?one needs to know what that validity claim is, and this requires some measure of transparency. At the same time, for position-taking to be free, one must be taking that yes/no position on grounds one would have chosen for oneself. Does this kind of argument leave us where we started? That is, does it imply that the only 143normatively acceptable kind of speech for deliberation is that intended to persuade? Not necessarily, since linking the model of communicative influence with an account of autonomy yields a much more complex account of the normative potentials and pitfalls of different types and patterns of speech, which I will now examine in detail.Using autonomy to evaluate types of communicative influenceFor judgments that result from inferential structures to be fully autonomous, they must be both transparent and reciprocal. That is to say, they must engage people?s capacity for reflection, and draw on authentic beliefs.19   To the extent that choice is constrained, it should be constrained only by the will and judgment of the influence target. While manipulation is a form of domination, and persuasion is autonomy-preserving, this insight also yields a number of normative propositions about the other six types of communicative influence: proof, rhetoric, discussion, explanation, assertion, and disclosure.In its purest form, proof is normatively neutral. Without minimal buy-in (accepting the starting premises) it can have no force. The transparency involved in a proof may be valuable for other reasons, but without this buy-in, it can have no immediate effect on autonomy. To the extent that actors do buy into the premises, they may be forced to accept what follows; but transparency ensures that this buy-in is reflexive and that it enables 14419 An interesting question arises here with respect to the grounds for persuasion. Should speakers still have to work within targets? commitments to generate inferential structures?reasons and arguments?if the target?s beliefs are themselves immoral or clash with the speaker?s worldview? What if he or she wants to avoid even giving these commitments any kind of airing, as he or she might wish in the instance of racist or homophobic views. There are two options for influencing the target that are consistent with his or her autonomy. One is to draw on those commitments that are not so objectionable, in order to secure incremental agreement. However, this option will not always be available. The second option is to create an inferential structure on the basis of the speaker?s own commitments, knowing that, without buy-in, it will have limited influence?e.g. a proof. The influence of a proof depends on if, and where, the target buys in. Leaving speaker commitments that are necessary inferences but which, if the target were aware of them, would cause him or her to reject the inferential structure, is inconsistent with the target?s autonomy (this reflection emerged from discussion with Mike Neblo).reciprocal, or authentic, endorsement. When transparency declines, influence tends towards manipulation. When reciprocity increases?thereby increasing the opportunities for authentic buy-in to the force of an inferential structure?influence tends towards persuasion. Like proof, rhetoric only partially captures the requirements of autonomy. In its purest form, rhetoric is entirely reciprocal. That is, its force derives from the beliefs and values that already make up the recipient?s web of commitments, even though it is not transparent and thus does not lend itself well to reflection. To the extent that a rhetorician introduces or imposes external commitments or logics, rhetoric becomes manipulative. Thus, the dimensions of non-transparency and rigidity enable us to separate the normative harms of a rhetorical demagogue from more acceptable uses of rhetoric. A demagogue may rely on premises (and emotional content) that derives from the target?s own web of commitments. When a speaker builds into an inferential structure implicit, forceful connections and inferences that do not derive from commitments the target would otherwise accept, however, rhetoric begins to move in the direction of manipulation. If, however distasteful, a speaker truly restricts himself or herself to the substantive content and connections among pre-existing commitments, that influence would have to be taken as consistent with the targets? autonomy?though it might violate other, non-communication based democratic norms. Conversely, to the extent that logics and implicit connections are rendered transparent, rhetoric begins to resemble persuasion, and thus favours autonomy. Non-rigid types?discussion, explanation, assertion, and disclosure?do not lend themselves to quite the same approach to normative evaluation because the inferential structures of speech they embody do not themselves constrain choice and judgment. Of course, they can still have effects, in that actors are free to take up their claims or not. But 145these cases are normatively benign or even positive, because choice is unconstrained by the structures of speech themselves. However, the uptake of claims, as well as the kinds of claims available to speakers in the first place, will be constrained by the characteristics of the discursive ecosystem. These ecological conditions and constraints can be evaluated in terms of whether or not they create positive conditions for autonomous judgment, as I explain in detail in the next section, and in detail in Chapters 7 through 9. For instance, discussion may favour autonomy by helping actors to identify their own commitments, thereby enabling authenticity. Explanation may help actors build, understand, and make connections among commitments in the first place. By contrast, assertion and disclosure may have the advantages of lowering the threshold to disrupt the influence of problematic inferential structures. For instance, if a dominant frame in the discursive ecosystem generates rigid influence that compromises the autonomous judgment, it should be possible for others to challenge that frame?and perhaps improve the overall conditions for autonomous judgment?without having to construct an inferential structure to do so. In this respect, assertion and disclosure may themselves have value for autonomy, in that they make it easier to articulate challenges to instances of communicative influence that do threaten autonomous judgment. But in order to determine whether or not non-rigid influence aids or undermines autonomy, we need to first understand how these ecological features themselves relate to autonomy, in terms of the three elements under consideration here: judgment, reflexivity, and authenticity.Evaluating Discursive Ecosystems from a Democratic PerspectiveChapter 3 identified three sets of ecological features that make a difference for communicative influence: structural/substantive, subject-based, and institutional. These 146features exert modification, magnification, and selection pressures on speakers as well as on the ways recipients process the information and arguments that speakers put to them. In this section, I revisit these three types of features in turn in order to show how the autonomy-based discursive and democratic goods identified above can ground a normative evaluation of the conditions of discourse.20 Structural/substantive ecological featuresA discursive ecosystem consists in part of the pre-given relationships among beliefs and values that structure actors? existing commitments. This structure includes the salience of some beliefs or dimensions of evaluation over others as well as the connections people tend to make among ideas and values. In Chapter 4, I discussed these ecological features under the concept of frames. The ecological effects of frames can have profound implications for autonomy. An account of frames grounded in pragmatics, as the model of communicative influence is, explains frames in terms of ?bundling? some commitments, to the exclusion of others. An actor who signs on to one commitment has effectively signed on to further commitments in ways determined by the substance of the frame, rather than the logic of individual inferences (A. Calvert and Warren 2012). Frames compromise reciprocity by limiting the accessibility of considerations not prompted by the frame, and they compromise transparency by pushing inferential links and assumptions into the background, where they are less subject to reflective testing. The modifying effects of frames?their tendency to undermine reciprocity and transparency?mean that deliberative democrats should be 14720 Klemp also turns his attention to what he calls the discursive environment, arguing that what distinguishes contexts, from a normative perspective, is whether or not they intensify manipulative speech?s capacity to interfere, while lessening the possibilities for contestation (2012, 67). While this is part of the story, I argue that a communicative influence-based account of democratic goods highlights a larger range of contextual conditions than are implicated in the normative judgment of speech.concerned about the presence of a strong frame in the discursive ecosystem around an issue, and about the non-rigid forms of speech that perpetuate such frames. Frames that function to establish the relative weight of considerations around an issue compromise authenticity, because targets do not access the full range of considerations they might in the hypothetical absence of a frame. For example, the Harper government framed their decision to proceed with a no-bid procurement of F-35 fighter jets frames the question of procurement in terms of the armed services needing the latest and best equipment, period. This frame operated to suppress arguments about costs, whether the government should consider cost as a factor in weighing alternatives, as well as important considerations about Canada?s future defense needs. When an inferential target triggers a frame, it can evoke responses that lack reflexivity and authenticity. It can damage reflexivity because it does not make explicit claims relevant to the issue, such that they cannot be subjected to conscious processing and critical reflection. It can damage authenticity, because, by keeping claims implicit, it prevents people from mobilizing their own reasons to evaluate an argument. To the extent that even non-rigid communicative influence types entrench a frame?for instance, a speaker explaining why this jet is the best plane, or a speaker disclosing his or her preference for ?whatever is best for the troops??their modification effects raise problems for autonomy for which institutions should correct. I return to this issue in Chapter 7. Framing is an inevitable part of how people think about issues, and non-rigid types of influence like assertion are part of that process. The key is to identify when the frames around which communication gets organized are particularly damaging?as when they enable significant rigidity while compromising reciprocity and transparency?and correct for those cases.148Speech that triggers frame-based reasoning does not preclude authenticity, and thus autonomy. Demanding that communicative influence avoid frames in order to leave open access to a full range of considerations around an issue is an impossibly high bar. On the other hand, frames that work by admitting only one type of consideration?for instance, constraining the F-35 debate to revolve only around the issue of equipment, and casting objections as unpatriotic?should and do raise serious concerns about the autonomy of the judgments that follow. In between these two extremes, we can offer normative evaluations of frames that become embedded and replicated within discursive ecosystems in terms of the implications for authenticity and reflexivity that follow from their conscious or unconscious use. That is, to the extent that a frame renders communicative influence less reciprocal by excluding considerations, less transparent by rendering inferences implicit, or both, it becomes problematic.21Subject-based ecological featuresRecall from the previous chapter that subject-based ecological features concern traits of speakers and targets. On the target side, the processes by which ecological features connect to autonomy are similar to those of structural/substantive features, in that they modify communicative influence. In Chapter 7, I return to a more specific account of the threats modification effects pose to discursive democratic goods. For now, and as I argued in Chapter 4, target features such as lack of knowledge, low engagement and psychological 14921 Recall that here we are talking about frames as features of discursive ecosystems. The ecosystem itself may contain a number of frames, each of which may be accessible to a speaker or target, and each of which might raise different considerations. To the extent that frame competition can compensate for the harmful effects of frames taken individually, we may be less concerned about autonomy considered from the system level. However, as Calvert and Warren (2013) suggest, highly polarizing frames in competition with one another may undermine or preclude any salutary effects of discussion that cuts across frames.predispositions can preclude or compromise reciprocity, independently of the arrangement of inferential structures in speech. The absence of commitments precludes reciprocity and thus authenticity, a core element of autonomy. The implication of the absence of commitments in a low-knowledge situation is this: if speaker?s articulate rhetorical structures?which are rigid and non-transparent?these structures become manipulative in the absence of target commitments that generate reciprocal force. Targets have only a gap in their web of commitments, where otherwise there would be beliefs and values for the rhetor to engage. This is why explanation is a critical corrective to rhetoric, as I argue in Chapter 9. Explanation can fill in actors? webs of commitments and, by articulating a number of connections, better integrate into the target?s existing beliefs to promote understanding. As a result, rhetoric will present less of a risk of creating manipulative inferential effects because actors will be better positioned to challenge or endorse inferential structures on their own, authentic terms.Similarly, I suggested in the previous chapter that having a stake promotes more deliberate processing, creating the conditions for greater reciprocity independent of the inferential structure itself. In terms of evaluating subject-based features from the perspective of democratic goods, having a stake is neither good nor bad: people are unlikely to reach for and actively engage their own commitments on a claim or argument if they cannot see how the issue affects them. As I develop in detail in Chapter 7, anti-democratic hazards arise when people cannot see their stake on issues that do affect them, whether directly or peripherally. Though autonomous judgments are generally better than non-autonomous ones, the issue of autonomy is particularly pressing in situations where the all-affected principle tells us that a choice ought to be empowered.150On the speaker side, ecological features magnify a speaker?s influence by conferring a status that favours uptake of their commitments due to their assumed credibility, whether deriving from trust, expertise, first-hand knowledge, shared identity, or shared interest. Under such circumstances, assertions can even exert influence beyond their status as single commitments. The conditions that govern credibility can alter communicative influence in two ways, both of which are normatively troubling. The first condition is relatively straightforward. The dangers of magnification effects depend on what is being magnified, and this simply kicks the issue back to the evaluation of different types of communicative influence?persuasion, explanation, discussion and the like. In other words, the magnification of persuasive influence may be acceptable, but the magnification of manipulative influence is not. The second condition concerns whether or not actors have sufficient grounds for making credibility judgments. One way of thinking about the issue of how targets make credibility judgment is this: magnification effects generate supplementary influence, over and above what inferential structures generate. This supplementary influence should be subject to the same conditions of authenticity and reflexivity as influence deriving from inferential structures. Actors need to be able to see how sources of credibility connect to their own commitments. Regardless of what speakers actually say, if the grounds for attributing credibility to them are non-transparent, then there is also cause for concern. For instance, where credibility derives from cues like party affiliation or issue-ownership, there is a risk that some commitments will remain implicit, and yet an inferential structure will nevertheless receive uptake.22  While substituting judgments of speakers for judgments of content has the 15122 Further, certain important arguments will not be aired at all because speakers know they are not perceived as credible on that issue. advantage of freeing up cognitive capacity and helping citizens allocate scarce resources for learning?a key argument in the trusted proxy literature (Lupia and McCubbins 1998)?the downside is that when such judgments are based on non-reflexive cues, influence may be magnified in a way that the influence target may not reflexively endorse. They are being moved, not by the force of reasons, but by the force of the cues themselves. The question of whether the influence that follows from magnification effects is consistent with the demands of autonomous judgment comes down to whether people have made judgments of credibility on the basis of reasons that they can reflexively identify and authentically endorse. In Chapter 8, I suggest that institutional arrangements can provide people with reasons. For instance, judges and juries are issue statements following a commitment to hearing arguments in a way that lends weight to their eventual judgments, and also creates space for advocates to focus solely on the needs of their clients and the demands of their case. In both cases, recognized institutional roles provide reasons that help people to decide how to use what they know about speakers to inform how they interpret claims and arguments?perhaps giving different weight to the content of a judges reasoning than a lawyer?s.Institutional ecological featuresInstitutional features of the discursive ecosystem determine who speaks to whom, when, and with what audiences. They determine how inferential structures translate into concrete outcomes, be they decisions, votes, policies or individual opinions. Institutional ecological features condition communicative influence by way of selection effects. In Chapter 4, I argued that there are two kinds of institutional ecological feature: generative and transactional. Generative institutional features are those that embed actors? commitments into 152the discursive ecosystem, rendering them accessible to interlocutors and audiences for attribution and, sometimes, reciprocal leverage. They might include, for instance, voting records on bills, signatures on petitions, or recorded comments. Generative institutional features, by themselves, are difficult to evaluate. On the one hand, getting more powerful actors? commitments on the record can be a tool available to disempowered actors seeking to leverage ?the civilizing force of hypocrisy? (Elster 2011). On the other hand, we can imagine situations in which actors are forced to acknowledge something, thereby generating a commitment to which they can be held absolutely and indefinitely. For instance, the coercive power of an interrogator combined with the generative aspect of a written confession risks creating a situation in which a forced confession creates extreme consequences. In a less extreme case, actors with less power certain rules or premises just to get a seat at the table, only to have those commitments used against them. Marginalized groups might have to capitulate to certain commitments in order to participate in an exchange, potentially in ways that place them at a starting disadvantage. Similarly, political representatives may be trapped into voting for amendments in order to get a policy passed. Generally speaking, generative institutions create opportunities to leverage reciprocity into rigid force because there are external, institutional pressures on actors to acknowledge their commitments. However, the normative evaluation of such practices depends on how those commitments were generated in the first place?through forced confession, to get a seat at the table, or to get a bill passed. Questions for the normative evaluation of such circumstances might include an evaluation of the communicative influence type that led the speaker to undertake a commitment in the generative environment153?was it manipulative? Or was there some other, non-linguistic form of coercion or inequality that ought to call into question the integrity of the generated commitment.Transactional features?those that determine the flow of influence from speakers to targets and whether they switch roles, the parties to an exchange, and the scale of audiences?similarly raise competing normative considerations. Deliberative democratic theory, as it currently stands, tends to favour dialogue in all situations. Without rational dialogue, one cannot be said to have genuinely tested propositions and norms. Yet focusing on the rational validation of arguments through ideal discourse is misleading. In principle, there is no reason that one-way communication cannot further the discursive goods envisioned in deliberative accounts of democracy (see Goodin 2003).Generative and transactional features combine with actors? incentives to select for different types of influence. This selection effect is key to a more systematic analysis of the normative potentials and pitfalls of institutions. As I suggested in Chapter 4, features of political institutions can select against transparency, against reciprocity, and towards rigidity?in other words, they can favour manipulative influence. To the extent that they do have this selection effect, they undercut autonomy, and for this reason they demand the kinds of direct and corrective institutional interventions I develop in Chapter 9.23  Beyond that, I argue that democratic theorists should look for pathological selection effects?which is to say, the effects of institutions that select for types of influence that go against their purposes. For instance, if a site of discourse was meant to produce explanations?for instance, a public 15423 Other accounts of manipulation focus on the level of agenda setting, and preventing those forms of manipulating the process in ways that avoid arguments and judgments?moving unfriendly amendments or filibustering, for instance?are important for many democratic values. But this is a different kind of manipulation from manipulative communicative influence. Both work to bypass or avoid reflexive, authentic judgment, but only the latter uses people?s own reasoning capacities against them. information session?and instead produced rhetoric, there would be cause for concern. Similarly, sites meant to produce discussion that turn to manipulation would, as I argue below, compromise the overall conditions for autonomous judgment.The idea of pathological selection, of course, raises the critical question of what types of communicative influence selection effects at different sites ought to aim towards, particularly if they are to constitute the conditions under which the discursive democratic goods of the motivating force of reasons and freedom in discursive will formation can thrive. Answering this question requires a theory of deliberative political systems that conceptualizes how sites of deliberation ought to compensate for and complement one another. Rather than seeking persuasion from every institution?and thus superficially appearing to meet the criteria of autonomy?I suggest that different sites should select for different types of influence in the broader deliberative system. In Chapter 9, I develop a detailed argument that communicative influence types should balance one another, since an excess of any one type can become pathological. Too much discussion, and communication cannot accomplish anything. Too much rhetoric, and people will never acquire new commitments. Too much disclosure, and people will never understand one another. In each case, I argue the pitfalls of an excess of one type can be compensated for by its opposite?with the obvious exception of manipulation. Finally, institutions can also correct for some of the anti-democratic hazards of magnification and modification effects, and can be judged as well in terms of their ability to do so, an argument I develop in Chapters 7 and 8.155ConclusionIn this chapter, I suggested that the model of communicative influence offers an alternative way of conceptualizing the relationship between speech and the normative aspirations of deliberative democrats. Ultimately, the democratic justification for reason-giving as a specific linguistic practice comes down to two discursive democratic goods: the motivating force of reasons, which allows people to apply obligations and constrain themselves; and freedom in discursive opinion and will formation, which ensures that people take on these obligations only on the basis of their agreement to the validity of the reasons. Taken together, these two discursive democratic goods reflect the broader principle of autonomy, or the idea of acting intentionally, for reasons that one can recognize as one?s own. The elements of autonomy?judgment, reflexivity, and authenticity?map on to the dimensions of communicative influence, providing a framework for its normative evaluation.The dimensions of rigidity, reciprocity, and transparency can be used to evaluate rigid forms of speech in terms of their consistency with autonomy. Less directly, we can evaluate non-rigid influence types in terms of their effects on the discursive ecosystem, because the discursive ecosystem ultimately feeds back into the autonomous quality of judgment in the face of communicative influence. In the third section, I explained this feedback in terms of how structural/substantive, subject-based, and institutional ecological features create or undermine the conditions for autonomy. The normative argument in this chapter suggests that one can achieve, or at least approximate, many of the goods identified with ideal theory, in a way that is attentive to the strategic conditions of politics. In Chapters 7 to 9, I develop the idea that one can look at institutions, not just in terms of how closely or distantly they resemble an ideal forum, but rather in terms of the kinds of influence they allow, incentivize, 156or penalize. In addition, in Chapter 6, I apply the model of communicative influence to an assessment of the pervasive patterns of political communication: first, in terms of how they use inferential structures to exert influence; second, in terms of the effects of their use on the discursive ecosystem; and third, in terms of how both of these features of political communication together cause these generic features of political communication to either enhance or compromise autonomous judgment.                                                                           157Chapter 6. Patterns of Communicative InfluencePart of the reason political discourse rarely resembles the straightforward exchange of reasons envisioned in much of deliberative democratic theory simply lies with the way people talk in everyday politics. Speech is full of metaphors, analogies, ad hominem arguments, stories, hyperbole, selective argumentation, and other devices, all of which can be deployed strategically. These recurrent features of political discourse structure reason giving, often with a strategic gloss that puts it at odds with much of deliberative theory. My aim in  this chapter is to treat these aspects of political discourse as patterns of communicative influence: identifiable, generic inferential structures that speakers fill in with specific content and commitments. When speakers use patterns of communicative influence successfully, it will often be because they are attending to the ecological conditions I detailed in Chapter 4. Speakers must be attentive to what resonates with targets: which analogies prompt them to draw the desired connections between claims? Which metaphors have entailments they will recognize? Speakers must also attend to how patterns of communicative influence alter discursive ecosystems, granting credibility, establishing interpersonal relationships, and perpetuating frames. In this chapter, I use the model of communicative influence to shed light on these and similar pervasive features of political communication, and I identify their implications for the theory of autonomy developed in Chapter 5. There has been a tendency to criticize deliberative democracy in the Habermasian tradition for undervaluing modes of communication that do not fit into the rationalist, logical understanding of ?reason-giving.? The corresponding call to admit alternative forms of communication?notably narrative (Young 2002; O'Neill 2002)?has received broad acceptance. At the same time, this move 158has not been accompanied by an account of what it is about what the act of storytelling?as opposed to its content?contributes to deliberation. From a pragmatic perspective, stories arrange claims in a specific way. As I show, however, storytelling in fact arranges claims and commitments in a very specific way. Deliberative democrats who idealize rational dialogue assume that the persuasive force of reasons is, in principle, separable from the form in which those reasons are conveyed. The notion of inferential structures I have been developing in this dissertation offers a way of moving beyond this assumption. In this chapter, I use the concept of a pattern of communicative influence to explore some of the ways in which the form of claims take gives them force that exceeds their content. In fact, a claim?s form can also alter the character of its influence?not just narratives, but also comparisons, evocative language, deception, and evading commitments. Elaborating the concept of an inferential structures?whose rigidity, reciprocity and transparency can vary?makes it possible to reflect on the way patterns of communicative influence generate force, so that we can pinpoint their implications for autonomous judgment. This chapter proceeds as follows. I offer a pragmatic analysis of five distinct patterns of communicative influence, each in turn: first, storytelling, anecdotes, and testimony; second, analogy and metaphor; third, persuasive definitions and loaded words; fourth, exaggeration and deceit; and fifth, evasion and hedging. I identify each pattern?s pragmatic properties and ecological functions. Pragmatic properties describe how patterns of communicative influence arrange commitments and inferential connections. Ecological functions refer to the patterns? effects on the broader discursive process. For example, through ridicule, a speaker can distance himself or herself from a position he or she does not find credible (Van Laar 2008). Based on both pragmatic properties and functions, I discuss 159each mechanism?s effect on communicative influence in terms of rigidity, reciprocity, and transparency. On this basis, I evaluate what each pattern implies for the theory of autonomy introduced in the previous chapter. Table 6.1 lists the patterns of communicative influence I cover in this chapter.Table 6.1. Summary of Patterns of Communicative InfluencePattern of Communicative Influence Key FormsNarrative Storytelling, anecdotes and testimonyComparative Metaphor, analogy, example and illustrationEvocative Persuasive definitions and loaded wordsDeceptive Lies and exaggerationAvoidance Evasion and hedgingNarrative Patterns: Testimony, Anecdotes, and Story-tellingCritics of overly rationalist accounts of deliberation have insisted that inclusive dialogue requires a valuation of alternative forms of communication, the chief among them  being story-telling, narratives, and testimony (O'Neill 2002; Ryfe 2005; Young 1997, 2001, 2002). Some empiricists have incorporated these calls for diversity into measures of quality of deliberation (Stromer-Galley 2007). Given the attention it has received, narrative provides a good starting point for my exploration of patterns of communicative influence. We can think of narratives, particularly stories, as linking fictional events or facts, often in ways that establish relatable dilemmas, develop character, or make sense of personal biographies. For example, success stories involving self-made men or women who pull themselves up by their bootstraps imply a simple connection between hard work and success. But in doing so, they tend to also downplay the role of external supports. This is because stories tend to revolve 160around a single central problem (Polletta and Lee 2006, 702) and, in doing so, they convey the overriding importance of that problem. Think of a representative tells a story of a constituent?s personal struggle in order to convey the immediacy of a problem he or she hopes to address. Here, the problem becomes the overriding theme of the narrative. Within the category of narrative, I distinguish anecdotes by virtue of their specificity and the implication that an anecdote truthfully reports events without embellishment. Their ?groundedness? tends to be one of their important features. For example, politicians? personal anecdotes can work to convey a sense of authenticity, or of familiarity with voters? concerns. This said, scholars, Young acknowledges that stories can also be used manipulatively (2002, 79). And yet deliberative democratic theory currently lacks meaningful criteria for determining when story-telling might actually be harmful to deliberation. Pragmatic properties of narrative patternsNarratives connect a series of assertions in ways that embed the relationships between them. They leave inferences implicit, although targets can infer the if/then and cause/effect relationships from the requirement that narratives form coherent wholes, rather than random sets of claims. For example, an anecdote about a 100-year-old voter making the trip to the polls on election day augments a claim about the importance of voting, with the implicit argument that if he or she can do it, so should others. Unlike straightforward reason-giving, stories integrate description, explanation, and evaluation (Polletta and Lee 2006). At the same time, stories remain detached from broader discourse, standing as relatively self-contained units. Because of this detachment, the merit of inferences among events and facts are instinctively first judged in terms of a story?s coherence, as opposed to in terms of their independent validity (Polletta and Lee 2006, 702) as an ?exchange of reasons? model might 161lead us to believe. Narratives can also explain experiences, which are necessarily unique to each person (Young 2002, 75). To the extent that narratives are articulated and accepted, they can enable reciprocity by establishing shared premises for joint reasoning (Young 2002, 7). Polletta and Lee also find that stories can function hypothetically as tools to work out the implications of a given position (2006, 712). Using narratives to spin out the implications of a position then provides others with a basis for challenging or agreeing with the commitment that led to that inference.Testimony, for its part, is distinguished from other narrative-based patterns by the speaker?s position of privileged access based on observation or direct experience to the propositions that make up a narrative. Testimony?s influence derives from this explicit, privileged entitlement. The attribution of privileged entitlement may derive in part from institutional settings and the roles they establish. For instance, a ?witness? or ?expert,? or anyone called to give an account of events, may enjoy privileged entitlement simply by virtue of having been selected for the veracity and insight of their particular account. They might be required to make an explicit commitment to truthfulness, or asked to give an account of their credentials as part of their testimony. But like storytelling, testimony is also judged in part in terms of whether it makes sense on its own terms?or whether it is self-contradictory or incoherent, as we know from the genre of courtroom dramas. Ecological functions of narrative patternsGiven these properties, the effects of stories, anecdotes, and testimony on the surrounding discursive process can vary widely. Narratives have the potential to become part of the substantive structure of the discursive ecosystem, particularly as they contribute to framing effects (Gamson and Modigliani 1989; Nelson and Kinder 1996). For example, 162stories about ?welfare mothers? perpetuate stereotypes that prompt inferences about welfare spending in general (Kuklinski and Quirk 2000, 170). ?Founding? stories like those around the US constitution can implicate values, often in a way that implies that what the founders had envisioned ought to trump competing considerations. On a smaller scale, in his study of group deliberation at AmericaSpeaks events, Ryfe finds that, left to their own devices, people tend to produce and then deliberate around competing narratives on an issue (Ryfe 2006, 81). Narrative patterns can guide and direct group reasoning processes by imposing some structure on the considerations actors deem to be relevant.In terms of speaker-based ecologies, narrative patterns may serve to establish credibility in relation to a given issue. Beyond enabling people to lay claim to first-hand experience, stories establish social identities (Young 2002, 73), and anecdotes allow speakers to show that they are ?no different? from those to whom their speech is directed by generalizing their experiences. Stories allow people to offer accounts that they hope will protect their reputations in the face of failure (McGraw 2001).24 On the target side, identities conveyed through stories may help people identify their own obligations and commitments, and to consider what their preferences should be in light of shared experiences (Polletta and Lee 2006, 704; Ryfe 2006, 76).25  16324 In the pragmatic terms of the model of communicative influence, we might think of the use of stories to give accounts as a form of explanation, in the sense that they enable a speaker to show why a given set of circumstances, encountered sequentially rather than all at once, provides adequate entitlement (a reason for) a given course of action. Even though the account might not have the same compelling force for a target to commit to that same outcome, they might nevertheless demonstrate the speaker?s entitlement. 25 If one expands Brandom?s notion of the ?institutional ought? to include not only institutional roles but social roles in general, this notion of identities incurring commitments links up nicely with Brandom?s account of obligations?patterns of practical reasoning that constitute the middle ground between preferences and moral, or universal, oughts...For these reasons, stories play key roles in activist ?consciousness raising? tactics. Finally, stories work to establish and preserve interpersonal relationships. Empirical findings show that deliberation enables people to disagree without having to express their disagreement directly. Instead, they may say something like ?I agree, but...? followed by a story that expresses an opposing point of view (Polletta and Lee 2006, 704; Ryfe 2006, 79). This tactic allows people to avoid either actually antagonizing one another or worrying about antagonizing one another, in part because stories establish ?personal? truths?expressions that belong to the speaker, and cannot be refuted by someone else?s experience. Implications of narrative patterns for autonomyConsidered by themselves, stories imply non-rigid forms of influence. Particularly in their premise-establishing capacity, stories are not typically arguments in themselves?although they may link up into larger arguments, filling places within rigid inferential structures. But because they do not limit choice or constrain judgment, I argue that there should be few concerns about the immediate or direct implications of stories for autonomy. But their ecological effects do have important normative implications. Stories work by providing a basis for mutual understanding. To the extent that they are relatable and resonate with people?s experiences, they create the conditions for reciprocal inferential structures, prompting others to listen empathically (Polletta and Lee 2006, 703) and to consider what is said in light of their own experiences. As Michael Morrell (2010) argues, empathy is critical to any account of deliberative democracy. The process of empathy, he argues, helps give truly equal consideration to all insofar as it enables people to overcome bias through perspective-taking, and to be attentive to a wider range of considerations?not just explicit reasons referencing a generalizable good, but to affective or particular claims (Morrell 2010, 162). 164Narrative patterns of communicative influence?stories and anecdotes?are particularly well suited to facilitating such an empathic process. From an entirely different perspective, Hafer and Landa develop a model that shows how pre-decision communication has consequences for interaction insofar as it can show people how what they already know is relevant to the decision at hand (2007, 332). Their proposition represents another way of thinking about the reciprocal functions of stories: they can prompt people to retrieve relevant considerations from their own experiences, using them to process or engage the claims a narrative pattern advances. In some instances, however, stories may themselves be rigid or argumentative. For example, tragedies or just-so stories that articulate lessons use narrative patterns to demonstrate the consequences of taking, or failing to take, some action. But to the extent that stories allow these features to remain implicit, they undermine transparency. These implicit connections may be fairly easy for the hearer to deduce; indeed, it is probably something we often do instinctively, otherwise stories would seem like random strings of facts and events. That being said, they do have the potential to be built into rigid inferential structures in a way that undermines transparency by rendering inferential connections among claims implicit. Such uses of stories are rhetorical to the extent that they work by mobilizing prior commitments. They become manipulative when they depend on implicit premises supplied by the speaker that, if made explicit, would not be acceptable on the target?s terms. Finally, the reciprocal potential of stories may depend on their cultural status. To the extent that stories have a ?sacred? status, they implicitly place them beyond challenge. Invoking them can thus work to exclude countervailing considerations, compromising reciprocity. Although some stories may themselves be rigid, rigidity may not actually be the 165primary issue of concern in most cases. As I suggested in Chapter 3, there are two ways for inferential structures to derive force: from their logical arrangements and from buy-in on the basis of the target?s own commitments. Stories are powerful because they can tap into a target?s commitments by invoking shared knowledge and values, and speakers can leverage reciprocal buy-in into influence using even flawed, limited, or non-transparent rigid connections?all while contributing framing effects that might discourage people from raising competing considerations. As such, while not necessarily damaging to autonomy, the influence they generate has the potential to be damaging with only a few imposed inferential connections, lending a structure force without transparency. Testimony raises slightly different normative issues from stories and anecdotes. To the extent that a target is being asked to accept something on the basis of a speaker?s privileged position to report facts, the premise that he or she speaks with authority is working implicitly in the inferential structure. Such authority premises lend themselves well to rigid inferential structures because they compel attributions of entitlement without contestation. Typically in circumstances where testimony is taking place, that entitlement should be sufficient to compel the target to undertake the commitment themselves. That being said, this compulsion may be less, to the extent that what the speaker is asking the target to accept is steps removed from his or her privileged entitlement?for instance, where the speaker reports something that he or she thinks follows from his or her direct observation, or a claim that is slightly removed from his or her direct expertise. Rendering the sources of entitlement that qualify speech as testimony transparent ensures that the influence deriving from the speakers? privileged entitlement to claims remains consistent with the reflexivity criteria for autonomy.166Testimony is distinct from storytelling in part because it is articulated under the assumption that it will be attributed a high truth value. Institutional circumstances may achieve this simply by inviting or prompting the kind of privileged claims that constitute speech as testimony, although they may vary in the degree to which they confine speakers to just the facts, or even a subset of the facts (excluding hearsay, for instance). The context in which a speaker testifies promotes the uptake of claims because of the speaker?s privileged entitlement. As a result, the hazards of testimony most resemble those of magnification effects. As I argue in Chapter 8, whether magnified influence is problematic for the autonomy of the judgments that follow depends primarily on whether targets have good reasons to attribute entitlement or credibility. Both stories and testimony, considered under the model of communicative influence, raise complex normative issues. Their pragmatic properties and ecological functions cut two ways, with the potential to enable and undermine autonomous judgment. Institutional arrangements can sometimes directly address the use of narrative patterns to promote autonomous judgments. For example, legal disputes in court are often presented in terms of competing narratives that might explain a set of facts. The rules of evidence are designed to ensure that narratives cannot exert undue influence simply as a result of being presented to a jury in narrative form. As such, advocates can use narratives to attempt to exert some measure of control over judgments, but their ability to leverage the power of the narratives is counterbalanced by grounding those narratives in shared, objective facts to which all parties must accede. At the same time, strict, neutrally adjudicated procedural rules cannot achieve this effect in the uncontrolled domain of the public sphere, where advocates also use narratives to shape judgments. That quality public sources undertake fact-checking roles is 167thus particularly important to protecting the conditions for autonomous judgment. Finally, institutions may push for more complete transparency in the articulation of inferential structures by establishing norms or rules against narratives, as, for instance, with the publication of scientific findings.Table 6.2. Narrative Patterns of Communicative Influence: SummaryStories and Anecdotes TestimonyPragmatic Properties- leave causal connections between events implicit.- judged first in terms of coherence- can establish shared premises- distinguished by privileged sources of entitlement (expertise or first-hand experience)- force derives from this entitlementEcological Functions- perpetuates frames- can establish social identities and preserve or build interpersonal relationships- can help establish credentials.Rigidity - generally non-rigid, except where morals or lessons are involved- consistent with relatively rigid or non-rigid structuresReciprocity - can prompt reciprocity through empathy and relatability- not reciprocal because sources of entitlement are unique to the speaker.Transparency - leaves many inferences implicit - power derives from transparency of sources of entitlementThreat to Autonomy?- They have the potential to be, because they generate reciprocal buy-in, typically non-transparently, that can be leveraged into rigid influence rhetorically or manipulatively. - Not typically, though testimony does favour uptake, on the model of magnification effects, because speakers are treated as credible. Institutions should be sensitive to abuses of this credibility.Comparative Patterns: Metaphor, Analogy and ExampleInfluenced by Lakoff and Johnson?s book Metaphors We Live By (1981), scholars now recognize metaphor as a critical part of our linguistic and cognitive structures. Metaphors are powerful linguistic moves that can have a direct effect on policy preferences 168(Bates 2009; Barry et al. 2009). Even if targets do not immediately accept a metaphor, research shows that they can nevertheless appear ?downstream? in a conversation, potentially altering its course (Cameron 2007, 109). What does this mean for communicative influence? Lakoff and Johnson concluded that metaphors are significant, not for their truth value, but for the inferences that follow from them (1981, 158), making a cognitive theory of metaphor a natural fit with an inferentialist theory of discourse. I explore this overlap, but restrict its domain. Since the broader purpose of this discussion is to construct an answer to the problems posed by strategic speech, I focus on deliberate metaphors, selected and deployed consciously to induce a change in perspective through comparison (Steen 2008, 214). For the purposes of this discussion, I define metaphor as a cross-domain comparison that is necessarily partial and highlights similarities (Barry et al. 2009). An analogy is generally viewed as a more complex version of this relationship (Musolff 2004) involving resemblances of relationships: A is to B as C is to D (Perelman and Olbrechts-Tyteca 1969, 371-72).26  Many of the considerations I identify below that apply to metaphors and analogies also apply to examples by virtue of their similar inferential properties.27 Pragmatic properties of comparative patternsMetaphors and analogies work by highlighting some considerations and, in the process, hiding others (Barry et al. 2009; Lakoff and Johnson 1981). Metaphors may bring 16926 Barry et al. (2009) offer a slightly different take, claiming that whereas metaphors work across domains, analogies are within the same realm of experience. I prefer Perelman and Olbrecht-Tyteca?s account, which emphasizes the resemblance of structures. Moreover, an analogy within the same realm of experience may be an illustration or an example.27 As Doury (2009) points out, the line Perelman and Olbrechts-Tyteca draw between analogy and example?that the former maps across domains and the latter within them?is more of a continuum. For example, are cases of the same type, but from different historical periods?like precedents?examples or analogies (Doury, 2009, p. 145)?about a change in perspective (Dascal and Gross 1999; Steen 2008, 2009), potentially introducing new sets of considerations to an inferential structure by getting targets to think of an issue in a different way. Both metaphors and analogies allow actors to draw inferences from situations with familiar facts to other cases in a process of cross-domain inference (Lau and Schlesinger 2005). Analogies also enable hypothetical reasoning about the consequences of a course of action (Musolff 2004). They enable actors to challenge one another by rendering explicit further, untenable entailments of analogies to which speakers have committed themselves. A key feature of both metaphors and analogies is that they have entailments. In pragmatic terms, entailments are inferences that can or must follow from endorsing a metaphorical or analogical relationship (Lakoff and Johnson 1981), with committive or ?must? type inferences rendering inferential structures more rigid. At minimum, committing oneself to a metaphor commits one to the pertinence of the features being compared (Bates 2009; Perelman and Olbrechts-Tyteca 1969, 390). Entailments are interesting pragmatic properties for strategic reasons as well. Targets can refute metaphorical and analogical claims, not just on the basis of the appropriateness of the metaphor, but also by pointing out entailments that would entangle the speaker in a contradiction. According to Musolff, some analogies are grounded ?source scenarios? that tend to generate many assumptions about actors? roles and states of mind (2009, 28). Such analogies in particular lend themselves to exchanges on the basis of extended entailment. At the same time, speakers can defend challenges based on entailments by shrinking the scope of the metaphor (Musolff 2004; Perelman and Olbrechts-Tyteca 1969, 387). 170Examples also fall under comparative patterns of communicative influence. They work by enabling a form of incomplete generalization among cases (Benoit 1980), allowing actors to apply previous experience to new situations (Arthos 2010). Examples can have a variety of functions in discursive exchanges, enabling generalizations, grounding analogies, adding forcefulness, or generating standards of judgment (Arthos 2010, 321).28  According to Beno?t, Aristotle thought that examples were more persuasive than deductive arguments, despite the latter?s compelling force, because of their accessibility and resonance with experience (1980, 190).Ecological functions of comparative patternsLike narratives, comparative patterns can exert powerful framing effects. Once a comparison is established, it structures the commitments around an issue, rendering certain inferences more or less likely, and certain considerations more or less salient. Analogies establish precedent and direct attention (Aronovitch 1997)?indeed, reasoning by analogy makes up much of legal dialogue. Lau and Schlesinger identify a similar structuring effect with metaphors, claiming that they set up ?archetype? solutions to social problems so that people can form preferences without necessarily understanding the technical details?for instance, on how they think families, communities, or markets should work (2005, 79).29   From the perspective of subject-based ecologies?which depend on the credibility and records of speakers?speakers must be cautious about the entailments of metaphors and analogies. Articulating a comparative pattern has consequences for the immediate context of 17128 Examples can also take the form of stories and anecdotes. As such, the latter can also serve these pragmatic functions, insofar as they are deployed as examples.29 They also ask what the existence of these non-technical archetypes mean for quality of deliberation (Lau and Schlesinger 2005, 106), a question this framework is intended to help answer.a discursive interaction because actors incur commitments to metaphorical and analogical entailments. Opponents can use these entailments to challenge the speaker by spelling out implications to which they take the speaker to be committed. Opponents might also challenge the speaker by constructing a parallel analogy intended to ridicule, along the lines of ?that would be like saying?.? (Whaley and Holloway 1997). Comparative patterns thus have implications for credibility in the short and medium-term contexts of discursive interaction because speakers may have unwanted entitlements attributed to them on the basis of the comparisons they draw. Implications of comparative patterns for autonomyMetaphors and analogies have high potential for rigidity. Because they structure inferences, they exert a powerful pull on the reasoning processes of those who hear them. For example, comparisons between the Kosovo crisis and World War II with respect to ethnic cleansing highlighted a reasoning pathway directly from a description of events to the decision to intervene (Bates 2009). Other analogies may leave more room for maneuvering, as Musolff?s example of EU negotiations demonstrates. To the analogical threat, ?the European train is leaving the station,? Margaret Thatcher replied that it would be better not to be on the train if it were heading the wrong direction (Musolff 2004, 31). In this situation, an analogy that was intended to be rigid was diverted by virtue of its reciprocity, that is, its appeal to common knowledge. As such, although comparative patterns have more directive or rigidity potential than narrative patterns, the threats these features pose to autonomy can be mitigated by ensuring that comparative patterns are reciprocal and, to the extent possible, transparent. In the above EU-as-train example, the fact that it was a familiar expression meant it had entailments to which opponents could latch on?in this case, to question the 172?destination.? Casting situations metaphorically in familiar terms may even be one way to enhance reciprocity, insofar as familiar metaphors enable people to bring their own knowledge and experiences to bear. At the same time, comparative patterns can sometimes threaten autonomy. Though they depend on reciprocity for force?targets must understand a comparative pattern and follow its entailments for there to be any effect?the considerations they exclude might be those with the potential to halt or redirect the force of an inferential structure. Strategic actors can build analogies into inferential structures in a way that makes disagreement difficult or unpopular, or they can silence criticism by appealing to seemingly unproblematic source scenarios (Musolff 2004, 174). Indeed, metaphors and analogies pose risks for transparency?and thus reflexivity?in that they render inferences and assumptions implicit. Non-transparency is a risk with examples, in that they can be used highly selectively. Of course, for reasons of expediency, and because of our cognitive biases towards confirming evidence (Mercier and Sperber 2011), this risk is not necessarily malicious or even strategically deliberate. Whether deliberate or not, insofar as implicit choices govern the selection of examples, analogies, and metaphors, democratic theorists should be attentive to the ways institutions might render these implicit aspects of comparative patterns explicit and subject to evaluation. In situations where targets have insufficient knowledge or information to weigh the suitability or accuracy of an example, its use cannot be reciprocal.30 In order to avoid having such practices damage autonomy, the grounds for the comparison must be either transparent, 17330 It is possible that, with an insufficient basis for understanding a metaphor or analogy, targets may simply slough off any of their effects. A more troubling possibility, though, is that they may understand a comparison sufficiently to make the inferences it prompts, without having adequate grounds to identify and evaluate the implicit basis of the comparison. or its use must be non-rigid. Institutions might also limit the influence of some types of comparative patterns. For instance, hypothetical analogies may be useful for working out the implications of a position, but because they are hypothetical, targets may not be able to judge their validity. Hypothetical examples and analogies, then, may be limited to providing points for discussion and explanation, and not enabling directive, rigid force. I summarize comparative patterns of influence in Table 6.3.Table 6.3. Comparative Patterns of Communicative Influence: SummaryMetaphor, Analogy, and ExamplePragmatic Properties- highlight some considerations and hide others- commit speakers to the relevance of the similarities- generate entailments?inferences that follow from the comparison- can introduce new considerations by shifting perspectivesEcological Functions- contribute to framing effects- help people form beliefs by rendering issues in familiar terms- in using a comparative pattern, speakers may incur commitments on the basis of which others can challenge themRigidity - potentially high, depending on the nature of the comparison. For instance, if case x is similar to case y, and there was only one answer to case y, there may only be one answer in case x.Reciprocity - comparative patterns can either contribute to or limit reciprocity. They can draw in new commitments by prompting people to think of an issue in different terms, but they can also exclude considerations by defining an issue as being of a certain type.Transparency - comparative patterns work against transparency, though perhaps to a lesser extent than the other patterns, depending on the nature of the metaphor/analogy. They do generally render assumptions about the appropriateness of the comparison implicit.Threat to Autonomy?- yes, because they have the potential to combine rigidity, non-transparency and non-reciprocity. To the extent that comparative patterns are deployed rigidly, there should be scope for targets to reflexively weigh them and consider whether the comparison is in line with the target?s own commitments.174Evocative Patterns: Persuasive Definitions and Loaded WordsDeliberative democrats have paid much attention to narrative and storytelling, while cognitive linguistics has tended to focus on metaphor and analogy. By contrast, work on persuasive definitions, along with loaded language and value-laden terms, has been fairly limited (Walton 2005, 172). Yet such persuasive structures are central and persuasive features of discursive practice. Evocative words have the potential to amplify communicative influence, extending the impact of inferential structures beyond what one might expect from their substance alone. And, of course, strategic actors fully recognize the value of well chosen words, especially in politics. For example, over the last decade, actors in the Canadian energy industry have avoided the negative image of the ?tar sands? by always using the more neutral term ?oil sands.? Normatively speaking, we need not have a problem with using value-laden terms (Walton 2005, 173). Indeed, we would be hard-pressed to escape them, particularly in political discourse that necessarily implicates values, stakes, and priorities. Still, definitions can function as a powerful form of communicative influence, extending or reducing the extension of concepts that confer value on the objects to which they are applied. The term ?democracy? is a case in point. Speakers often use the word ?undemocratic? without necessarily intending to make reference to the process through which the decision was reached. Rather, they mean to cast the decision in a negative light, invoking public value judgments in an effort to delegitimize a decision with which they do not agree. An analogous pattern concerns what we might call unloaded words: acronyms or bureaucratic terms that replace language that might otherwise evoke strong responses, for 175instance ?enhanced interrogation? for torture or ?collateral damage? for civilian casualties.31 We can think of unloaded words as working in the opposite way of evocative patterns in general. They fail to evoke the relevant legal, moral, or ethical considerations that a more direct term might be expected to. While I focus primarily on loaded language and persuasive definitions here, where possible I identify theoretical similarities and dissimilarities with unloaded words.Pragmatic properties of evocative patterns of communicative influenceEvocative patterns of speech like word choice or implicit definitions can also exert substantial communicative influence. Although my analysis here is grounded primarily in work on persuasive definitions, the unifying feature across evocative patterns is that words serve as implicit arguments, evoking inferences by virtue of how people understand and react to them. Macagno and Walton conceptualize a definition as ?persuasive? when it leads to a commitment through either argument from classification or argument from value (2008b, 204). In evocative patterns, words function as predicates, imposing conditions on subsequent argumentation and reasoning (Macagno and Walton 2008a). For example, in US politics the Republicans? false labeling of the end-of-life counseling provision as ?death panels? in President Obama?s Affordable Healthcare Act evokes the notion that a committee decides whether patients will or will not receive care (Neblo 2012, 181). The very definition of end-of-life counseling as ?death panels? evokes a chain of inferences ending in the conclusion that the bill obviously goes against common sense, ethics, and respect for the elderly. When word choice exerts influence through classification, evaluations associated with a term in its original meaning can be applied to an object covered by a new or expanded 17631 I thank Chris Kam for this suggestion.definition.32 For instance, conservative efforts to attach the label ?carbon tax? to the NDP?s proposal for an emissions trading system was based on the insistence that it was a tax based on the fact that it would lead to government revenue?and that the comparable conservative plan was not a tax, simply because it would not. They attempted this maneuver despite an earlier claim that anything establishing a price on carbon was by definition a tax (Wherry 2012). The evocative strategy was to use the label ?tax? to differentiate the two plans in a way that draws on the general aversion to taxes to make one seem automatically better.Strategic speakers thus seek to change the extension or coverage of a term, while preserving the evaluations associated with that term?and the inferences from value that follow (Macagno and Walton 2008a, 546). Pragmatically, definitions work by opening up some inferential possibilities and closing off others. For example, defining abortion as murder allows the speaker to transport judgments of severity, blame, and legitimate punishment without having to justify their application. Speaking more broadly, by selecting words, people can select premises that generate specific inferences and conclusions (Ilie 2009, 39), without explicit justification, argument, or explanation. By contrast, unloaded words suppress moral, legal, logical and ethical implications of practices by diverting attention, creating distance between how a target might understand and reason about a different practice, and the language of discourse about that practice.Ecological functions of evocative patterns of communicative influenceDefinitions, whether implicit or explicit, affect discursive ecosystems by imposing conditions and boundaries on discourse. Similar to the modifying effect of frames, widely 17732 In early work on persuasive definition, attaching old values to new terms was a result of leaving the emotional connection with the original term intact (Stevenson in Walton 2005). accepted defini