1 Introduction

Argumentation theory seeks to understand and evaluate real-life argumentation. But the world is messy, and people argue in ways that are often difficult to classify and evaluate. If the main goal of argumentation is to resolve disagreementsFootnote 1, then many circumstances make that goal extremely difficult to realise: power imbalances, emotional attachments, unfair time constraints, cognitive biases or hidden interests are, among many others, circumstances that can heavily impede the resolution of disagreements. I call these circumstances suboptimal settings, and they can be defined as settings that are unfavourable for a reasonable exchange of reasons.

When people disagree under these circumstances, these disagreements tend to be persistent; that is, they are disagreements “that likely cannot be resolved by persuasive argumentation” (HIDDEN FOR REVIEWERS). Since the disagreements are persistent, and the conditions for a fair exchange of reasons are not satisfied, the parties will likely look for other means of overcoming their disagreements. For example, they might engage in a negotiation dialogue (van Laar and Krabbe 2018) or opt for a non-argumentative settlement of the disputeFootnote 2. But, sometimes, they need or want to stick to rational persuasion. In that case, the circumstances of the dialogue, and the fact that disagreements within such circumstances are hard to resolve, might incentivise looking for solutions that require contributions to the conversation that might clash with normative rules of resolution-oriented dialogue such as the “rules for critical discussion” (van Eemeren and Grootendorst 2004).

The main question of this paper is, then, the following. When the parties in a suboptimal setting exchange reasons, thus aiming at persuasion, do the rules for critical discussion apply in the same way as in normal settings? My interest is not in the rules themselves but in what they should enable: properly conducted argumentation should facilitate a reasonable disagreement resolution. Then, when an argumentative move hinders the chances of arriving at a rational resolution, it should be considered unreasonable. But is it reasonable to assess argumentation from the same normative stance when the pre-conditions for dispute resolution are mostly not in place? The main contribution of this paper is not to develop a completely novel argumentation theory for suboptimal settings. Instead, the ambition is to add to the existing literature on persistent disagreements and difficult argumentative settings by pointing to the usefulness of the concepts of ‘suboptimality’ and of a very general ‘supernorm’ that is even applicable when higher-order conditions are not satisfied.

I claim that there are two extreme positions and a middle-ground position regarding this question, each of which I take to propose a specific kind of ‘policy’ regarding how to evaluate contributions to the exchange of reasons. The first position is connected to what I call the “anything goes policy,” which comes down to the view that, since the conditions for a reasonable exchange of reasons are not satisfied, the discussion is outside the domain of reason. Accordingly, anything goes for the parties, and we are recommended not to apply the rules for critical discussion. The second position, connected to the “business as usual policy”, says that it doesn’t matter if the setting is suboptimal or not. Either way, the rules for critical discussion should apply in the same way. The middle ground position that I would like to elaborate on leads to the “supernormal policy” and implies that, in suboptimal circumstances, we need to adopt a more comprehensive approach to reasonableness: some moves that in other cases would be considered fallacious, shouldn’t be considered as such if they help the parties to arrive at a mutual understanding and, ultimately, create better conditions for reasonable dialogue. For the supernormal policy to properly work, we would need (instead of the rules for critical discussion) a kind of supernorm in place that helps the parties restore the proper conditions for reasonable dialogue or compensate for irredeemable cracks in the argumentative setting. Some authors seem to accept such a general norm (Gilbert 1995, 1997; Jacobs 1998, 2006) and hold the view that we should evaluate argumentation using this norm at all times, not just when the higher-order conditions for resolving disputes remain unsatisfied. I propose to use this approach only in a more limited way: to prepare the parties for a normal exchange of reasons that might help them advance towards resolution.

The supernormal policy requires a two-step processFootnote 3: during the first step, the parties will either try to restore the suboptimal setting or compensate for the suboptimality of the setting in some way. They will try to resolve the disagreement that prompted their discussion during the second step. The first step, then, is a procedural precursor for resolution in which the parties attempt to enhance the setting by bringing their positions together, while the second step is a proper resolution process, as it is understood in pragma-dialectics. That is, resolution is a process by which: “the argumentative discourse has resulted in an agreement between the parties involved on whether or not the standpoint at issue is acceptable” (van Eemeren et al. 2014, p. 528). I label the first step “preparation” and the second step “resolution”.

So, while for the preparation process, the parties can use moves that would be considered fallacious under normal settings, that is not possible for the resolution process. Here the dialogue must be adequately regimented to allow the parties to resolve their difference of opinion (cf. van Eemeren and Grootendorst 1984, p. 17).

I consider this problem from the standpoint of pragma-dialectics (van Eemeren and Grootendorst 1984; Eemeren and Grootendorst 2004). The main reason for adopting this approach is that it is a well-worked model that provides a plausible account of the norms of reasonableness for argumentative exchanges. However, the third policy departs from pragma-dialectics in two important respects: first, it establishes the necessity of a broader normative approach to argumentation, mainly inspired by Michael Gilbert’s (1995) coalescent argumentation, and by Scott Jacobs’s normative pragmatics (2000), and implies that some dialogue moves that for pragma-dialectics would be unreasonable, should not be considered as such.

The paper is structured as follows: in the second section, I define and discuss the fundamental concept of a suboptimal setting. The third section introduces and criticises the first solution: the anything-goes policy. In the fourth section, I evaluate the second solution: business as usual policy. In the fifth section, I introduce the supernormal policy. The sixth section applies this policy to some case studies that have been analysed before in the literature. Finally, I present some concluding remarks.

2 Basic Concepts

It is necessary, first, to clarify some concepts.

2.1 Persuasion dialogue

As Walton and Krabbe (1995) argue, there are many dialogue types, and persuasion dialogue is just one of them. Besides persuasion dialogue, we find negotiation dialogue, eristic dialogue, deliberation dialogue, information-seeking dialogue, inquiry dialogue, and other mixed dialogue types. Persuasion dialogue only occurs if the parties disagree and not in other cases like cases where they need to know something, where an information-seeking dialogue is more suited.

And even if they disagree, the parties can do many things besides attempting to persuade each other. They can negotiate a compromise, deliberate, or even toss a coin. Elsewhere (HIDDEN FOR REVIEWERS), I’ve specified how parties overcome persistent disagreements by shifting from one to another dialogue type. But in this paper, I want to concentrate only on persuasion dialogues because not every persuasion dialogue can be the subject of shifts. Notably, disagreements about what to believe, in contrast to disagreements about what to do, can’t be shifted to other dialogue types (as shown in my paper HIDDEN FOR REVIEWERS).

There are two dimensions of persuasion dialogues: descriptive and normative. The descriptive dimension enables us to identify a certain type of dialogue among others and show its main features. The normative perspective requires that we provide specific rules to evaluate persuasion dialogues and prescribe how they should be conducted in the future. When Walton and Krabbe descriptively analyse a persuasion dialogue, they show the main features of such a dialogue: the initial situation is disagreement, the main goal is to resolve the disagreement, and each of the participants’ aim is trying to persuade the other party (1995, p. 68). When they analyse the normative dimension, they use their “systems of dialogue rules” (1995, pp. 123–172) to formalise norms to evaluate persuasion dialogues.

When I say “persuasion dialogue”, I refer to a dialogue in which the initial situation is disagreement. The main goal is to resolve the disagreement, and each of the participants’ aim is to persuade the other party (1995, p. 68). Only if we allow this minimal definition of persuasion dialogue we can adequately say that such a dialogue can be conducted in a suboptimal setting. The parties can have other aims in mind, but we still call it persuasion dialogue if persuasion is the main one. On the other hand, the normative dimension is “critical discussion”.

2.2 Critical discussion

A critical discussion is a dialogue in which:

The parties attempt to reach agreement about the acceptability of the standpoints at issue by finding out whether these standpoints are tenable against doubt and other criticism, given the mutually accepted starting points ( van Eemeren et al. 2014, p. 528).

In a critical discussion, the parties begin with a difference of opinion regarding a specific standpoint, and if one of them succeeds in defending her standpoint or attacking the one presented by the counterpart, then one of them needs to retract her original standpoint or criticism, and the parties can reach an agreement. If they conduct the process reasonably, the parties will arrive at a resolution.

Critical discussion is a normative ideal for conducting argumentation when the parties find themselves in a disagreement. Then, let’s consider persuasion dialogue as a dialogue in which the parties try to persuade each other to reach an agreement; and critical discussion as a normative model that provides us with rules to reach an agreement reasonably.

According to the pragma-dialectical school, to conduct a critical discussion reasonably, the parties need to comply with the rules that authorise the performance of specific speech acts in the four stages of the process (confrontation, opening, argumentation, and concluding stage). These standards can evaluate whether a contribution to the dialogue helps the parties reach a reasonable resolution. These standards are known as rules for critical discussion (RCD).

These rules “constitute a dialectical procedure for the performance of speech acts in a critical discussion” (van Eemeren et al. 2014, p. 539) and are presented as a set of 15 rules (van Eemeren and Grootendorst 2004, pp. 136–157) or, in its shorter version, as a set of 10 commandments (2004, pp. 190–196). By “RCD”, I understand, in this paper, the list of ten commandments.

2.3 Fallacies

Along with the establishment of do’s and dont’s for a critical discussion, the RCD serve to reinterpret and unify most of the classical fallacies inherited from the Aristotelian tradition. Accordingly, a “fallacy” is defined as “a discussion move that violates in some way a rule for critical discussion applying to a particular discussion stage” (van Eemeren et al. 2014, p. 523). In other words, a fallacy is a “speech act that prejudices or frustrates efforts to resolve a difference of opinion on the merits” (van Eemeren 2010, p. 198). Therefore, it is impossible to resolve a difference of opinion based on fallacies since the resolution will not arise from the merits of the arguments. Thus if a fallacy is presented, it needs to be retracted or compensated for.

If fallacies imply violations of rules, then any fallacy can be reconstructed as the violation of such or such rule. For instance, if a party uses a threat to keep her counterpart from presenting or defending a standpoint, she commits an ad baculum fallacy. The reason is that she violates the freedom rule, which says that “discussants may not prevent each other from advancing standpoints or from calling standpoints into question” (van Eemeren and Grootendorst 2004, p. 190).

In conclusion, critical discussion is a normative model that reasonably enables the parties to resolve a difference of opinion. The RCD regulates the model, whose observance ensures that, if the parties arrive at an agreement, it a resolution based on the merits and whose inobservance will impede or, at least, hinder them from resolving their difference of opinion. Fallacies are violations of the RCD, so the parties need to avoid using them if they want to resolve their difference of opinion on the merits.

2.4 Higher-order conditions

Critical discussion, more than a theory to describe actual discourses, “is a theory of how discourse would be structured if it were purely resolution oriented” (van Eemeren, Grootendorst, Jackson, & Jacobs, 1993, p. 26). It plays a role as an ideal to which actual dialogues should be compared to. However, “the system described above assumes that certain conditions hold” (1993, p. 30). Those conditions have been called higher order conditions (HOC). They distinguish twonkinds: second and third-order, considering that the RCD are the first-order conditions for resolving a disagreement.

Second-order conditions refer to an “idealised set of attitudes and intentions” (p. 31) of the parties, which imply that they “wish to resolve, and not merely to settle, the disagreement” (p. 31). Naturally, if the parties want to score points or damage the adversary instead of resolving the disagreement, a reasonable resolution is not possible. But sometimes the problem is not that they are unwilling but, instead, that they are incapable of resolving the issue since they lack the “ability to express their opinions, to listen to the opinions of others, and to change their own opinions when these fail to survive critical examination” (p. 33).

However, willingness and ability are not enough to conduct a critical discussion, so besides the second-order conditions, there is a third-order that refers to external circumstances of the dialogue. Then, the parties must not only satisfy mental conditions for critical discussion but also “they must be enabled to claim the rights and responsibilities associated with the argumentative roles defined by the model” (p. 33). Therefore, to put forward and criticise standpoints, the parties need to “have the right to advance his or her view to the best of his or her ability” (p.33). This right can be thwarted in many ways: in a particular context, there might be taboo topics, unfair time constraints, authority relationships or dogmatic positions, among others.

Zenker (2007, p. 12) has provided a list of HOC. Among the essential 2nd order conditions: people must be willing to listen to the other party, be prepared to accept that the other party’s point of view might be justified and theirs might be wrong, be disinterested in the outcome of the discussion. Among the 3rd order conditions, people must be free to speak their minds, have no authority relationships, and have equal time constraints.

2.5 Suboptimal settings

Considering the above, a persuasion dialogue that takes place in a suboptimal setting can be defined as a persuasion dialogue that fails to meet, to some substantial degree, one or more higher order conditions.

Suboptimal settings are situations that stand in between the total abandonment of reasonable dialogue and minor glitches. I call a setting ‘normal’ in situations where the higher order conditions are, for the most part, met (though small departures would be permitted). I call “hopeless settings”, those cases when the parties are entirely unwilling to argue persuasively (and, maybe, more than a persuasion dialogue, we are in the presence of an eristic one). Finally, when parties do try to persuade each other, but the setting is not optimal, we will be in the presence of suboptimal settings.

Suboptimality, however, is not a region of dialogues with clear borders. Instead, it is located in a sort of continuum. On one extreme, we find normal settings and, as we move away from the normal settings, we approach cases where the HOC are less and less present. At some point, the dialogue becomes purely eristic (Walton and Krabbe 1995) or the disagreement too deep beyond repair (Fogelin 1985). So, the best way to characterise sub-optimal settings is to show what they are not. They are not “normal” settings, that is, the type of settings where the parties have the will, abilities, and social conditions to reach a resolution through critical discussion. But, neither, they are hopeless settings, where the parties don’t want to persuade each other at all, so persuasion is not one of their goals, or there isn’t any way to restore the setting and allow for resolution, as it happens with Fogelin’s “deep disagreements” (Fogelin 1985).

If the RCD are related to the HOC, it is not clear whether the RCD are the best way to describe reasonable contributions to the dialogue in suboptimal settings. Therefore, we could re-phrase the research question of this paper in the following terms: are the rules for critical discussion the best way to describe reasonable contributions to a persuasive dialogue if one or more of the higher order conditions for critical discussion are not met to some substantial degree? Aakhus (2003, p. 263) presents this question as follows:

When second and third order conditions are not satisfied, is it reasonable to conform strictly to the ideal model of critical discussion to understand whether a move fosters progress toward solving the conflict? Or, if the second and third order conditions are not met, then are the standards for judging argumentation based on the assumptions for critical discussion the best standards to use to interpret and evaluate argumentation in practical settings?

In what follows, I develop three solutions to this problem.

3 First Proposal: Anything Goes Policy

The “anything goes policy” can be characterised as follows: When parties try to persuade one another in suboptimal settings, the rules for critical discussion don’t play a role in evaluating contributions to the dialogue.

The main reason to support the anything-goes policy is that the higher order conditions are enabling conditions for a critical discussion. Therefore, if one or more of them are not met, we might have a persuasion dialogue but not a critical discussion. As a result, the RCD are only applicable in normal settings. Zenker characterises (while not necessarily defending) this solution as follows:

The point to note, then, is this: If a text appears argumentative, but its setting fails to comply to some higher order condition, then this discourse must not be interpreted as one that is aimed at a resolution of a difference of opinion, to begin with (cf. van Eemeren and Grootendorst 2004). In such cases, the discourse simply ceases to be a proper object for the Pragma-Dialectical theory (2007, p. 13).

If that is the case, then the RCD are not in place anymore, so we would have two options: (1) there are no appropriate rules for evaluating the argumentative quality of dialogue contributions in these settings; or (2) we need another, a maybe more general norm to understand why a fallacy is a forbidden move in these cases. This section deals only with the first option, while the second one will be re-examined in Sect. 5 of this paper. Anything goes policy, then, refers only to the first option.

We could look at the term “condition” to back the first option. The Merriam-Webster dictionary defines it as: “something essential to the appearance or occurrence of something else”. Considering this definition, a synonym of this meaning would be “prerequisite”. But suppose the higher order conditions are a prerequisite for critical discussion. In that case, the consequence is quite clear. If the conditions are not satisfied, it would be inappropriate or even unfair to analyse the reasonableness of contributions using the RCD.

While somewhat appealing, an anything-goes policy seems wrong for two main reasons. The first is that we can recognise fallacious moves even in a suboptimal setting. The second is that it could be argued that every persuasion dialogue is suboptimal, so the RCD would never be the best way to evaluate whether a contribution to the dialogue is reasonable. In real-case scenarios, the parties will have some reluctance to recognise they have been proven wrong, some interest in the outcome of the discussion, some lack of emotional restraint or some authority relationship.

In the end, in real-life argumentation, the conditions are never totally met, so saying that anything goes when the higher order conditions are not met implies that anything goes in any discussion, and that’s an absurd outcome. If that is the case, we can think that the RCD are always in place in persuasion dialogues, which would be the second solution.

4 Second Proposal: Business as Usual Policy

The “business as usual policy” can be defined as follows: When parties try to persuade one another in suboptimal settings, the rules for critical discussion are the suitable model for evaluating which contribution is reasonable in a dialogue.

Someone could argue that there is not a problem here, and the only problem is that the word “condition” seems to entail a requirement or essential condition. But if we consider critical discussion to be just an ideal model, the higher order conditions are, by definition, never met. Therefore, the distinction between the higher-order conditions being slightly or substantially not met is irrelevant. If that is the case, we could always reconstruct persuasion dialogues in terms of critical discussion and call violations of the RCD “fallacies”.

Pondering the above, the solution to the problem posed should be business as usual policy. Then, if we reconstruct a persuasive dialogue of an exact academic debate or a marital disagreement, the situation would be the same: the parties must avoid fallacies to arrive at a resolution and, if they commit them, they might arrive at an agreement, but it will not be on the merits.

Business as usual seems to be the standard position taken by the pragma-dialectic school:

When analysing argumentative discourse, the normative ideal of a critical discussion serves as a kind of template against which experience can be compared and a kind of standard against which it can be judged. As we will see, actual human interaction is not “naturally” resolution oriented. People involved in disagreement are not normally disinterested in the outcome but have a heavy interest in one outcome or another. They do not generally enter into discussion willing to subject all of their thinking to debate but treat certain things as so fundamental as to be beyond challenge. They have deficiencies of skill. They argue within social conditions that virtually assure some degree of inequality in power and resources (…). Actual practices are not described by such a model, but certain of their features can be given interesting explanations in terms of the model. (van Eemeren F. H., Grootendorst, Jackson, & Jacobs, 1993, p. 34)

For these authors, the higher order conditions are never fully met, so the RCD are an ideal model. Critical discussion is a sort of template, a blueprint used to evaluate actual practices (Aakhus 2003). Therefore, the fact that the higher order conditions are not met is not a reason not to evaluate contributions to the dialogue using the RCD. Then, even if the parties arguing are, for example, intimates with strong emotional attachments, the dialogue can be evaluated using the RCD.

But there might be problems with this interpretation. In the case of the anything-goes policy, stressing the term “condition” too much makes the system excessively permissive; in the case of business as usual policy, stressing the term “ideal” too much could make it too strict and not adaptable to circumstances.

There are two main objections against this conception of the business as usual policy. The first one is that it doesn’t make any difference between the optimal and suboptimal settings when it comes to the issue of whether the rules for critical discussion are the best standard for evaluating reasonableness; the second is that it leaves situations in grey areas unresolved.

The first objection says that the RCD are too strict about evaluating dialogues in suboptimal settings. But, for example, we would consider that appeals to emotions are a proper and nonfallacious way of arguing when a mother argues with her daughter.

The second reason to reject the business as usual policy is that it does not tell us what to do with grey areas. Pragma-dialectics does recognise that, in some cases, the RCD are not the best to determine whether a contribution is reasonable, especially when the arguers defend incommensurable standpoints or there is a clash argumentative points of departure. In such cases, “much of what is wrong appears to result from the absence of an essential second-order condition for critical discussion—a serious, resolution-oriented attitude on the part of the participants” (van Eemeren F. H., Grootendorst, Jackson, & Jacobs, 1993, p. 166). But pragma-dialectics doesn’t tell us what to do in those cases. Can we apply the rules for critical discussion, or we should just look for another way to move forward? And which way could that be?

5 Third Proposal: Supernormal Policy

5.1 The supernormal policy

Normative models of argumentation need to account not only for standard settings but also for suboptimal settings. In other words:

normative models must take into account that discussants are limited in their reasoning capacity and susceptible to fallacious reasoning; that they are usually each committed to a particular point of view and susceptible to prejudice; and that they are each motivated by a desire to win in a context of controversy. (Krabbe 2009, p. 122)

When confronted with suboptimal settings, what does a party that wants to resolve a disagreement reasonably need to do? According to the supernormal policy, the parties need to test each other’s standpoints in the resolution step rationally. But in the preparation step, there are two possible contributions, depending on the nature of the suboptimality.

Some suboptimal settings can be restored to normal by the parties. To do this, at least one of the parties must acknowledge that the dialogue is sub-optimal and that it could be restored (but, of course, she could be mistaken in these appreciations). For instance, lack of trust can sometimes be remediated by talking openly to each other; cognitive biases can be sometimes revealed and corrected, lack of time or opportunities to present a good argument can be sometimes fixed, unwillingness from the parties to recognise when they are wrong can be worked out. When this is the case, the priority of the parties should be to restore the normal setting and put the discussion back on a reasonable track. For instance, as Kloster (2018) proposes, when there’s a lack of trust, the parties will have a hard time resolving their disagreement. Therefore, they need to know each other first to start trusting and then argue. For restoring the dialogue to a normal setting, I will mainly use Gilbert’s (1995) framework.

But in other cases, the setting cannot be restored. Authority relationships, for instance, will not go away even if the parties want them to, argumentative injustice cannot be changed within the timespan of a single discussion, emotional dispositions are just part of the character of some people, and so on. In such cases, the parties’ aim should be to compensate for the circumstances that make the disagreement suboptimal. Since the problem cannot be taken away, we need to learn to live with it, to make it as less harmful as possible. To understand how this compensation works, I will use Jacobs’ (2000) normative pragmatics as a framework.

The distinction between restoration and compensation is not clear cut, but it seems worthwhile. Both processes require that the arguers lean towards rhetoric but in a different way. Following Gilbert, the restoration process is more emphatic and seeks to coalesce the positions of the parties; it is more about mutual understanding than anything else. The compensation process, on the other hand, and following Jacobs, is inevitability more assertive: it doesn’t necessarily seek to bring the parties together but to call attention to the suboptimality of the setting. In any case, both processes use rhetorical techniques that could be considered fallacious, at least from a pragma-dialectical standpoint.

In the dialogue’s preparation process, the RCD doesn’t serve the same function in a normal setting. While the RCD are a set of specific norms, this process would need a sort of more general supernorm, which should capture the spirit that animates the rules for critical discussion, but in a general and non-specific way, so that it provides guidance also when the conditions for critical discussion are not satisfied.

It is also relevant to note that, as described before, suboptimality represents a sort of continuum so that some settings would be more suboptimal than others. That distance from the normal setting should be one of the guiding factors for proceeding in the argumentative process. Then, the closer we are to a normal setting, the stronger we should be committed to following the RCDFootnote 4.

The supernormal policy would not be followed, for example, when parties use arguments that might escalate the disagreement and not restore or compensate the suboptimality of the setting. To evaluate what works, they need to assess the distance to the normal setting. Then, for example, an argumentation process between intimates close to an eristic dialogue should allow more emotional appeals than a political debate with a politician who cannot accept when he has lost an argument.

At this point, an objection can be formulatedFootnote 5: if departure from a normal setting is extreme, how extreme might a tactic be? Is it acceptable, for example, to censure or isolate people? That is a good question that exceeds to topic of this paper, but, at the very least, I can say the following: first, argumentation processes have obligations and norms of moral nature attached to them (Blair 2015). That is, to hit someone with a stick because we disagree is not only unreasonable for the resolution of a difference of opinion but also morally wrong, and argumentation is (also) related to moral duties. Secondly, if a party departs too much from a reasonable exchange, then maybe we are not talking about a suboptimal setting but a hopeless case. In that case, the supernormal policy is of no use. Finally, the supernormal policy implies that we should try to restore or compensate the setting. The aim here is to overcome disagreements peacefully and reasonably, so anything that deviates from this goal shouldn’t be considered desirable for the supernormal policy. That provides a limit to the strategies that can be used.

We can formulate this supernorm in the following terms: When parties try to persuade one another in suboptimal settings, their contributions to the dialogue must be evaluated considering how much they help restore the dialogue to a normal setting or compensate it when the restoration is not possible.

I will now explore the two contributions that this supernorm allows to restore and compensate.

5.2 Restoring the normal setting

To understand how the restoration of normal settings works, it is helpful to understand the concept of coalescent argumentation, developed by Michael Gilbert.

According to Gilbert (1995; 1997; 2002), one of the mistakes of dialectical theories of argumentation is to think that argumentation is solely about presenting arguments through language and logical connections. He claims that this belief can be called a logocentric fallacy, which “is committed when language, especially in its most logical guise, is seen as the only form of rational communication” (2002, p. 32).

To avoid this mistake, he begins by understanding an argument as “an exchange of information centered on an avowed disagreement” (1995, p. 839). Consequently, when people disagree, they not only disagree about a claim but, somewhat, about a position. A position is “a matrix of beliefs, attitudes, emotions, insights, and values connected to a claim.”(Gilbert 1995). The claim is just the tip of the iceberg, while the position (that mostly remains underwater) is what is really producing the disagreement.

For example: if a couple disagrees about the best school for their children, they not only disagree about the claim “school p is more suited than school q for Maria” but, rather, about a whole set of beliefs and attitudes that are connected to that claim: therefore, they have clashing positions, not only clashing claimsFootnote 6. For pragma-dialectics, what matters in this example are the clashing claims. But, probably, analysing only those claims will not allow us to know what is happening. If the claim is rooted in the feelings, life experience, or situation of the parties, it is unlikely that the disagreement can be resolved only through an exchange of logical reasons, so we need to look beyond that exchange.

So, if the parties want to reach an agreement, they need to understand their respective positions. In that sense, argumentation is a process in which the parties uncover the positions that back or warrant their respective claims: “one must garner not only the facts that support the claim but the values, emotions and attitudes that go along with the outlook attached to the claim.” (Gilbert 1995). Such a process of understanding a certain position is instrumental to what, in this paper, has been called restoring a normal setting.

To do so, the parties must explore different modes of arguing. Gilbert claims that there are four modes: “In addition to the classical logical mode. . there are the emotional, visceral (physical), and kisceral (intuitive [& non-sensory]) modes” (1997, 75). If the parties explore these modes, they are better positioned to arrive at an agreement. This process of mutual understanding through different modes has been called “coalescent argumentation”: “The aim of coalescent argumentation is to bring about an agreement between two arguers based on the conjoining of their positions in as many ways as possible” (Gilbert 1997, p. 70).

In a way, suboptimal settings are mostly suboptimal for argumentation in the logical mode, but not necessarily for argumentation on one of the other modes. Sometimes people argue with emotional appeals that are not prima facie wrong but only wrong when analysed from a logical perspective. Gilbert’s criticism of pragma-dialectics, informal logic and other normative approaches is that they focus way too much on the logical mode (Gilbert 2002), reducing the chances of the parties to resolve their disagreements. When faced with suboptimal settings, the parties should explore these modes because they will have more means to coalesce and look for a mutual understanding.

Arguing with people in echo chambers is a good example of a suboptimal setting that could eventually restore to normal. “An echo chamber is a social epistemic structure from which other relevant voices have been actively excluded and discredited” (Nguyen 2020, p. 141). That is, people in echo chambers have learned to distrust certain epistemic sources, so arguments coming from that source are not persuasive by definition. This distrust implies that the setting is suboptimal (and sometimes hopeless), so the parties cannot expect to resolve their disagreement through normal argumentation. However, by “cultivating trust between echo chamber members and outsiders” (2020, p. 158), it might be possible to restore the setting to normal. In my view, this might be achieved through many ways, among them, argumentation that seeks coalescence and building trust, not resolution. As Nguyen (2020, p. 159) argues:

Thus, the route to undoing their influence [of the echo chambers] is not through direct exposure to supposedly neutral facts and information; those sources have been pre-emptively undermined. It is to address the structures of discredit – to work to repair the broken trust between echo chamber members and the outside social world.

In cases like these, the way to restore a dialogue to a proper setting is to use empathy to build trust and understanding, for example, through different modes of arguing, and not merely use a logical approach to analyse statements.

5.3 Compensating for a suboptimal setting

If the suboptimal setting can be restored into a normal one, it should be restored. But that is not always possible since some settings cannot be changed. Therefore, we need to work out a way to compensate for the misbalances of the suboptimal setting. To do that, it might be necessary to call the counterpart’s attention through a rhetorical move. This move could imply, among other things, the use of moves that under normal settings should be called fallacious. In the Topics, Aristotle seems to be arguing in this sense when he claims:

… often the person questioned is the cause of the argument not being properly discussed, because he does not concede the points which would have enabled the argument against his thesis to have been properly carried out; for it is not within the power of one party only to ensure the proper accomplishment of the common task. It is, therefore, necessary sometimes to attack the speaker and not the thesis, when the answerer is on the watch for points against the questioner and also employs abuse. (Aristotle, ca. 350 B.C.E. 2014,(8.11, 161a17-24)

By calling it necessary, Aristotle suggests that the parties can fulfil their dialectic goals by including such an eristic move (Krabbe 2009). But in which sense? Jacobs’s normative pragmatics (2000) can be helpful to understand this point.

Rather than considering only the arguments presented, normative pragmatics focuses “on the communicative properties of actual argumentative messages”. (p. 262). Messages are more complex than just arguments. While arguments are what was said, messages also include “the way it was said, when it was said, who it was said to, by whom; all in contrast to what could have been said but was not” (p. 263). By contrast, Jacobs criticises dialectical approaches for only considering the arguments given and forcing their normative model to descriptive messages.

So, according to Jacobs, when looking at arguments in this way, we cannot just label some moves as “fallacies” if they violate certain norms that, we claim, are required to conduct a dialogue properly. Instead, when we consider the whole message, some rhetorical strategies that might appear as fallacious, when regarded only in the sense of what was said, can be constructive for resolving the disagreement. For instance: “emotional appeals can play a constructive role in deliberation and may be positively required by the situation” (p. 277). Or, elsewhere: “apparently fallacious rhetorical tactics can be defended as playing a constructive role in the debate that’s actually going on” (p. 278).

Suppose one of the parties uses rhetorical strategies to frame the discussion in a certain way that benefits her or constantly uses fallacies. In that case, it might be necessary to use rhetorical strategies that we could call fallacious to counterbalance the rhetorical effects produced by their message. Therefore, these strategies can be constructive: “They respond to the demand to find ways to place people in more open, critical, resolution-oriented frames of mind and to make the conditions for argumentation conducive to reasoned deliberation.” (p. 281).

Accordingly, suboptimal settings can be circumstances where the use of rhetorical strategies (fallacious or not) is recommended: “Rhetorical strategy can be a reasonable way of overcoming or working around the practical insufficiencies of the situation” (p. 282).

Considering the above said, the compensation strategy seems to be more assertive than the restoration strategy. While restoration is about building trust and understanding, compensation is about calling the counterpart’s attention to counterbalance the situation, considering that the suboptimality of the situation is probably unable to be restored. In that sense, the parties might even use counter-fallacies (van Eemeren and Houtlosser 2015). In the same way, Van Laar and Krabbe (2016, p. 331) consider that fallacies could be sometimes used as a sanction against a party acting unfairly.

6 Some Case Studies of the Supernormal Policy

I want to apply the above ideas to real-life case studies in this section. I have expressly chosen case studies that have been analysed before in the literature. I aim to fulfil three main objectives: first, provide realistic examples of the supernormal policy; second, to contrast my diagnosis with the one made by others in the field; and third, to specify the normative force of the supernormal policy.

To test the supernormal policy, it would be necessary to answer “yes” to the following questionsFootnote 7: (1) Is the setting of the dialogue suboptimal? (2) Does one of the parties use a dialectical move that could be considered fallacious or unreasonable, at least from a pragma-dialectical framework? (3) Does the allegedly fallacious move help the parties restore the dialogue to a normal setting or compensate for its suboptimality when the restoration is not possible? If these three questions can be answered affirmatively, it can be concluded that the supernormal policy fulfils its expected effects, at least in the cases presented.

6.1 Case study of restoration of the normal setting

The following case is an example of suboptimal settings where the parties make certain moves to restore the dialogue to a normal setting. Following Gilbert, in this process, the parties will need to use different modes of arguing (some of which might be considered fallacious in a normal setting) to build trust and understanding between the parties, thus restoring the dialogue to a normal setting.

6.1.1 Case study 1: GMO debate

This case study has been taken from Goodwin (2016) and concerns the rhetorical strategy used by the scientist Kevin Folta. Folta is a well-known biologist working on transgenic crops (GMO’s), actively defending their use in the public sphere. Goodwin points out that, in cases like these, cooperation from the counterpart cannot be assumed:

“We do not assume that participants in an argumentative transaction or ‘dialogue’—especially in civic life—will inevitably be cooperative, reasonable, or possessed of any of the multitude of virtues that make transactions run well. Instead, we examine what arguers themselves can do to design or modify the contexts in which arguments are exchanged”. (Goodwin 2016, p. 1)

According to Goodwin, parties need to design the dialogues’ contexts to frame their contributions to the dialogue in a light that serves better their goals. In Folta’s case, he needs to make his objectivity apparent: part of his audience thinks that he is partial towards the biotech industry, so they don’t grant him enough trust to believe that his scientific discourse is objective enough. His challenge is to be seen as non-partisan or unbiased.

The dialogue in question is a talk at a University in which Folta is trying to defend the use of GMO’s by addressing three topics: “what GMOs are, why people resist them, and what future holds for biotechnology” (2016, p. 2). The interactions with the public are varied, but some interlocutors were specially aggressive towards the speaker, as the following interaction shows (Goodwin 2016, p. 9):

One of the interlocutor’s of Folta is “hostile man” (HM). At several occasions he interrupts Folta (KF). For instance:

(1) HM: “you have to stop right there because that is the biggest lie that there ever was.”

Instead of dealing with the accusation of “lie” or shutting down the interruption, Folta answers:

(2) KF: “This is a great opportunity for us to have a conversation, I’d love to be able to follow up on you. You can be the first person at the end when we talk about it. I’d love to answer that question.” (2016, p. 9).

For Goodwin, what Folta is doing is the “re-characterisation of HM’s intervention as a ‘question’ that’s part of a ‘conversation’.”(2016, p. 9). Folta seeks to reframe attacks into welcome questions, presuming good faith against the evidence and addressing those attacks according to such a presumption. In other words, “he aggressively presumes their good faith, even when they themselves don’t display it” (2016, p. 10). His ultimate goal is to give the audience reasons to believe that he’s being objective.

Is the dialogue between KF and HM suboptimal? The interaction just transcribed doesn’t give us the full picture. Still, from the tone of the interaction (and the general context described by Goodwin) it can be assumed that we are dealing with a sub-optimal setting: HM is unwilling to listen or change his mind, so he doesn’t fulfil HOCs Nº 1 and 3 described in Sect. 2.4.

Can be KF’s answer and general strategy considered fallacious or unreasonable from the standpoint of critical discussion? As Goodwin points out, Folta is presuming good faith against the evidence. This means that he is not addressing HM’s real commitments but, instead, watering down his interventions. He is also evading HM’s accusation of “liar”. Accordingly, it could be argued that KF’s intervention move violates the “freedom rule”, that states that parties “may not prevent each other from advancing standpoints or from calling standpoints into question” (van Eemeren and Grootendorst 2004, p. 190) and/or the “relevance rule” that states that “Standpoints may not be defended by non-argumentation or argumentation that is not relevant to the standpoint.” (2004, p. 191).

Does Folta’s strategy help restore the dialogue to a normal setting? The dialogue isn’t only directed towards HM but towards a diverse audience. HM leaves the room before the dialogue is over. Still, as Goodwin recognises, Folta’s strategy does help to create trust with members of the audience that were originally sceptical, thus enabling an eventual resolution. In that sense, I think that his strategy does help the parties restore the setting to normal. What Folta tries to achieve, ultimately, is to bring coalescence between him and HM, and towards the rest of the audience, even if that implies not following the RCD by the book.

6.2 Case study of compensation of the suboptimal setting

In this case, the suboptimal setting is unlikely to be changed through mutual understanding, so the parties need to call the counterpart’s attention to compensate for the rhetorical situation. The moves might seem, thus, a little more aggressive than in the previous examples.

6.2.1 Case study 2: Partial-birth abortion

Jacobs (2000, pp. 275–279) analyses the following ad concerning abortion that appeared in American newspapers:

“Killing mostly-born infants.

with a pair of scissors.

We’re debating this?

This is not a bad dream. It’s real. Every year, thousands of infants are forcibly dragged from their mother’s wombs, then killed. Brutally. And our nation is actually debating whether or not this should be allowed to continue.

Congress calls this ‘procedure’ partial-birth abortion. Doctors say it’s never medically necessary, and is even dangerous for the mothers. As religious leaders, we say it’s infanticide. And morally bankrupt.

This is a defining moment for America. We stand on the brink of a moral abyss. Others once stood on that brink – others who tolerated the killing of innocent people, including children. It happened in Germany half a century ago. It has happened in Russia, China and Cambodia. It’s happening in Rwanda today. What will historians say about us if we cannot summon the moral will to ban the killing of infants, moments before birth?

When we first hear of atrocities, our natural reaction is disbelief, then outrage. But so many lies get told and repeated, it becomes hard to know what’s true. Eventually, we tune it out. Sadly, that’s all it takes. Good people, confused, and with little time to sort things out, sit on the sidelines.

The truth is in your heart. Listen to that small voice within. For the sake of the littlest children, speak out. We must not turn a blind eye and a deaf ear toward those so helpless they cannot even ask for our help.

Next week, your Senator will be voting on whether or not to ban this ‘procedure.’ Tell him or her to put a stop to partial-birth abortion in the United States of America. Please. Call the Senate switchboard today at (202) 224–3121.

THE NATIONAL CONFERENCE OF CATHOLIC BISHOPS.

SECRETARIAT FOR PRO-LIFE ACTIVITIES”.

Jacobs claims that the ad is full of arguments that can be labelled as fallacies according to textbooks in argumentation. However, he argues that “these apparently fallacious rhetorical tactics can be defended as playing a constructive role in the debate that’s actually going on” (Jacobs 2000, p. 278). In his view, these arguments and rhetorical figures create space for the parties for exploring different approaches to the problems at issue. This is presented as a process of “self-regulation of the disagreement”. He claims that the rhetorical strategy used in the ad is an appropriate counterpoint to the rhetorical tactics used by the other party.

Indeed, this ad is a sort of response to a previous communication by National Organization for Women (p. 278):

“Myth: If late-term abortions are so rare, it doesn’t really matter if we ban them.

Fact: Most late-term procedures involve wanted pregnancies that go tragically wrong when the woman’s life or health is endangered or the fetus develops abnormalities incompatible with life.

These women deserve protection, even though their situations are rare.

Myth: The ‘partial-birth’ abortion procedure is unnecessary.

Fact: Medical experts state that the safest method of late pregnancy termination for some women is the intact dilation and extraction (D&X) procedure.

Bills that would ban the D&X procedure would place legislators in the unprecedented position of regulating medical decisions and would require doctors to use a less safe method in some cases.”

Jacobs claims that the language used by this organisation is “positively anaesthetising”. They frame the discussion from the pregnant woman’s perspective, which in Jacob’s opinion is not less biased than the ad presented before, that frames it from the unborn child. That’s why the apparent fallacies do play a role in helping the parties to find new ways to look at the problem. I agree with Jacobs here, and from the viewpoint adopted in my paper, I claim that a way to understand the role of those moves is to see the setting as suboptimal and justify the use of apparently fallacious tactics using the supernormal policy. So, let’s review our three questions:

First, is the setting suboptimal? The topic of abortion has been, for long, considered an example of intractable or even deep disagreement (Fogelin 1985). The problem, according to Fogelin, is that when people disagree about the morality of abortion, they disagree about a whole set of connected propositions that constitute a “way of life”, that is, a way in which we see the world. Now, even without going as far as Fogelin goes (who thinks that there’s not a rational way to resolve this disagreement), we can say that a disagreement about abortion is, at least in this case, suboptimal. Therefore, it is not to be expected that the “National Conference of Catholic Bishops”, the “Secretariat for pro-life activities” or the “National Organization for Women” are “willing to accept that their points of view can prove to be wrong” nor are they “disinterestedness in the outcome of the discussion”. In fact, they have a firm conviction regarding their main standpoint that is unlikely to be changed easily. The same can be said about the “National Organization for Women”.

Second, is the strategy fallacious from the standpoint of critical discussion? Jacobs properly develops this point and claims that the ad published by pro-life organisations can be accused of at least committing three fallacies: question-begging, declaring certain standpoints sacrosanct, and abusing emotional appeals. For pragma-dialectics (van Eemeren and Grootendorst 2004), question-begging is a violation of the “starting point rule”, declaring sacrosanct a standpoint is a violation of the “freedom rule”, and abusing of emotional appeals is a violation of the “relevance rule”.

Third: does the apparent rule violation succeed at compensating the situation’s suboptimality? According to Jacobs it does “call out and counterbalance the biases and defects in the argumentation of the other side” (p. 278). As I see it, the rhetorical strategy serves, at least, to show that the apparent “neutrality” of the counterpart is also biased. In that sense, it does compensate for the suboptimality. However, we don’t know if the strategy helps the parties resolve in the long run, but that doesn’t mean that the strategy is unreasonable; it could be reasonable, but just inefficient.

6.2.2 Case study 3: Trump on Charlottesville

The third case study is a press conference held by former President of the U.S. Donald Trump, and analysed by Jackson (2019, pp. 634–635). The topic of the press conference is Trump’s reaction to the Charlottesville incident, in which a group of white supremacists violently protested the removal of the statue of Confederate General Robert E. Lee and clashed against groups supporting the removal. On August 15, 2017, at a press conference, the following dialogue ensuedFootnote 8:

72 DT: [Okay, what about the alt-left that came charging at ‘em, excuse me, what about the alt-left that came charging at the, as you say, the alt-right? Do they have any semblance of guilt?

73 R6: ((inaudible clamor)) What are you saying?

74 DT: Let me ask you this. What about the fact that they came charging that they came charging with clubs in their hands, swinging clubs. Do they have any problem,=I think they do.=.

75 R6: Sorry [are you.

76 RX: [Mr. Trump 77 DT: =[As far as I’m concerned, that was a horrible, horrible day.

78 R7: (But) you’re not putting [((inaudible)) on the same level = as the Neo- Nazis and white supremacists,

79 DT: [Wait a minute. I’m not = finished. I’m not finished, fake news.

80 R7: Sir.

81 DT: That was a horrible day,

82 R7: you’re not putting these protestors on the same level as [neo-Nazis and white supremacists.

83 RX: [Is the alt-left as bad as white supremacy?

84 DT: I will tell you something. I watched those very closely, much more closely than you people watched it. And you have uh you had a group on one side that was bad, and you had a group on the other side that was also very violent. And nobody wants to say that, [but I’ll say it right now.

85 R7: [((inaudible clamor)) (you’ll say).

86 DT: You had a group you had a group on the other side that came charging in, without a permit, and they were very, very violent.

Jackson analyses this case as an example of naturally occurring discourse. That is, it is a “discourse that is clearly argumentative, but that participants themselves have difficulty parsing into claims and reasons.” (2019, p. 632) She claims that argumentation occurring in these settings is evaluated by the parties using “natural normativity”, which consist of “intuitions about what is and is not helpful in managing disagreements” (2019, p. 640). Sometimes, a clash between well-articulated normative systems and this “natural normativity” can occur, and then “the natural normativity of argument as a disagreement management system comes into play” (209, p. 641).

She claims that in the case presented, to capture what happens, we cannot be focused only on statements because what happens is much more than just statements. Specifically, the reporters are using “call-outs” which are challenges to what Trump is “presumed to hold but that he does not openly state” (2019, p.635). Call-outs are sometimes necessary for getting to productive disagreement since “callout-by-callout, a case on each side does appear” (2019. P. 636).

The supernormal policy does not contradict Jackson’s analysis but only adds a new perspective. Naturally occurring discourse might or might not be sub-optimal, but in any case, the parties need to evaluate what helps manage disagreements. And just as “natural normativity” might clash with “well-articulated normative systems”, the supernormal policy might clash with the rules for critical discussion.

There are three questions to answer: First, Is the setting sub-optimal? Trump’s discourse illustrates much of what happens when higher order conditions are not met in this and other dialogues. He doesn’t seem to be willing to accept that his points of view can be wrong, to provide justification for his views or lose face. We can say, then, that the press conference setting is suboptimal.

Second, do the reporters violate rules for critical discussion? It can be argued that that is the case. At least in interactions 78, 82 and 83 the reporters present “call-outs” that try to show Trump as committed to things he actually hasn’t said: “Criticism of the President’s response to Charlottesville is not aimed at the claims to which he commits himself (…) but at dispositions he is suspected of holding” (Jackson 2019, p. 636). In that sense, it could be argued that the move violates the “standpoint rule” that states: “Attacks on standpoints may not bear on a standpoint that has not actually been put forward by the other party” (van Eemeren and Grootendorst 2004, p. 191).

It could be argued that the journalists’ call-outs are not assertions but mere questions: they are not misrepresenting the President’s position but merely trying to find out what that position is. We can interpret the text in both senses, and some further context will be needed to settle the matter. However, at least contributions 78 and 82 look a lot like assertions that try to make conclusions from what the counterpart hasn’t expressly said. In contribution 78, the journalist presents what can be interpreted as a rhetorical question: “but you’re not putting on the same level as the Neo- Nazis and white supremacists?” which is repeated in 82. But rhetorical questions aren’t really a type of question but a type of assertion (Han 1998).

Third: does the apparent rule violation succeeds at addressing the suboptimality? I think that is the case. First, from Jackson’s perspective, it could be argued that the move helps the parties manage their disagreement by revealing where they stand. Second, from the perspective of the supernormal policy, the move might be considered “a reasonable way of overcoming or working around the practical insufficiencies of the situation” (Jacobs 2000, p. 282). The journalists are calling attention to the true nature of what Trump is implying but doesn’t want to say. Therefore, it is a way of compensating for the suboptimality of the situation.

7 Conclusions

To answer the question “what is a good argument?” argumentation theory needs to mediate between a fundamental tension. On the one hand, it has to present a normative model of what “good argumentation” entails. But on the other hand, it has to connect to real-life argumentation. When we consider suboptimal situations, that tension forces us to make a choice. How can a normative model account for this kind of cases? In this paper, I’ve presented a proposal.

This proposal adds to others because it stands in a kind of middle ground between, on the one hand, saying that every persuasive dialogue must conform to a model, and on the other hand, that the parties themselves must create or design that model. That is, it does recognise that we need normative models such as critical discussion to evaluate reasonable contributions to the dialogue. At the same time, it recognises that suboptimality might make the rules for critical discussion partially inapplicable, so we need a wider conception of reasonableness for those cases. This is not to say that other approaches to argumentation theory don´t deal with the application of normative models to real-life argumentation. Still, they don’t focus on faulty or suboptimal cases as I’ve tried to do here.

I do recognise, however, that many ideas advanced in this paper need further development. Among others, the concept of suboptimality could be better exposed, for which the higher-order conditions also need to be better developed. The supernormal policy could be further tested in case studies, and the connection between this policy and other strands of argumentation theory, especially those that lean towards rhetoric, further explained.