1 Introduction

The interpersonal exchange of reasons and critical considerations, or argumentation, is widely thought to enhance deliberative interactions that generate and justify reasonable policy decisions. In deliberative democracies, citizens in their various roles—as voters, opinion leaders, policymakers, activists, lobbyists, journalists, consumers, stakeholders, etc.—are indeed expected to debate proposed or enacted policies and to exchange views on them. Contributions to such discursive exchanges invariably fall under a range of norms, some of which apply to any communicative context (e.g., norms of intelligibility). Others apply specifically to argumentative contexts (e.g., norms of cogent reasoning). And yet others are specific to contexts of public policy argumentation (e.g., norms that promote citizen participation).

Our main goal is to outline the state-of-the-art on norms that should govern public argumentation. We focus on norms associated with, on the one hand, the ideal of correctness—a term we use as a placeholder for a range of specific ideals (e.g., moral or social optimality, permissibility, truth, validity)—and, on the other hand, norms associated with the ideal of participation, a term we use to denote such ideals as inclusion, diversity, and equality. Both ideals are understood as independent, yet related sources of political legitimacy.

The ideal of correctness requires conformity to certain standards. Standards of a substantive kind concern external yardsticks such as the (lack of) acceptability of a policy standpoint or the (lack of) optimality of a proposed policy. Relevant critical questions thus include whether the information used is reliable and whether inferences drawn are valid or sufficiently probable. By contrast, standards of a procedural kind address the processes and procedures that lead to the (lack of) acceptability of a policy, given a communicative process among citizens and other stakeholders. Relevant critical questions thus include whether all concerned parties have a say, whether motivating considerations are well explained, and whether objections are integrated into the deliberative outcome.

Substantive and procedural accounts of argumentative correctness can be developed in different ways, lending a distinct “flavour,” even a distinct meaning, to the concept of correctness. Specific instances of argumentative exchange may thus be associated with different evaluative criteria (e.g., truth, legitimacy, acceptability, etc.). The concept of correctness nevertheless has the relatively stable function of indicating an ultimate evaluative stance driving the back and forth of argumentative exchanges. ‘Correctness’ thus provides a placeholder for terms that reflect a variety of evaluative criteria.

Comprising both political and epistemic aspects, the participation ideal demands the inclusion in an argumentative exchange of anyone affected by, or in part responsible for, a policy, in ways that encourage them as deliberating participants to express opinions, preferences, and concerns. In a public policy deliberation context, the participation ideal thus tasks participants as individuals with inquiring into a range of competing perspectives on a policy issue, while participants-as-collectives are tasked with developing a reasoned standpoint on the policy issue, specifically a standpoint that commands the assent of a relevant majority. Like the correctness ideal, the participation ideal can be specified in diverse ways. But given that participation is a means toward a correct deliberative outcome, this ideal plays a subordinate role in argumentative practices.

The correctness and participation ideals are widely thought to be mutually coherent. In a well-designed democratic system, “deliberative rationality” and “democratic legitimacy” align and mutually reinforce one another (Benhabib 1994). If participation in a policy deliberation is instrumental, or even intrinsic to a correct deliberative outcome, and if information that is necessary for a correct outcome is distributed among agents, then participants must have an opportunity to adapt or improve their opinions and preferences in response to learning about those of their political peers. This extends to preferences on the rules and procedures for deliberation and argumentation, which participants must be able to co-determine. Particularly if a policy decision is viewed as good simply because it was agreed upon, participation becomes a requisite for good policy.

While maximising participation is therefore conducive to approximating a correct deliberative outcome, social deliberation has its pitfalls. An increase in participation may bring about an increase in distortion due to social biases that silence dissent or create polarisation. Although the ideals of correctness and participation can cohere, they can also be in tension and sometimes clash. Under less-than-ideal circumstances, the epistemically valuable “freedom of inquiry” clashes particularly with the basic democratic “freedom of participation” (Jacobs 2003). Disagreements, after all, may be so numerous or deep that maximising participation makes it difficult, perhaps impossible, to arrive at a deliberative outcome that gains the assent of a relevant majority (Fogelin 1985).

For instance, if participants lack access to necessary information or cognitive capacities, if they reject expert opinions, act overly self-serving, or fall prey to herd mentality, then maximising participation is likely to result in a suboptimal deliberative outcome. Maximising participation may thus impair correctness. Forms of participation that are broadly non-conducive to correctness, therefore, must be limited, redirected, or curtailed. This, indeed, can be described as “the classic puzzle of deliberation”, that is, “a trade-off between the quantity (of participants allowed to express opinions, arguments, and criticisms) and quality (of arguments advanced and decisions reached in the process of deliberation)” (Aakhus & Lewiński 2011, p. 170).

When aspiring to a correct policy decision, then, what counts as genuine participation and how ought one to participate? Specifically, with the ideals of correctness and participation regularly in mutual tension, how can a balance between the associated norms be struck? Of course, we cannot hope to provide a final answer, but rather present the pertinent complexities from the viewpoint of argumentation studies. We begin by elaborating on the concept of a norm of public argumentation (Sect. 2), turn to salient views on the ideals of correctness and participation (Sects. 3), and finally survey linguistic, logical, dialectical, rhetorical, and epistemic norms for public argumentation (Sects. 4, 5). Before concluding, we stress that the norms of public argumentation must “materialise” in the legal and political institutions that are tasked with policy-making (Sect. 6).

2 Norms of public argumentation

The communicative practice of exchanging reasons and critical considerations, or argumentation, can be considered public, insofar as it is “about a public issue (content clause) and […] is typically expressed by people in their public capacity, e.g., as citizens or politicians (speaker clause), in an open forum (context clause), while addressing a larger audience whose members are people in their public capacity (addressee clause)” (Zenker et al. 2020, p. 4).

As a form of deliberation, public argumentation relies on practical reasoning to justify opinions and decisions on a public policy as correct—in the sense of being evaluated as legitimate, optimal, reasonable, etc.—or to invalidate other options as incorrect. While the purpose of public argumentation as a communicative practice is to enhance the quality or the value of a deliberative outcome (Habermas 2003), it also serves a wide range of more specific purposes. It enables participants to inform or persuade each other, place issues on the agenda, integrate viewpoints (among political peers or academic experts), transform private motives into public considerations, express subjective attitudes about a public decision, and create new policy options (Zenker et al. 2020).

Norms of public argumentation thus provide individual or collective agents with prescriptions regarding their speech acts, the discussion procedures they follow, or the kinds of arguments and critical reactions they advance. Specific evaluative criteria can be understood as underlying these prescriptions, respectively as being explainable by them. Although the norms are general, applying a given norm in a context typically requires some amount of specification and disambiguation. Even if some norm is uncontested, a policy deliberation can therefore pivot not only on its acceptability but also on how it should be interpreted and applied. (See Cooke 2002 on the transformative role of argumentation.) For example, participants who charge each other with lying, “bullshitting,” or spreading fake news appeal to the norm of truthfulness or honesty. But they apply this norm to different objects.

The wide variety of norms, and how they relate to the ideals of participation and correctness, are usefully discussed along two dimensions. First, by specifying the complexity of the object or aspect that a norm targets (current section). Second, by characterising norms according to their specificity, e.g., whether a given norm addresses public argumentation, argumentation more broadly, or communication in general (Sect. 4).

Norms targeting contributions to public argumentation can be distinguished at increasingly complex units of action, i.e., at the micro, meso, and macro levels. A speech act generally belongs to the micro level. Linguistic-communicative norms at this level secure mutual intelligibility, understanding, and communicative effectiveness, thus providing normative conditions for any meaningful communicative practice. Key examples include the maxims of conversation (Grice 1975; 1989), the felicity conditions of a speech act (Austin 1962; Searle 1969), or norms for assertion (Goldberg 2015; Williamson 1996).

The meso level includes argumentation as a complex speech act, comprising the assertion of a claim (or thesis) and one or more reasons supporting it. Examples of norms at this level are that good arguments fulfil a characteristic function, e.g., to convince an addressee of the thesis (Perelman & Olbrechts-Tyteca 1969; Tindale 1999; 2004), to reach a consensus (Habermas 2003) or resolution (van Eemeren & Grootendorst 1984; 2004), or to guide the addressee or audience addressed to recognize the rational justification for a thesis (Goldman 2003; Biro & Siegel 2006; Lumer 2005; Siegel & Biro 1997). Such norms emphasise not only consistency, logical validity, epistemic justifiability, dialectical acceptability, or rhetorical persuasiveness, but also attitudes such as open-mindedness or critical engagement, and the transparency of an argumentative strategy.

The macro level is made up of discussions and debates between individuals and institutional agents. Whether norms for evaluating policy decisions rest on moral, prudential, or utilitarian considerations, they are invariably part of the shaping and reshaping of society and are thus located at the macro level. Also at this level are norms with a clear procedural component, for instance norms regulating how a discussion aims at reaching consensus, norms concerning the cooperative “ironing out” of mistakes and the search for truth (Lumer 1988; Walton & Krabbe 1995; van Eemeren & Grootendorst 1984; 2004), as well as norms addressing political requirements (e.g., transparency and accountability), legal requirements (e.g., the right to speak freely or rights reserved to parliamentarians), or moral requirements (e.g., representation of the interests of all concerned or respect for discussion partners). The macro level also includes norms that target more complex, often institutionalised features of argumentation, such as legal procedures, guidelines for good journalism, educational curricula, or conventions regulating a political culture. Examples are norms that require clarity of legal language, fair journalistic reporting, recognition of the need for critical thinking, or the absence of hegemonic political speech.

The distinction between micro, meso, and macro levels helps to understand how the norms of public argumentation interact. For instance, some scholars of deliberative democracy acknowledge certain “uncivil” speech acts as justifiable at a lower level for purposes of countering hegemonic speech at the macro level (Chambers 2017, p. 268; Mansbridge 2012). Norms of public argumentation may thus compete on both the horizontal dimension (correctness vs. participation) and the vertical dimension (micro, meso, macro). We refrain from assumptions about the degree to which such norms are in fact operative in public discussion, or regarding the benign or harmful effects of increasing awareness of them (Aikin 2020).

On this background, we turn to the ideals of correctness and participation.

3 The ideals of correctness and participation

3.1 Overview

We start by discussing the correctness of a public policy decision according to both substantive and procedural criteria (Sect. 3.2) and address whether ‘participating in public argumentation’ means merely to express feelings, preferences, or opinions, or rather to open-mindedly inquire and deliberate how public policies serve the common good (Sect. 3.3). To exemplify the combination of a substantive view of correctness and an expressivist view on participation, we turn to Condorcet’s jury theorem and his voting paradox as discussed in decision theory and judgement aggregation (Sect. 3.3.1). Finally, we address the combination of a procedural view on correctness and a deliberative view on participation by turning to the virtues and vices of deliberation as discussed in deliberation theory (Sect. 3.3.2).

3.2 The correctness ideal

Because a standpoint on a policy option can in different ways be understood as correct or incorrect, these terms can be replaced by terms denoting the motivating grounds for participants to argue for or against a policy. To support the adoption of a proposed policy option or the rejection of a rival option, for instance, participants may evaluate it as acceptable, fair, legitimate, optimal, sustainable, democratic, reasonable, etc. Even if the exact sense of ‘(in-)correctness’ is not well-specified, some such regulative ideal is presupposed whenever participants engage in the reason-based back-and-forth of argumentation.

This holds in both the informal and the formal public spheres. In the informal public sphere, individual citizens are meant to debate whether policies are worthy of their approbation by expressing their ideas and preferences, understanding those of others, and considering how to profit from modifications or refinements (Habermas 1996). In the formal public sphere, by contrast, elected (or appointed) representatives, and in some cases, the people themselves decide on policy options (e.g., through referenda). What must here be additionally accounted for are such institutional constraints as legal admissibility, or parliamentary or public support.

The term ‘correctness’ can be elaborated in substantive or proceduralist terms. According to a substantive stance on evaluating a deliberative outcome, a public policy decision is good to the extent that it satisfies evaluative criteria that are independent of whether any particular procedure has been followed. These criteria may refer to values such as welfare, truth, the public will, or the sheer preferences of (a majority) of citizens. Public argumentation in the informal public sphere would thus serve the primary function of allowing citizens to express their considerations and judgments, and to refine and transform them in view of their peers’ to arrive at a substantially correct outcome. Subsequently, as actors in the formal public sphere engage in policy-making, citizens’ considered judgments could complement the results of opinion polls or voting procedures.

This substantive stance aligns equally well with emphasising the political importance for the formal public sphere of expressing preferences via public argumentation in the informal public sphere. After all, a public policy decision may well be considered (likely) correct if it reflects the expressed preferences of the majority or serves the common good. For a proceduralist stance on evaluating deliberative outcomes, by contrast, what matters for the correctness of a policy decision is the procedure by which participants arrive at it, including its argumentative aspects. A major reason for emphasising the procedures of public argumentation, of course, is the typical absence of non-controversial criteria to evaluate a policy in substantive rather than procedural ways. Moreover, it is prima facie only plausible that a well-designed deliberation procedure likely results in an approximately correct outcome, for example, by providing participants with tools that limit undue pressure from advocacy groups, business, or state powers, or by ensuring input from both citizens and experts.

In the formal and informal public spheres alike, a moderate proceduralist stance treats a sound public argumentation procedure as an important but possibly dispensable indicator of the correctness of a deliberative outcome. Whereas a more radical proceduralist stance treats this as a necessary condition. In this consensualist view, the correctness of a deliberative outcome resulting from an impeccable deliberative process is partly or entirely constituted by participants agreeing on it (Barry 1989, pp. 268, 292; Habermas 2003; pp. 257–58; Lafont 2012, p. 288; Scanlon 1982, pp. 103–28; 1998, pp. 1–5).

A substantive view instead emphasises the importance of evaluative criteria that are independent of the deliberative procedures for public argumentation. While procedures can provide reliable routes to reaching correct deliberative outcomes, they are not guaranteed to do so, and may even lead participants astray. And even if procedures do provide such reliable routes, this may be owed to substantive criteria being applied implicitly, so that procedural correctness derives from substantial correctness. Relevant substantive criteria, although themselves controversial, are found in welfare ethics, for example, based on insights into how the general good can be aggregated from individual benefits, specifically how opposing interests can be balanced fairly (Brandt 1979; Hooker 2000; Parfit 1997; Smart 1973; Temkin 1993). But since the application of substantive criteria typically requires expert knowledge, and given the risks of an expert-based technocracy for the common good, this merely underscores the need for participation, i.e., for citizens being included in contexts of public policy argumentation.

3.3 The participation ideal

Although the participation of citizens remains important for public argumentation, different emphases can be placed not only on why it is important but also on the forms it should take. While participation plausibly has virtues besides contributing to the correctness of a deliberative outcome, the correctness ideal largely explains why participation is considered valuable for public argumentation. Consequently, at least two central questions arise: First, who should participate (e.g., stakeholders, experts, representatives, citizens, institutions, artificial agents), and at which stage (e.g., in agenda setting, in drafting, ratifying, or implementing policy proposals, or in monitoring, evaluating and revising policies)? Second, what counts as genuine participation, and how ought one to participate? Our focus will be on the latter, conceptual question.

3.3.1 Expressing opinions and voting

A natural, if minimal way of participating in public argumentation is to express one’s preferences or opinions. An important class of methods for the use of preferences or opinions involves their aggregation (e.g., via opinion polling or voting). As to the level of detail at which preferences or opinions are expressed, however, and regarding deliberative activities beyond mere expression, Condorcet’s (1976) jury theorem and his voting paradox show that contributing to the correctness of a deliberative outcome requires more than a minimal form of participation. While the literature on decision theory, judgement aggregation, and social deliberation is currently dominated by an expressivist and proceduralist view, it is worth examining whether the long-recognized problems for judgement aggregation are more fruitfully addressed from a deliberative and substantive view on social deliberation.

Within large democratic bodies, Condorcet (1976) argued, voters (or jurors) may or may not possess the decision-making competence required for a majority decision to be correct more often than not. If the minimum amount of competence for an individual agent is identified with a probability of at least 50% of making a correct decision, and if more than 50% of voters command this minimal competence, then Condorcet’s jury theorem shows that the majority verdict is likely to be correct. Because this probability increases rapidly with increasing group size, the theorem may seem to provide a true boost for mass democracy. As additional conditions for the jury theorem, individuals are assumed to cast an all-or-nothing vote on a binary issue and to reach their voting decision independently.

Like earlier authors, however, among whom is Plato (Steinberg 2010), Condorcet observed that voters often lack this minimal decision competence. In those cases, Condorcet showed, the jury theorem reverses: “the more numerous the assembly, the more it will be exposed to the risk of making false decisions” (Condorcet 1976, p. 49). Worse yet, his voting paradox showed that, if voters are free to order their preferences between more than two options, then preference cycles may block a unique correct collective outcome.

To explain, suppose that an electorate tasked with deciding on one of three policy options (A, B, C) is divided into three equally sized sub-groups with distinct preferences, or what Ingham (2019) calls a case of “multiple majorities.” One-third of voters prefer A to B and B to C, the other third prefer B to C and C to A, and the remaining third prefer C to A and A to B. Pairwise majority voting would yield the outcome that a two-thirds majority prefers A to B (the first and third group), while a two-thirds majority prefers B to C (the first and second group). Coherence requires that A be collectively preferred to C. Yet a two-thirds majority does at the same time prefer C to A (the second and third group). Maximizing participation would in this case enhance the electorate’s diversity in terms of preferences but also increase the risk that the collective decision lacks coherence.

If social deliberation generally and particularly public policy argumentation shall mitigate the risk that a collective decision lacks coherence, then voters’ competence must (somehow) be enhanced, as this increases the likelihood of a correct aggregate outcome. Social deliberation may promote a specific structure in voters’ preferences that ensures the coherence of pairwise majority voting, namely if individuals’ preference profiles are influenced such that they evaluate policy preferences by initially reaching a meta-level agreement on a single evaluative criterion. For instance, if deliberation results in an agreement that what really is at stake are the costs or the candidates’ leadership skills, then preference cycles cannot surface even if different opinions about costs, or skills, persist. The aggregation of preferences can thus produce a single outcome.

If deliberation makes a suitable evaluative criterion available, individuals can in principle be rationally convinced to drop preference orderings that are incoherent relative to this criterion. The resulting set of individual preference orderings is “single-peaked” (Gaertner 2002), i.e., they can be ordered objectively on a given spectrum of preferences. As empirical results from deliberative polls (List et al. 2013) indicate, moreover, if a voting system is not gerrymandered or otherwise contrived to yield “low-fidelity” feedback, then deliberation can lead groups to approximate single peakedness. Although this increases the chances that majority voting generates a coherent deliberative outcome, voters need not be aware of the underlying common ordering. Instances of single-peakedness therefore need not be interpreted as indicating a meta-agreement (Roy & Rad 2021).

A principled way of reconciling the participation and the correctness ideals thus is to first create more cohesive preferences. This reconciliation may also result in the applicable aggregation rules becoming a matter of social deliberation (Miller 2007). Empirically, this is well described in accounts of how decisions were reached in the ancient Roman Senate, for instance, how monasteries and convents chose their leaders during the Middle Ages, or how democracy developed in France and America in the Enlightenment period (McLean & Urken 1995). In any case, a reconciliation of the participation and correctness ideals requires a high-quality deliberative setting that maximises the availability of reliable information.

Conversely, inserting disinformation would create a setting that likely leads to less cohesive preferences. A salient example is the recent partisan effort, in the USA as elsewhere, to cast doubt on the acceptability of public health measures by denying or downplaying the effectiveness of face masks in reducing viral spread. As some agents even appear to “double down” if scientific evidence conflicts with their personal beliefs (Nyhan & Reifler 2010), this only underscores the need for a well-designed, norm-governed deliberative setting.

What historically enabled the systematic formal investigation of the class of aggregation methods satisfying such desirable properties as that each individual can contribute any ordering of options they prefer, was Arrow’s (1951) axiomatic framework for social choice theory (List et al. 2013). Although Arrow’s impossibility theorem has sometimes been interpreted as suggesting that democracy is self-contradictory (e.g., Riker 1982)—because the theorem shows that no method of aggregation where this and other plausible properties hold is guaranteed to result in a unique social choice, or what Arrow calls a “social welfare function”—focussing on the impossibility aspect of Arrow’s theorem risks losing sight of important possibilities for democratic decision-making. Social choice theory beyond Arrow’s axiomatic framework rather shows that the ideals of participation and correctness are reconcilable.

A reconciliation of these ideals pivots on applying a method known as voting by grading, where participants disclose more refined information about their preferences and how these are interconnected (Balinski & Laraki 2010). If social deliberation assists participants in developing the intensities of their preferences (Dryzek & List 2003), and if voting proceeds by grading six policy options, say, then preference cycles cannot arise even in the absence of single-peakedness, provided that the collective deliberative outcome is identified with the median grade. Under this condition, moreover, at least some of the issues arising from Condorcet’s “unenlightened” majorities can be avoided (Morreau 2020).

Social deliberation in the form of argumentation can thus help citizens to refine their preferences and opinions. Moreover, social deliberation occurring before an aggregation procedure can compensate for relaxing some conditions of aggregation procedures formulated by Arrow. If social deliberation promotes the truthful expression of preferences, this can compensate for suspending the condition that preferences be expressed truthfully, a condition keeping the aggregation procedure “strategy-proof.” And social deliberation plausibly “induces individuals to reveal their preferences and views truthfully” (Dryzek & List 2003, p. 9) for two reasons. First, since epistemically vigilant participants can test the veracity of information (Mercier & Sperber 2011) and given that communication is typically recurrent, a profitable lie today can damage one’s credibility tomorrow (but see Sect. 5.3). Second, if social deliberation encourages participants to reach meta-agreement on an evaluative criterion that ensures “single-peaked” group preferences, then this may compensate for suspending Arrow’s maxim that any other preference profile can serve as the input for aggregation.

Given a high-quality deliberative setting, then, more elaborate aggregation procedures like voting by grading can improve the alignment between the ideals of participation and correctness. From voters, this alignment demands engaging in public argumentation to test the reliability of information (including information on what is preferred) and to identify criteria that can potentially settle a policy dispute. This in turn requires that preferences are based neither on epistemic or social deference, nor on mere herd behaviour, but on independent fact-checking, and the capacity to reason critically and switch perspectives.

3.3.2 Participation by deliberation

Beyond contexts where social deliberation precedes the aggregation of preferences and opinions, the positive or negative effects of social deliberation as a sui generis form of participation merit separate treatment. Social deliberation may be required, for instance, if questions of public policy cannot be settled by voting, or if formulating the policy options to be voted upon itself requires deliberation. In such contexts, social deliberation typically cannot be understood as part of a procedure to aggregate individual preferences and opinions into a correct-qua-legitimate collective decision, because social deliberation is here more strongly associated with community-oriented (vs. individualistic) communicative processes and procedures that transform preferences and opinions in view of the common good. It should therefore rather be understood as an attempt at reaching or approximating a substantively correct decision (see Sect. 3.2).

First turning to the positive effects of social deliberation, an increase in participation in public argumentation cannot only go along with an increase in the necessary information for a well-considered, correct policy decision. It can also enhance the quality of the argumentation by strengthening citizens’ control over the deliberative processes and procedures, to thus “weed out” errors or misleading framings and phrasings. Just like a legally binding verdict should obey the classical dictum ‘you must listen to the other side’ (audiatur et altera pars), a deliberative outcome of an inquiry, for instance, is rational only if all relevant evidence and perspectives are taken fully into account. Given time and resources, public argumentation can thus serve to obtain more complete, veridical information from reliable sources before participants form a final opinion. (See the principle of total evidence (Carnap 1947) and peer agreement as higher-order evidence (Kelly 2010; Smart 2018; Surowiecki 2005).)

If increased participation leads to increased diversity of inputs that improve the quality of information as a deliberative process unfolds, public argumentation epistemically enhances the opinions that participants form (Chambers 2017). A key mechanism is the mutual criticism of preferences and opinions, as well as their supporting reasons, including their presuppositions and specific formulations. If this dialectic interaction functions well, errors, flaws, and fallacies can be diagnosed and corrected. The “surviving” preferences and opinions, therefore, can withstand a serious level of critical testing.

However, participants in public argumentation who take the deliberative process to be impaired or distorted are likely to be sceptical also about the quality of the deliberative outcome. These participants are presumably less likely to display tolerance towards any remaining disagreement or to trust whichever outcome results from an impaired process. Not only are such impairments diverse, but they also occur at various levels. Examples include unequal access of political parties to monetary resources during an election campaign, the participation of “unwanted” actors (e.g., foreign secret service members or artificial agents), algorithms that determine levels of audience attention to messages based on commercial interests, or political myths that appeal to gut feelings (e.g., racist or nationalistic tropes) that potentially block critical thinking.

But norms for public argumentation that allow such impairments to be avoided are broadly lacking today. And even if these norms were available, disagreement may persist about their relevance or about how they should apply to a specific case. Some scholars hence contend that a deliberative outcome merits the qualification legitimate only if participants have sufficient control over the specifics of the deliberative procedure (Habermas 2003; Lafont 2012). In the formal and informal public spheres alike, this requires a high degree of collective autonomy as to who may participate in social deliberation, what deliberative procedures are used, and how to evaluate deliberative contributions and outcomes (Cooke 1997).

The need for collective autonomy entails acknowledging that norms for public argumentation cannot simply be taken for granted. If participants find that an applicable norm must be improved, a well-designed process of public argumentation must (somehow) allow for this by providing room for meta-argumentation on norms and procedures. Participants can generally shape this dynamic in diverse ways. The formal public sphere, for instance, must even leave room to improve constitutional rights such as freedom of expression (assuming procedural obstacles are overcome). In the informal public sphere, by contrast, the need for collective autonomy can be addressed not only through sustained efforts at civic education that enhance the quality of deliberative inputs (Guttmann & Thompson 2004) but already through more mundane measures such as a “deliberation day” that attracts citizens to engage in public policy argumentation (Ackerman & Fishkin 2004).

The need to control the procedures of public argumentation is most obvious given the potential adverse effects of leaving these procedures unconstrained. On the assumption that specific discourse contributions ought to be democratically inadmissible (e.g., racist, sexist, or otherwise derogatory speech), an unconstrained procedure is likely to see some participants behave in ways that are undemocratic or democratic in appearance only. Moreover, social or cognitive failures may lead participants to defer uncritically to majority opinions or to let fear of reputation damage hamper open disagreement (Sunstein & Hastie 2008). And if participants enjoy not only different levels of skill as arguers but also different levels of being acknowledged as reasonable deliberators (Sanders 1997), then presumptions favouring the status quo may for the wrong reasons be reinforced. Securing high deliberative quality thus requires that participants reflect critically on procedural issues and also engage with the design of the deliberative process of public argumentation (Aakhus & Jackson 2005; Aakhus & Lewiński 2011; Jackson 2015; Lewiński 2017).

Probably, the most important epistemic consideration against maximising participation is that if too many participants know too little about the true quality of a policy proposal—itself an instance of Condorcet’s “unenlightened majority”—or use strongly biased linguistic resources (e.g., towards the status quo), they are more likely to be misled by demagoguery or ideology. Whether increased participation improves the deliberative process or its outcome thus depends once again also on the quality of the deliberative input.

Against this background, we now survey norms of argumentation that bear on participation, the quality of the argumentative process, and the correctness of a deliberative outcome.

4 Norms of argumentation

4.1 Overview

This section surveys norms for communication in general and public argumentation in particular. We begin by discussing how linguistic norms affect the argumentation process, then turn to norms as developed in the three classical approaches to argument: logic (including epistemology), rhetoric, and dialectic. (For historical background on these approaches, their conceptual frameworks, and applications, see Wagemans 2021). The subsequent section (Sect. 5) addresses legal and political norms as examples of practice-dependent norms for argumentation in public deliberation.

4.2 Linguistic norms

Public argumentation would be impossible were it not for linguistic norms that allow speakers to express and mutually understand their thoughts and intentions. Competent users of a natural language can be thought to share syntactic and semantic rules determining what a well-formed, meaning-imbued utterance is (Chomsky 1965; Katz & Fodor 1964), as well as pragmatic rules enabling them to coordinate on non-literal meanings given how an utterance functions in context (Alston 2000; Austin 1962; Green 2009; Grice 1989; Sbisà, 2019; Searle 1969; Williamson 2000). As rules that provide evaluative criteria for all communicative practices, these linguistic norms are directly relevant to how the ideals of correctness and participation inform and shape communicative practices. This is presumably best appreciated when focusing on the functions that a norm serves and the penalties that are deemed appropriate if a norm is violated.

A constitutive rule provides the concepts required to understand utterances as meaningful contributions to communicative practice. This holds independently of whether a communicative practice is well-suited to produce a correct deliberative outcome or to invite participation. A constitutive rule thus makes a communicative practice possible by defining whether an utterance counts as the performance of a specific illocutionary act, respectively whether a rule violation voids it. The different argumentative burdens that speakers incur as obligations, therefore, can be grounded in the structure of different kinds of illocutionary acts (Kauffeld 1998). A constitutive rule for asserting, for instance, is that the propositional content asserted by a speaker (S) is new information to a hearer (H). As a result, if S utters “I am unconvinced of the merits of this policy” in contexts where S knows that H knows as much, then S’s utterance is more likely to request that H justifies this policy to S.

Different pragmatic frameworks share the idea that reasonable or appropriate conversational contributions are normatively regulated. For instance, Grice’s maxim of quality (Grice 1989), Searle’s sincerity condition for speech acts (Searle 1969), and Williamson’s knowledge rule (Williamson 2000) all roughly convey that a speaker ought to speak truthfully. By contrast, to “quietly and unostentatiously violate a maxim” (Grice 1975, p. 49) normally damages a communicative process, e.g., by potentially deceiving the addressee with false information. If detected, such violations normally reduce the speaker’s credibility (Green 2009), functioning as a kind of sanction. Similarly, truthfulness as an objective requirement defines (without recourse to participants’ subjective perspectives) whether a speech act token is a correct instance of its type (Sbisà, 2019). An objective requirement being violated can result in a negative assessment of the speaker’s utterance as being the wrong thing to say.

At some point in a public policy context, a proposed policy and the reasons for and against it are typically asserted (Goldberg 2015). Norms of assertion target the properties of a declarative sentence that rationally warrant its assertion. Prominent norms of assertion dominating the philosophy of language literature cite doxastic or epistemic criteria, e.g., belief: assert that p only if you believe that p (Bach & Harnish 1979); justified belief: assert that p only if you justifiably believe that p (Lackey 2007); truth: assert that p only if p is true (Weiner 2005); knowledge: assert that p only if you know that p (DeRose 1996; Williamson 1996). Norms of assertion thus bear on the epistemic setting where the premises and conclusion of an argument are expressed. Among the growing empirical research on norms of assertion, some results are consistent with the knowledge account (e.g., Turri 2015), others with the justified belief account (e.g., Kneer 2018).

What the philosophy of language has seemingly left unexplored is the connection between assertion and argument. By asserting, speakers undertake commitments to the truth or justifiability of an asserted content and so incur an obligation to offer reasons when challenged (Camp 2016; Fricker 2012). Because a seamless argumentative exchange requires that speaker meanings are identified correctly, this makes norms of commitment attribution central to analysing and evaluating public argumentation. For instance, the conversational score-keeping model (Brandom 1983; Lewis 1979) treats a speaker commitment as a form of obligation determining what a speaker is directly accountable for (MacFarlane 2011; Walton & Krabbe 1995). Beyond what is uttered overtly, commitments extend to “what the speaker can be said to have taken for granted in making his or her utterance” (Katriel & Dascal 1989, p. 286), including commitments to implicit contents, illocutionary forces, and to having performed the speech act itself (De Brabanter & Dendale 2008).

Commitment attribution is of special methodological relevance because reconstruction and evaluation of argumentation regularly require the identification of implicit premises or conclusions (Hinton 2021; Oswald 2016; van Eemeren & Grootendorst 1992). Even if linguistic norms are nonspecific to argumentative exchanges, they bear critically on how public argumentation as a form of social deliberation unfolds. For in what is generally to the detriment of deliberative practice, speakers can covertly twist a linguistic norm by manoeuvring within the space of plausible deniability, and thus reduce the credibility or inclusion of another party, avoid conversational liability, or simply mislead.

How the voices of underrepresented groups are potentially silenced or oppressed in unfair ways is the focus of contemporary work in speech act theory. In the conversational score-keeping model (Langton 2018; Lewis 1979; Witek 2015), appropriateness rules define the scope of permissible conversational moves, and kinematics rules define how the performance of a speech act affects the state of the conversation. This provides a model to trace how speakers can knowingly exploit how hearers accommodate by adapting their mental states to interpret utterances as aligning with the appropriateness rules. Especially problematic forms of exploitation occur if speakers transmit presuppositions that potentially influence public opinion without incurring accountability for them (Langton 2018; McGowan 2018; Saul 2018). An extreme example is a claim by former U.S. President Donald Trump that he had won the 2020 presidential election on the presupposition that the African American vote does not count (Snyder 2021).

Other forms of exploitation include “dog whistles” (i.e., audience-specific code words that convey discriminatory or ideologically loaded content), backdoor speech acts (Langton 2018), or acts of plain hate speech. If participants in public argumentation fail to expose and challenge these forms of exploitation, current appropriateness rules for acceptable speech are likely to be modified. Failure to challenge the use of a slur in public discourse, for instance, signals its acceptability in that context (McGowan 2018). This challenges the supposedly active norm of equality (Mendelberg 2001), according to which discussants should enjoy an equal level of consideration and credibility.

Reference to speech act norms also allows for the detection of more nuanced ways of discursive exclusion. For instance, speakers from underprivileged social groups (e.g., in terms of their ethnicity, gender, or education) may be invited to join a debate, with no overt exclusion taking place. If the group’s demands are treated as mere suggestions, however, then their speech acts’ illocutionary force can nevertheless be blocked or downgraded (Hornsby & Langton 1998; Kukla 2014). The participation ideal is thus upheld merely in appearance. Proper forms of participation, therefore, require critical monitoring of speech act norms.

4.3 Logical and epistemological norms

According to the Aristotelian idea of logic as a tool that allows humans to come to know, logical investigations are inspired by natural language argumentation. In formal symbolic logic, instances of natural language argumentation are treated as propositional or sentential structures, the premises of which are claimed to establish, or support, a conclusion in virtue of the argument’s logical structure (logical form). Since formal symbolic logic deals with how information is processed rather than what information is processed, logical norms generally concern the validity of the inference relations between bearers of information. These relations are what a formal symbolic logic defines, in the precise terms of an artificial language with a well-defined syntax and semantics (Tarski 1983), as formal structures.

Despite a plethora of logical systems having been developed to serve specific purposes, the top-down perspective of logic entails that formal structures break with the polysemy and ambiguity of natural language. Given that “logics are in the mind” (Stenning & van Lambalgen 2008, p. 41), the meta-logical study of ecologically valid normative reasoning standards for a context pitches algorithmic norms concerning the application of a normative standard of inference against epistemic-regulative norms concerning the choice of the applicable standard itself (Ricco & Overton 2011). The most prominent relation of inference, of course, is that of logical entailment in classical logic.

Classical logic is two-valued (true/false), extensional (the truth value of a compound sentence is derivable solely from those of its compounds), and monotonic (validly derived conclusions are stable under premise addition). Thus, a set of sentences or propositions A logically entails a sentence or proposition B if, and only if, provided each element of A has the semantic characteristic of being true, then B is true too. But the variety of information processing methods entails that formal logic has many faces. Even deduction need not be classical. Many-valued logics, for instance, go beyond truth and falsehood by introducing three or more truth values. Intensional logics account for the phenomena of argumentation and reasoning using more refined concepts that cannot be dealt with only in terms of extensions (e.g., Kripke’s possible world semantics). And non-monotonic logics aim at modelling defeasible reasoning, where new information may undermine or falsify previously held beliefs (e.g., default logic). Finally, other formal tools account for various ampliative processing methods inspired by real-life information processing, such as abduction, induction, or reasoning by analogy. This makes a given logical system one among other yardsticks for reasoning, problem-solving, argumentation, or judgement and decision-making (Evans 2012).

While natural language cannot easily be translated into an artificial language, and although semantic relations rarely offer a sufficient reason for an argument’s acceptability (and syntactic relations even more rarely do so), the logical structure continues to play an important role in evaluating natural language argumentation. A prominent example is the account of logical validity in dialogue logic, where the rules for a dialogue game let participants systematically examine whether the proponent’s thesis (conclusion) follows logically from the opponent’s concessions (premises) (Krabbe 2006). For a specific dialogue game, an argument’s validity is defined as the existence of a winning strategy for a proponent defending the conclusion against an opponent conceding the premises. The proven equivalence between this dialogue-logical concept of validity and “classical” logical concepts of validity (e.g., in terms of truth preservation) shows that a procedural approach to logic can in principle recover the evaluative result at which a non-dialogical, or substantive, logical account would arrive.

As another example, the logic of questions variously offers an interrogative model of inquiry (Hintikka 1999), an inquisitive semantics (Ciardelli & Roelofsen 2011; Groenendijk & Roelofsen 2009), or an inferential erotetic logic (Wiśniewski 1995) that formally accounts for the guiding role of questions in information processing. With the concept of logical validity here supplemented by a component of goal-directedness (e.g., solving a problem, settling an issue, or deciding on a course of action), the combination of the logic of questions and dialogue logic yields models of linguistic interaction that focus on cooperation between arguing parties (Łupkowski 2016). For instance, scenarios for answering a question offer structured models of information search aimed at identifying gaps or requesting more information (Wiśniewski 2013).

This formal and top-down perspective contrasts with the informal and bottom-up perspective of informal logic. This research program primarily offers tools for the analysis and evaluation of complex arguments in real-life contexts that do not require translating natural language argumentation into a formal language (Blair 2012; Johnson 1996; Johnson 2014; Johnson & Blair 1977; Johnson & Blair 2006). Probably most influential is the distinction between norms labelled as relevance, acceptability, and sufficiency (Johnson & Blair 1977); acceptability, relevance, and good grounds (Govier 2010); or premise adequacy, premise relevance, and inferential sufficiency (Godden & Zenker 2018). More recent work on a systematic procedure to evaluate natural language arguments even makes do with only two properties: premise acceptability and the solidity of the argument lever, i.e., the inferential warrant that expresses the support relationship between premise and conclusion (Hinton & Wagemans 2022; Wagemans 2020).

Informal logic generally supplies arguers in real-life contexts with tools to critically evaluate their own and others’ persuasive attempts and, where needed, to engage in “logical self-defense.” These tools primarily serve not only to encourage but to actively support critically engaged participation aimed at approximating a correct deliberative outcome. Arguably the main tool is a collection of argument schemes describing the structure of common forms of (typically defeasible) argument. Associated with a scheme are critical questions meant to test the acceptability of a form’s instances (Walton et al. 2008).

In the context of fighting online disinformation, logical self-defence has recently been developed into a critical pedagogy approach to digital literacy, called “argument-checking” (Brave et al. 2022; Plug & Wagemans 2020). The approach builds on the argument categorisation framework of the Periodic Table of Arguments (PTA) (Wagemans 2016; 2019; 2020). The PTA provides an identification of the argument type that can subsequently be used in the Comprehensive Assessment Procedure for Natural Argument (CAPNA) (Hinton 2021; Hinton & Wagemans 2022). CAPNA evaluates various aspects of individual arguments using procedural questions.

A sound argument can generally be understood as the externalised, linguistic and interpersonal counterpart of an intrapersonal justification of belief. Conversely, an unsound (or poor, weak) argument fails to express a genuine justification. An argument thus is sound if it guides the addressee toward a rationally justified belief. This has been regarded as the standard function of argumentation in the epistemic approach to argumentation (Goldman 1999; 2003; Lumer 2005). Epistemic norms target the quality of argumentation or decision-making processes indirectly, while directly targeting the outcome, often in terms of a justified belief (Biro & Siegel 1992; Feldman 1994; 1999; Goldman 1999; 2003; Lumer 2005; Siegel & Biro 1997). Indeed, the epistemic approach recognizes procedural norms as rational only if they are instrumental in producing true and justified beliefs while avoiding false and unjustified ones. Since good arguments thus depend on general principles of knowledge that originate in logic, probability theory, or rational decision theory (cf. Goodwin 2007; see Lumer 2005), epistemic considerations are treated as more basic than dialectical or rhetorical considerations.

Specific epistemic norms, or criteria, are available for deductive (Feldman 1999; Lumer 2005), probabilistic (Lumer 2011), and practical argument forms, e.g., arguments addressing the justification of prudential value judgements (Feldman 1999; Lumer 2014). Heeding these epistemic norms generally improves a deliberating body’s truth orientation, and thus the epistemic position from which public argumentation occurs. Perhaps most relevant is the application of epistemic norms to practical arguments for judgments about policy options of the form: ‘for group G, policy option O is the best available alternative in view of the common good’ (Feldman 1999; Lumer 2014). More specific criteria for judging the sufficiency of reasons for standpoints can for instance be found in welfare ethics (see Sect. 3.2).

4.4 Dialectical norms

Natural language argument can be interpreted as a cooperative, yet mutually critical contribution to a dialogical inquiry into, or discussion about, the correct answer to a question or a problem. When arguments are studied from a dialectical viewpoint, they are analysed and evaluated as attempts at rationally persuading an interlocutor who is reluctant to commit to a proposition and advances critical considerations to be overcome by argument. Dialectical theories naturally emphasise the importance of procedural norms for reasonable dialogical engagement, where participants aspire to a correct outcome by critically engaging their standpoints, proposed answers, or solutions. A key example is the “code of conduct for reasonable discussants” (van Eemeren & Grootendorst 1984; 2004).

Hamblin (1970) had shown how to conceive of the fallacies as violations of rules for dialectical systems, a conception that the pragma-dialectical theory (van Eemeren & Grootendorst 2004) subsequently developed into a comprehensive theory of the fallacies. According to this theory, the rules governing argumentation are the rules for critical discussion, a normative model of the kind of dialogue where participants attempt to resolve their difference of opinion on the merits of the case. Fallacies, here, are seen as violations of these rules. Pragma-dialectics deals not only with the connection between the fallacies, argument reconstruction, and the contrast between reasonableness and persuasiveness, but also addresses how argumentation is embedded within institutional contexts (van Eemeren 2010).

Following Hamblin (1970), the scope of dialectic as a theory of argumentation can be extended beyond a critical discussion. Next to persuasion dialogue (a type of dialogue close to critical discussion), Walton & Krabbe (1995) characterise negotiation dialogue, deliberation dialogue, inquiry dialogue, information-seeking dialogue, and eristic dialogue (e.g., a quarrel or polemical discussion). Each dialogue type is associated with a different main goal and specific uses of reasoning. Negotiation dialogue, for instance, aims at developing an agreed-upon compromise solution based on offers and counter-offers, rather than at a resolution based on argument and counterargument (van Laar & Krabbe 2018a). Walton (1998) presented these dialogue types as normative models, similar to a critical discussion.

Walton and Krabbe (1995) discuss not only norms for mixed types of argument, where a dialogue of one type can be functionally embedded in a dialogue of another type. They also discuss how to evaluate shifts from one type of dialogue to another. A politically relevant example is the case where participants, upon ceasing to believe that a disagreement resolution is feasible, shift to a negotiation dialogue to reach a compromise solution (van Laar & Krabbe 2018b).

Dialectical theories thus provide insights into how to engage optimally in the dialogical back-and-forth, in ways that exploit the resources at hand, to thus arrive at deliberative outcomes that are deemed correct from the viewpoint of participants. As part of this, computational dialectical approaches provide different semantics for abstract argumentation systems, where arguments may attack, defeat, or undercut other arguments (Dung 1995). A semantics for such a system identifies sets of arguments that share desirable properties such as being part of a set containing a defence against attacking arguments.

Besides illustrating the logical structure of arguments, the dialectical approach recognizes that argumentation schemes (Walton et al. 2008) feature both procedural elements that show how scheme-instances function in dialogue and how they are to be evaluated by interlocutors, as well as substantive elements that focus on an argument’s structural properties. Once a scheme’s variables are saturated to yield an instance comprised of specific premise(s) and a conclusion, evaluating the dialectical quality of individual arguments typically proceeds via critical questions that target the instance’s propositional content, inference validity, or context (de Jong 2019; van Laar and Krabbe 2013). Drawn from a wide background, critical questions are always formulated generally as norms for any domain, if often on merely intuitive grounds (Yu & Zenker 2020). Insofar as reasonably convincing a critic requires that discussants answer all relevant critical questions satisfactorily, these questions provide binding norms for accepting the instance’s conclusion, for the reasons offered.

4.5 Rhetorical norms

Arguments regularly provide the means for orators to attempt to persuade audiences, including (larger) non-interactive audiences that cannot take turns with, or even interrupt, the orator. Featuring a practical (rhetorica utens) and a theoretical side (rhetorica docens), rhetorical norms thus pertain to how orators adapt to audiences in non-interactive contexts (Perelman & Olbrechts-Tyteca, 1969; Tindale 2015). In the context of public policy argumentation, a seemingly passive audience nevertheless can, and regularly does evaluate an orator’s argumentation indirectly and typically in retrospect (e.g., by signalling applause or by voting.)

The level of adaptation or effectiveness of the argumentation can be understood as an empirical notion, in which case the critic “determines whether the speaker has succeeded in establishing or increasing the audience’s adherence to their point of view” (Wagemans 2021, p. 575). It can also be understood relative to rhetorical instructions or rules of the art. In this case, the critic “determines whether the speaker has applied the rhetorical instructions in a felicitous manner to the context at hand, taking into account the genre conventions, the specific situation of disagreement, and the beliefs and stylistic preferences of the audience” (ibid.).

Rhetorical approaches stress that rhetorical norms must not be seen as prefabricated guidelines, ready for use in a wide variety of contexts. They are rather construed within the contextual specifics of a historical situation (Zarefsky 2006; Kock 2008; 2013; Villadsen 2020). Rhetorical norms are therefore highly sensitive to how situational or contextual features constrain argumentative choices. As rhetorical norms thus acquire their normative force vis-à-vis specific arguers and audiences (Tindale 2015), rhetorical scholarship is particularly well placed to shed light on how participants themselves develop the dialogical norms by which they evaluate their own and others’ discourse contributions.

What distinguishes rhetorical norms from outcome-oriented epistemic norms and procedure-oriented dialectical norms is the acknowledgement that arguers and audiences need to dynamically navigate a shared cognitive environment. This temporal dimension allows for the social constitution of rhetorical norms, as well as for their modification and possible improvement (Perelman & Olbrechts-Tyteca, 1969; Scott 2020). Some theorists even claim that the rhetorical features of an argumentative context underlie or ground their logical and dialectical features (Tindale 1999).

This insight lies at the heart of a range of naturalist approaches to rhetoric in communication studies called design theory (Aakhus 2003; Goodwin 2007; Jackson 2015; Jacobs 2006). Design theory accepts that participants themselves construe rhetorical norms entirely within the argumentative context in ways that serve their ends, including the evaluation of arguments (Goodwin 1999; 2007). As regulative or constitutive norms, these rhetorical norms must be discovered or constructed (using descriptive or analytical methods) within the rhetorical situation rather than be imposed on it. This requires a study of argument assessments in specific contexts, among specific audiences, given specific information.

4.6 Conflicting norms

As the foregoing norms presuppose different conceptions of what arguments are and how they can be evaluated, they inevitably come into conflict. No doubt, linguistic norms of communication are fundamental, because they underlie the possibility of engaging in argumentative communication. In the context of public policy discourse, therefore, the main question is whether one of the main approaches (logic including epistemology, rhetoric, and dialectic) merits priority.

Dialectical norms may seem primary because the outcome of a deliberative process acquires value to the extent that participants interact in a collaborative process of making proposals and responding, in regulated or spontaneous ways, to others’ reasoned criticism. Since dialectical norms are grounded in what it is reasonable to say in a discussion context, they may be thought to provide a middle way between the objectivism of the epistemological approach and the audience relativism of the rhetorical approach. Yet if policy decisions are to be defensible as knowledge driven or fact based, then a commitment to objectivism, or inter-subjectivism, appears to be the only way for disagreeing participants to transcend a subjective stance in epistemically qualified ways.

Priority would thus seem to lie with the epistemic approach. However, “rock-solid,” widely-accepted epistemic norms, whether fundamentalist or coherentist, are either uninformative or frequently unavailable (Hamblin 1970), thus explaining discursive impasses or deep disagreements (Fogelin 1985). And once fundamental principles of truth, knowledge, and evidence are at issue, even agreed-upon procedures to establish epistemic credentials may not suffice to reach a resolution (Hinton 2019). Particularly in the absence of a fruitful coherentist epistemology, the rhetorical approach to argumentation alone seems to acknowledge that arguers must ultimately live with contingent decisions of their own making.

As rhetorical norms also fail to provide a sufficiently robust justification for deliberative outcomes, a priority ordering over these three main approaches is sure to remain problematic. This ordering is not strictly necessary either. Different normative frameworks may suggest conflicting evaluations in the context of public deliberation. The accompanying norms, however, focus on different aspects, whence a seeming norm conflict may well be spurious. At any rate, it is doubtful that a single normative approach to argumentation merits priority.

A similar claim holds regarding the priority of procedural (or process-oriented) norms and substantive (or outcome-oriented) norms. Process-oriented norms provide practical guidance for a deliberating body that aspires to a high-quality decision. Whereas outcome-oriented norms provide evaluative criteria for argumentative contributions that inform what a high-quality outcome is. As both classes of norms may sometimes pull in different directions, scholarly debate persists over the more basic class of norms, for instance among protagonists of the pragma-dialectical theory (van Eemeren & Grootendorst 2004), who adopt a procedural stance. Whereas critics who consider this stance to invite relativism advocate an epistemically oriented objectivist approach (Biro & Siegel 1992, 2006).

Some of the basic norms of public argumentation nevertheless fruitfully overlap pragmatically. Regarding one of the key types of argument in public deliberation, the argument from expert opinion, for instance, recent studies (Lewiński 2022; Zenker & Yu 2023) show its argument scheme to be associated with critical questions that are largely co-extensive with the public norms epistemologists have proposed to evaluate expertise (e.g., Goldman 2001). Specifically, an expert should be competent in the relevant discipline, be unbiased and recognized by other experts, and have verifiable credentials and a good track record of correct judgements. Not only do these substantive norms provide readily available criteria for public deliberation to test the acceptability of expert opinions. The criteria themselves also cannot be detached from procedural norms of public argumentation guaranteeing that arguments raised by real or putative experts receive thorough critical testing (Goldman 2001; see Hamblin 1970).

We now turn to norms specific to the practice of public argumentation.

5 Practice-dependent norms

5.1 Real-life argumentation

A useful assumption in studying real-life argumentation is that arguers manoeuvre strategically between a dialectical objective of making a reasonable and non-fallacious contribution to a resolution-oriented discourse, and a rhetorical objective of making that contribution maximally effective and persuasive (van Eemeren 2010). These objectives can cohere or clash against the background of idiosyncratic conventions, points of departure, or procedures that characterise specific institutional settings like those for legal decision-making or political deliberation (van Eemeren 2010). We begin by addressing how the conduct, assessment, and perception of argumentation in institutional settings are influenced by legal and political norms.

5.2 Legal norms

Beyond considerations of the linguistic, logical, epistemic, dialectical, or rhetorical kind, discourse contributions in public policy argumentation are subject to a wide range of additional norms. This includes moral values such as dignity or respect, for instance, ecological values such as sustainability, or political values such as legitimacy. As public argumentation is exercised in specific institutional settings, some norms are generally formed and established in the interplay between argumentative ideals and practices (Habermas 1996). A particularly significant and sophisticated set of institutional argumentative practices is provided by the law (Dworkin 1986; MacCormick 2005).

Theoretical perspectives on norms of legal argumentation can be classified along two dimensions. First, the types of arguments that are regarded as valid (here meaning: acceptable) depend on whether one takes an exclusivist or inclusivist approach. Exclusive legal positivism claims that the legality of a legal decision consists in its relation to purely legal arguments, grounded in formally binding laws (Marmor 2002). Inclusive positivism, by contrast, allows for a reference to moral, political, or other practical arguments, whereas deliberative theories and natural law theories require such references (Dworkin 1986; Finnis 1979; Habermas 1996; Himma 2002). The second dimension repeats the controversy discussed above, as to whether argumentation should be assessed according to logical, epistemic, dialectical, or rhetorical norms. The current literature on legal argumentation also acknowledges a partial convergence of these different theoretical accounts (Feteris 2017; Walton 2018).

Since legal argumentation is institution-dependent, different argumentative norms may operate in different contexts, depending on the jurisdiction (in particular the civil law vs. common law legal systems), the field of law (private, criminal, constitutional law, etc.), or the particular argumentative practice (e.g., legislation, constitutional courts, standard judicial decision-making). In legislation, itself the most inclusive among institutionalised legal practices, arguments of legal, political, ecological, and other types may be mutually intertwined. Argumentation in constitutional courts, by contrast, is usually more exclusive in that respect, yet still less restrictive than standard judicial decision-making, where clear-cut distinctions between legal and extra-legal arguments are expected, and the latter arguments are admitted only exceptionally, in so-called hard cases. In the ‘culture of justification’ of modern democracies, nevertheless, a reasonable expectation prevails that legal decisions across all these practices be backed by arguments that are not only legally correct but also convincing and persuasive for a general audience (Bencze & Ng 2018).

When classifying norms for legal argumentation, one can distinguish formal norms that relate to an argument’s universality and consistency, substantive norms that relate to its content, and pragmatic norms that relate to the process of argumentation (Atienza 2020; Feteris 2017). The meta-norm of decisiveness—the ‘need for immediate, final decision’ as a result of argumentation (Ellsworth 2005, p. 697)—may be added as defining the specificity of legal argumentation.

Universality as a formal norm requires that claims be expressible in the standard form ‘whenever A, then B’, “even if the universal is acknowledged to be defeasible” (MacCormick 2005, p. 99). What the law genuinely adds to the political version of the universality norm (cf. Habermas 1990; Rawls 1971) is the public accessibility of the general legal standards in advance of a case (Fuller 1964). As formal norms, universality and accessibility thus support the objectivity and impartiality of argumentation.

Coherence as a formal norm requires that arguments “belong in a body of practical thought and commitment that is internally consistent, and characterised also by a certain overall coherence” (MacCormick 2005, p. 22; see Bertea 2005). Whereas consistency plainly demands that a body of norms display an absence of contradictions (of the logical form ‘p and not p’), coherence allows for some level of inconsistency, provided the overall number of connections between norms (somehow) compensates for it. What remains controversial is whether such a formal norm is purely formal and thus content-neutral, or whether the promotion of content-neutrality is a liberal political ideal, so that content-neutrality would at the same time be a substantive norm.

Substantive norms define the common starting points for argumentation in law (Feteris 2017) and require that arguments are based on valid standards, the recognition of which a rhetorician associates with the inventio stage (Perelman & Olbrechts-Tyteca 1969). In a given legal case, the proper selection of such standards (external justification) is often perceived as the crucial phase of legal argumentation (Aarnio 1987; Wróblewski 1974). Key examples of substantive norms that provide foundational dogmas of argumentation in modern legal orders are human rights, human dignity, and the rule of law. Canonical validity in law is generally informed by legal rules and principles, past judicial decisions, and canons of legal interpretation and reasoning (Aarnio 1987; Alexy 1989; Provenzano & Larson 2020). Insofar as these traditional standards are eventually supplemented, or replaced, by relatively novel standards, such as goals (Westerman 2010) or risks (Black 2010), the traditional shape of legal argumentation is modified.

As for pragmatic norms, of particular importance in a legal context are justificatory norms, norms for allocating the burden of proof (onus probandi) (Gordon & Walton 2009; Prakken & Sartor 2009), as well as transition norms that regulate the back-and-forth movement between general public argumentation and its legal counterpart (e.g., Alexy 1989). A convenient example is climate change litigation, where the basic question is whether and how to translate general policy goals, as well as ethical or moral duties of environmental protection (e.g., conserving biodiversity or preventing climate change), into legal obligations and rights, and what standards of proof to employ. With the landmark cases of Urgenda from The Netherlands and Leghari from Pakistan (both of 2015), one can observe a tendency of growing acceptance for such transitions by courts worldwide. A generalisation in this field is difficult, however, because exact norms regarding the admissibility of these argumentative strategies vary from one jurisdiction to another (Peel & Osofsky 2018; Voigt & Makuch 2018).

5.3 Political norms

Political norms of argumentation are normally connected with the aims and ideals of the political communities they are intended to serve. A political community run by experts, for example, would presumably be motivated by the orientation to truth, while maximising the participation of non-experts would not be an overriding norm. Indeed, an expert-ruled political community would likely consider the maximisation of participation harmful, because political wisdom is thought to be unevenly distributed among citizens with rights to political action and expression. Various critics of democratic arrangements since Plato have pointed this out, in an attempt to oppose the interest in truth to the audience-based norms of political argumentation in democracies. Even a champion of liberal society such as John Stuart Mill (1865/1977, ch. VII) proposed limiting full suffrage to citizens with a certain level of work experience or education.

Among the manifold responses to this perennial problem, one influential position states that the most important norms for deliberation in a democracy should be neutral regarding truth. Given the inherently pluralistic nature of (modern) democratic polities, what has instead been deemed appropriate is a certain abstinence regarding the truth and correctness of outcomes of political argumentation and deliberation (Rawls 1993). On this understanding, the norms of political argumentation are primarily formal and call—without appeal to the concept of truth—for resolving political disagreement by what Landemore (2017, p. 277) critically refers to as the “respectful exchange of arguments.”

Not least because of recent disinformation campaigns, however, norms of political argumentation for liberal democratic polities plausibly require substantive qualification, i.e., a qualification with relation to truth. Substantive norms are developed in epistemic proceduralism, a branch of political theory that seeks to defuse the critiques of democracy and the deflationary account of truth, while saving a substantive epistemic role for procedures in democratic institutions. In critiquing expert-ruled political regimes, epistemic proceduralists submit that even reasonable citizens regularly disagree about who counts as an expert (Estlund 1997; Goldman 2001; Lewiński 2022; Zenker & Yu 2023). This points to a central problem in justifying normative accounts of political argumentation in democracies: explaining how rule by non-experts can still result in epistemically valuable outcomes. If some people have more political wisdom than others and the views expressed in democratic procedures by many (perhaps even most) citizens fail to be correct, then how can the epistemic authority of decisions reached in such procedures be accounted for (Estlund 2008)?

The disinformation crisis in contemporary democracies suggests a different approach to this old problem. For the family of institutional arrangements referred to as liberal democracy, the pressing question regarding the norms and practices of political argumentation concerns the degree to which untruth may be tolerated. Democratic publics and their institutions seemingly can and, given freedom of expression, even must sustain some low level of disinformation while maintaining the common ground required for constructive policy argumentation. But as a result of long-term disinformation strategies that seek to make public argumentation across political differences effectively impossible (e.g., by questioning not only other agents’ authority but also the moral and political legitimacy of core procedures such as elections and their outcomes), public discourse has recently witnessed a severe fracturing. The key epistemic problem for democracy, therefore, is arguably not a lack of expertise among a politically (dis-)enfranchised population. It is rather the pollution of the cognitive environments and the breakdown of the infrastructures that provide citizens with (true) information and enforce the norms of its acceptability (Dutilh Novaes & de Ridder 2021).

This process, observable in historical contexts such as Weimar Germany as much as in several contemporary liberal democracies (Snyder 2018), yields a salient insight concerning the implicit norms of political argument. Liberal democratic institutions pivot on practice-dependent norms about what citizens are free to do. These norms preserve a core of common knowledge and shared standards of acceptability in argument and inference, a core that undergirds the fact of disagreement or value pluralism in liberal democracies. Yet these norms are what liberal democratic institutions themselves cannot legislate for, not even by constitutional measures.

As these norms are eroded through their consistent denial by salient agents and their proxies, common ground is destabilised concerning what reasonable citizens would pretend to know or defer to as knowledge (e.g., scientific institutions, statements of official record), and concerning what is generally acceptable or assertable in political argumentation. Often referred to as a state of “post-truth,” its characterising features are a lack of consensus on what is acceptable, paired with conflicting, incommensurate viewpoints and “alternative facts” supporting a variety of views (MacIntyre 2018).

Causes contributing to the disintegration of public discourse into epistemic bubbles or echo chambers include the rise of digital media and the erosion of such institutions of public record as public television and newspapers (Nguyen 2018). The cognitive environment of political discourse also includes long-term features that are especially formative regarding the norms and practices of argumentation. Examples are voting systems, political parties, and the institutions that support them. Particularly the platforms of political parties provide a method for the identification and articulation of citizens’ preferences. Political parties as institutions that recruit the political elite in liberal democracies can thus influence the salience of meta-norms that govern how existing object-level norms of political disagreement are enforced and transmitted. Such meta-norms apply where members of a group (e.g., a political party) reward or sanction each other for adhering to, or diverging from, the object-level norms of that political community.

For instance, the object-level norm to “take someone’s word” (i.e., to trust personal or expert testimony, or news venues of record) facilitates the exchange of information not only in democracies but to some degree also in autocracies. This norm can be suspended by a meta-norm that commits group members to a position that must be maintained at all costs. A recent example is for group members to claim that “the 2020 US. presidential election was stolen,” while group members inconsistently accept other results of the same electoral procedure (e.g., the election of Congressional representatives through the same ballot). Readily chastised as irrational behaviour, upholding this position allows members to (rationally) signal their adherence to specific group goals. The meta-norm of pursuing a group goal can thus override an otherwise widely accepted object-level norm.

Given that elections are the most important mechanism for the democratic transfer of power, contesting the outcome of an election is a paradigmatic case of questioning norms through meta-norms. Indeed, elections themselves can be conceived as meta-normative because they reward or punish agents based on their performance relative to the normative commitments of various groups. But electoral outcomes are pathway-dependent (i.e., they depend on the voting system), and voting systems as tools for social coordination and response can differ in crucial respects. Recent examples in the U.S. and U.K., among others, show that the rise of political discourse polarisation remarkably corresponds to first-past-the-post voting systems. Their most salient feature is that the electoral input (votes cast for a party) is asymmetrically related to the representational output (number of parliament offices won). Consequently, strategic geographic districting allows political parties to effectively select their voters, while the creation of socio-geographic echo chambers limits citizens’ exposure to messages from political opponents. This had already been identified and criticised by Dummett (1997) in connection with the notion of a “wasted vote.”

A feature of the environment that shapes norms of political argumentation in practice thus become insular tendencies of first-past-the-post voting systems and “gerrymandering” to create an environment in which parties focus more strongly on enforcing norms of their group, and less on appealing to norms across political divisions. (For a recent attempt to develop a measure of such environments, see Stephanopoulos & McGhee 2017). Though this is but one example of how seemingly unrelated public-sphere settings inform political argumentation, it is not precluded that the practice-dependent norms of argumentation may also be affected in other voting systems or institutional frameworks. In any case, both voting and political argumentation are part of the larger cognitive environment of policy argumentation. Even if voting is understood not as argument-making but as a means of expressing preferences or opinions on the public good, voting remains a constitutive part of the infrastructure for political argumentation that ideally reinforces relevant meta-norms.

Game-theoretically, one can think of voting and political argumentation as involving “team reasoning,” with different norms holding across various cultural, national, institutional, and linguistic contexts (see Lewis 1969; for problems of social coordination, see Sugden 2003; 2005). The norms of political argumentation would thus depend upon cultural-institutional features as diverse as the available topoi in a public’s collective memory, or the system of electoral rewards or sanctions for argumentative behaviour. Under conditions of extreme polarisation and widespread disinformation, political argumentation appears to be a zero-sum game in which the optimal strategy for some agents is to undermine the institutions that make policy deliberation maximally inclusive and correct. Of primary importance for the study of public argumentation and its norms, therefore, is an improved understanding of the conditions of extreme polarisation and widespread disinformation, and of how the design of the institutions that affect public argumentation can be improved.

6 Conclusion

Rather than leading to a single preferred normative framework, bringing together the broad range of scholarship on the norms of public argumentation has illustrated the expanse and complexity of relevant considerations. Many of these considerations would merit a dedicated treatment and are indeed central to fields bordering on, or partially overlapping with, the field of argumentation. If anything, scholars of argumentation are thus reminded of their own field’s interdisciplinary character.

For public deliberation in a democratic polity to be robust and its outcomes justifiable, the ideals of correctness and participation must be respected, nurtured, and developed. The correctness ideal, as we saw, can be developed in substantive or procedural ways, whereas the participation ideal may be understood narrowly as the expression of preferences and opinions, or more widely as their transformation through reasoned discourse. We saw that the norms of public argumentation can, and should, be applied at the micro-, meso-, and macro-levels of argumentative practice—from individual speech acts to institutionalized standards—and that each approach to evaluating public argumentation—the linguistic, logical, epistemic, dialectical, and the rhetorical approaches—raises distinct normative demands, demands that nevertheless partially converge.

As to the “classic puzzle of deliberation”—that is, whether the aims of improving correctness and increasing participation are mutually reinforcing or rather conflicting—limiting participation (e.g., to experts) not only leaves community members whose concerns are ignored to feel excluded. The societal resentment that is bred when expert decision-making occurs without community input is also paid in lost political authority or legitimacy. Less fully informed or equipped to make the complex decisions that experts can make, however, community members are particularly vulnerable to the misinformation that current technology spreads. For difficult, technical issues, therefore, direct democracy is often impractical and ever susceptible to hijack by parties who command a strong media influence.

The ideals of correctness and participation no doubt must be cherished to improve public deliberation. It is a further challenge to identify a normative framework and an adequate institutional design for aligning correctness and participation in a mutually advantageous way. The task of identifying that alignment is sufficiently important to not let it be dominated by partisan politics or the mass media. This task must be addressed not only for democracy to survive its current challenges, but also for argumentation studies to be relevant for the public realm.