1 Introduction

According to the call for papers for this special issue, Pragma-dialectic’s “dialectical conception of reasonableness” holds that “a difference of opinion is reasonably resolved if, and only if, the parties reach a joint conclusion as a result of putting the acceptability of the standpoints at issue to the test by applying criteria that are problem-valid and intersubjectively valid.” We have criticized this conception of reasonableness on several previous occasions. In this paper we place those criticisms in the context of speech act theory. We argue, first, that PD’s reliance on that theory is problematic in that it fails to acknowledge the distinction between the speaker’s perlocutionary object and the act’s perlocutionary sequel, and that this distinction is essential not only to that theory but to PD’s account of argumentation. We then recast our criticisms of PD’s account of reasonableness in terms of that distinction. For PD, the act of arguing takes as its perlocutionary object the achieving of agreement between the parties, and it counts the total speech act as non-defective (felicitous) only if that object is accomplished. But perlocutionary object and perlocutionary sequel may, and often do, come apart. We argue further that the perlocutionary object of arguing is not the achievement of agreement; rather, it is the establishing of the truth or justificatory status of a claim, proposition, or conclusion. As a result, we argue, PD incorrectly counts as reasonable epistemically unreasonable arguments, and counts reasonable arguments as defective if they fail to produce agreement or resolution.

2 PD and Speech Act Theory

Among the chief desiderata the PD theory of argumentation must meet, van Eemeren and van Haaften list the following:

Empirical research has to make clear which kinds of communicative (‘illocutionary’) and interactional (‘perlocutionary’) acts are in actual argumentative discourse instrumental in making argumentative moves that are relevant to resolving a difference of opinion8 and which kinds of factors play a part in determining the outcome. (2023, p. 3, emphases in original)

They go on to explain in the embedded footnote:

Communicative and interactional acts are speech acts simultaneously performed in oral or written speech events. They are aimed at bringing about the communicative effect of understanding and the interactional effect of acceptance.

We will argue that this appeal to speech act theory is problematic.

To begin, we need to be clearer than van Eemeren and van Haaften are on the tri-partite structure of speech acts as described by Austin. In a single concrete act of speaking, one performs a locutionary, an illocutionary and a perlocutionary act. The first is the uttering of a linguistic expression that can be thought of – though Austin does not put it this way – as the vehicle via which the second and third are performed.Footnote 1 An act of arguing, at least of the central linguistic kind (we do not address the interesting question of whether there are any other, for example, pictorial, kinds), requires the performing of a locutionary act, and to perform one of a certain sort is typically to argue. (Not always: one may do so as part of an elocution lesson.) Arguing is the illocutionary act one may perform in saying something, and by saying what one does, one may be performing a perlocutionary act which has to do with the effect of one’s illocutionary act on one’s hearer(s). This much is uncontroversial. The crucial point for our purposes here is that the connection between locutionary act and illocutionary act is conceptual; the connection between the former and the perlocutionary act one performs is not. (Austin 1962, pp. 108-9, 120-1) With perlocutionary acts, we have to distinguish between the speaker’s perlocutionary object and the act’s perlocutionary sequel. The former is the intended effect on the hearer, the latter the actual effect ensuing. These may come apart, as when my promising to be at the party or lecture (the illocutionary act I perform) is intended to please but causes dismay.Footnote 2 (ibid., p. 120)

The problem with the PD account of the act of arguing is that it takes its perlocutionary object to be the achieving of agreement between the parties. And it counts the total speech act as non-defective (felicitous) only if that object is accomplished. But, first, perlocutionary object and perlocutionary sequel may come apart here, just as with other speech acts. And while the object of arguing on a given occasion may be the achieving of agreement, it need not be. I may give you an argument for p without caring a jot whether it will result in your agreeing that p; indeed, I may do so knowing full well that it is unlikely to. Even if my aim is to get you to agree with me, it is so, as we may say, only derivatively, only by way of your coming to see that my argument supports p. Whether the arguer or the audience regards the premises as true and thus comes to believe that the conclusion is true is another matter. (Giving an argument that some conclusion follows from or is supported by some premises without believing those premises is common in philosophy.) Suppose that I advance an argument for a certain conclusion p that I believe to be false. My aim is not to support or establish p, but only to make the case that it follows from the premises or, more modestly, that the premises support p. I may hope that achieving that aim will result in your coming to agree with me, and thus that if I think that p is true, you, too, will think so. Note, though, that my object is not your coming to agree with me simpliciter, but your doing so as a result of recognizing the cogency of my argument. That is what the PD account overlooks.

Compare the illocutionary act of warning. We claim that it is a conceptual truth that its perlocutionary object is alerting. It may fail to achieve that object; its perlocutionary sequel may or may not be that its hearer is alerted – he may be hard of hearing or not paying attention. But nothing is a warning if it is not aimed at alerting. I may, of course, utter a linguistic expression typically used to warn, such as ‘Look out!’, without performing the illocutionary act of warning. I may utter the expression precisely to teach someone that it is typically used to warn. The claim is not that there is a conceptual connection between the vehicle used to perform the illocutionary act A and a particular perlocutionary object, only that there is such a connection between the illocution and a certain perlocutionary object. A linguistic vehicle typically used to perform A may not be so used on a particular occasion.Footnote 3 In the case when it is and A is a warning, that object is alerting, whether the act’s perlocutionary sequel is the hearer’s being alerted or not. If A is an act of arguing, the illocutionary act being performed is the supporting of a conclusion. The perlocutionary object may or may not be to get you to agree, and even if it is, the perlocutionary sequel may or may not be agreement. Just as I have not failed to warn you even if you are not alerted, I have not failed to give you an argument even if you are not persuaded. Whether the argument I give you is a good one is a matter that has nothing to do with your view of it but depends on whether it supports its conclusion.

The connection between an illocutionary act and its perlocutionary object is conceptual, that between an illocutionary act and its perlocutionary sequel is contingent. Even if it is true that my warning you typically results in your being alerted (notice that in saying, as we sometimes do, that it puts you on alert, we are conflating illocution and perlocution), it need not. But I cannot be warning you without aiming to alert you. (If I utter ‘Look out!’ with any other aim, the act I perform is not a warning.) My object in performing the illocutionary act of warning when I utter ‘Look out!’ is to alert you to a danger even if you are not alerted – as noted before, you may not hear me or may not be paying attention. I have, nonetheless, warned you. (I can, of course, alert you without performing the illocutionary act of warning by my cowering or pretending to.) Illocution, perlocutionary object and perlocutionary sequel can vary independently of each other.

We have argued that there is a conceptual connection between illocutions and their perlocutionary object. The point of warning is to alert, and if things go well, one’s hearer is alerted. What analogous conceptual connection can be thought to hold on the PD account? It would have to be a connection between arguing and the perlocutionary object of achieving agreement. But there is an important disanalogy between the two cases. The perlocutionary object of an act of arguing is not, as such, to achieve agreement. It is to provide premises to one’s target (which can be oneself) that entail or support the conclusion. One may have any number of other objectives – not infrequently to prompt another act of arguing (especially if the target is a philosopher). The crucial point is that there is no conceptual connection between arguing and aiming to achieve agreement as there is between warning and intending to alert, or arguing and aiming to support a conclusion. There is no combination of illocution and perlocutionary object of the happy sort possible in the case of warning and alerting, and, indeed, of illocutionary acts in general. Of course, my perlocutionary object may be to get you to agree with me and to do so because you see that the argument I am giving is a good one. If this happens, I achieve my perlocutionary object. However, as with other acts, the perlocutionary sequel of my giving you a good argument may be quite different. This is true with all illocutionary acts. While the perlocutionary object of threatening is to frighten, I can threaten you without your being frightened (or blame you without you coming to feel guilty, and so on). Similarly with arguing. I can give you an argument intending that you be convinced by it, and my doing so does not require that you be.

Does this not mean, after all, that there is a conceptual connection between arguing and intending to convince just like the one between warning and alerting? Yes, but only if the intention is to convince by way of your recognition of the cogency of the argument I give. That is what the PD account leaves out, its repeated invocation of ‘dialectical reasonableness’ notwithstanding. (More on this below.) On that account, the perlocutionary object is your coming to agree with me regardless of your reason for doing so. You may regard me as an expert and take the mere fact of my arguing for p as sufficient for accepting p without understanding the argument. We are not saying that it is impossible for this to be one’s perlocutionary object in giving an argument – perhaps it is more often than it should be. But such cases are not cases of convincing by argument.

Our claim, then, is that the perlocutionary object of arguing is not, as PD has it, the achievement of agreement. It is, rather, the supporting of a conclusion. The perlocutionary sequel of an act of arguing may or may not be agreement, but its perlocutionary object cannot be. On the other hand, the perlocutionary object of arguing cannot fail to be the supporting of a conclusion.

As we have noted, for PD, the perlocutionary object of the speech act of arguing is to resolve a disagreement. Indeed, it may seem that for PD, resolving disagreement is the primary, perhaps the sole, function of arguing. (Biro and Siegel 2016, pp. 268-9) That, however, is not what we are challenging here. Perhaps that is the primary – though, surely, not the only – reason why people argue. And, perhaps, it is a mistake to think that argument has a function at all, as Goodwin maintains. She characterizes what she calls the function claim in this way:

  1. (a)

    The context of an argument should be conceived as a joint activity.

  2. (b)

    That joint activity has the function of achieving a social good.

  3. (c)

    The norms of argument include those rules (principles, values, standards, etc.) an argument must follow (live up to, instantiate, meet, etc.), in order for the joint activity in which it is embedded to achieve its function. (Goodwin 2007, pp. 70–71, note deleted)Footnote 4

Both Walton and Goodwin use ‘argument’ to mean an act of arguing by way of giving arguments.Footnote 5 We take no stand on whether arguments in this sense have a function. But why people argue is one thing, what the arguments they use in arguing are for is another.Footnote 6 Our claim is that it is a conceptual truth about the arguments people use in arguing that they – that is, the arguments – consist of propositions purporting to support the proposition the arguer is advocating. That is one’s goal when one gives an argument, whatever further objectives one may have.

Thus there is no way to fit the PD account into a speech-act framework.Footnote 7 The mistaken characterization of the perlocutionary object as the securing of agreement, rather than providing support for a claim, is fatal to the attempt. The underlying problem can be articulated independently of that failure. It is that there is no conceptual connection between arguing and consensus-building. There is such a connection between arguing and trying to prove (perhaps to oneself), support, or discover something. But none of these things is the same thing as trying to eliminate disagreement, even if one has that as one’s reason for engaging in arguing.

3 Dialectical Reasonableness v. Epistemic Reasonableness

van Eemeren and van Haaften mis-state what they call the crucial problem for their view in saying that it is to explain how reasonable arguments “can lead to effective convincingness” (our emphasis). If ‘effective convincingness’ means achieving agreement, and doing so is the perlocutionary object of arguing, what the theory requires is that the perlocutionary sequel be achieved only by way of reasonable argument. Thus it must show that it does not endorse unreasonable arguments that produce agreement. It must also show that it does not count reasonable arguments that do not as defective. But it shows neither of these. On the contrary, as we show next, PD counts as reasonable epistemically unreasonable arguments, and counts reasonable arguments as defective if they fail to produce agreement or resolution.

van Eemeren and van Haaften offer an account of what they call “dialectical reasonableness”, and explore the relationship between it and “rhetorical effectiveness”, which they also call “effective convincingness”. (2023, pp. 1, 3) Dialectical reasonableness as they conceive it involves a “critical conception” of reasonableness, according to which “being reasonable means observing certain standards for the tenability of a standpoint against systematic criticism.” (p. 12, emphasis in original) What are these ‘standards’? They are “problem-valid standards of reasonableness that are also acceptable to the audience.” (p. 12) And what are these ‘problem-valid standards of reasonableness’? On the critical conception of reasonableness that PD favors, they are determined by PD’s “ideal model of a regimented argumentative exchange between two parties involved in a difference of opinion exclusively directed at resolving the difference based on the merits of the argumentative moves that are made.” (p. 13) So it is the argumentative moves that are meritorious or otherwise; they are meritorious, that is, reasonable, to the extent that they are problem-valid – tend to resolve the difference of opinion at issue – and conventional-valid – are acceptable to the participants. The regimented argumentative exchange is ideally to be carried out in accordance with PD’s “procedural code of conduct”, which is intended to secure “effectiveness through reasonableness”. (p. 30)

As we have argued elsewhere, this sort of ‘move merit’ is quite distinct from the epistemic merit that in our view is the heart of argument quality, since a move might well be both problem-valid and conventional-valid yet fail to provide justificatory support for the standpoint at issue. But notice how limited the PD view of critical reasonableness is. First, it is aimed completely at resolving differences of opinion and reaching agreement. There is no room for argument as inquiry, where arguers are trying to figure things out. There is no room for arguments as case-making, where an individual arguer attempts to establish a conclusion or thesis in the absence of a disagreeing interlocutor. Perhaps most importantly, there is no concern for the epistemic quality of (the substance of) a move – if the move is problem-valid (tends towards resolution of the dispute) and conventional-valid (is agreeable to the parties), it is ‘reasonable’. The epistemic quality of the reason(s) offered in the move is neither here nor there as far as dialectical reasonableness is concerned. This is an anemic conception of reasonableness.

We might also query what the “tenability of a standpoint” comes to on the PD account of reasonableness. As is well known, PD embraces critical rationalism, according to which reasonableness is a matter of critical testing: if a proponent’s standpoint survives the criticism offered by an opponent, it is ‘reasonable’ in that it has meant relevant challenges. Unfortunately, critical rationalism is a failed account of both argumentative reasonableness in particular and epistemic normativity more generally. Suppose a given standpoint p survives such critical testing. What can we then say about p? Not that it is true, of course, but more importantly, also not that it enjoys any positive epistemic status at all. This is because there is no such thing as positive support according to critical rationalism. As Popper says, our conjectures “may survive these tests; but they can never be positively justified: they can neither be established as certainly true nor even as ‘probable’ (in the sense of the probability calculus)… None of [our theories] can be positively justified”; “there are no such things as good positive reasons”.Footnote 8 So according to critical rationalism, even given successful criticism of not-p, we have no good reason to believe or accept that p. Worse, we likewise have no good reason to think that successful criticism of not-p shows that not-p is false or that we have good reason to reject it. We do not have any good reason to reject it, despite such successful criticism, because according to critical rationalism there is no such thing as a good positive reason. A criticism of not-p, however strong it may be, gives us no good reason to regard not-p as false or to reject or withdraw it, because critical rationalism denies that there are or can be good reasons – good reasons to regard not-p as true or to regard it as false. So even when we take ourselves to have a strong criticism of not-p, this strong criticism in no way discredits it, even in principle, according to critical rationalism. As Bart Garssen puts it, “A critical regimentation based on a critical rationalist philosophy of reasonableness involves critical testing aimed at checking whether the standpoint at issue should be rejected.” (2023, p. 534) But such testing cannot, in principle, determine that a standpoint should be rejected. It cannot provide good reason to reject, because according to critical rationalism there can be no such thing as a good reason to reject. Critical rationalism forbids a critical test to have any epistemic implications whatsoever. Even a critical test in which a standpoint is thoroughly trounced provides no good reason to reject that standpoint. Critical tests are powerless to provide good reasons for rejection, just as they are powerless to provide good reasons for acceptance. This point has been familiar for quite some time, as discussion of Popper’s infamous need for ‘a whiff of inductivism’ has shown. (Siegel and Biro 2008, pp. 195–199)

This is a deep problem that PD theorists have not overcome. PD’s recognition of pro-argumentation is no help here, because pro-argumentation affords no positive justification to the standpoints that it argues for. Nor does con-argumentation afford any positive reason to reject the standpoints it argues against. Criticism has no epistemic import if it doesn’t offer positive reason to reject its target. According to critical rationalism, it cannot offer any such reason. Critical rationalism is thus a tempting but ultimately untenable theory, both of the epistemology of science (as most philosophers of science have recognized for decades)Footnote 9 and of the epistemology of argumentation.Footnote 10

van Eemeren and van Haaften may protest, insisting that the epistemic merits (or otherwise) of an argumentative move are reflected in their account, since the moves are “directed at resolving the difference based on the merits of the argumentative moves that are made.”Footnote 11 But what are these ‘merits’? Are they utterances that assert their own epistemic quality, along the lines of ‘here’s my reason for saying that we should resolve this difference of opinion in my favor, because my reason is epistemically determinative – that is, it justifies my view’? This would obviously not work, since saying so does not make it so, and a dialogical partner may simply disagree that it is so determinative. More to the point, even when that partner agrees that the reason is a good one, it may not be, since people are not infrequently convinced by bad reasons and arguments. Are these then reasons and arguments that are actually meritorious, that is, that afford justification to the claim or standpoint at issue? If so, their merit is independent of the problem- and conventional-validity of the move. Rather, that merit is strictly epistemic.

van Eemeren and van Haaften introduce the “Principle of Reasonableness” (p. 12; cf. van Eemeren and Grootendorst 2004, p. 2), and opt for their ‘critical’ conception of reasonableness, as we have seen. They emphasize their pragmatic, dialectical focus:

Whatever conception of reasonableness is adopted, in real-life situations a difference can only be resolved on the basis of argumentation if the argumentation advanced is in agreement with problem-valid standards of reasonableness that are also acceptable to the audience. This means that the argumentation not only needs to be reasonable according to the prevailing standards of reasonableness, but also acceptable to (and accepted by) the intended audience – and therefore potentially effective. To put it more strongly, in order to fulfill its purpose adequately the argumentation has to be effective on the basis of its reasonableness. (pp. 12–13, notes deleted)

We applaud PD’s focus on resolving actual disagreements in practicable and reasonable ways. But, as we have argued above and elsewhere, effective convincingness need not, and often does not, track epistemic quality, and their conceiving of reasonableness in critical rationalist terms is problematic. (Siegel and Biro 2008, 2010; Lumer 2010Footnote 12) Perhaps more fundamentally, PD’s preoccupation with dialectic and dialectical resolutions, while salutary in its efforts to resolve differences of opinion in reasonable ways in actual argumentative exchanges, leaves out the heart of argumentation by ignoring the fact that argumentation involves arguments, whose quality is fundamentally and conceptually an epistemic matter, independent of their ability to resolve disagreements. In so far as PD restricts itself to the study of dialectical exchanges, it is limited in a way that argumentation theory should not be, because it rules out of bounds the study of both non-dialectically instantiated arguments, and the quality of arguments as they occur in dialectical exchanges considered independently of their dialectical role. In particular, their epistemic quality is ignored. But that quality is at the heart of argumentation and thus should be central to argumentation theory. Argument quality is not solely a dialectical matter. (Siegel 2023) van Eemeren and van Haaften insist that “…the rationale for advancing argumentation is to resolve a difference of opinion” (p. 31). But, first, this is not its only rationale, and a theory that focuses only on such resolution misses out on other, central, rationales. Second, even when the goal is dialectical, dialectical virtues do not line up with epistemic ones.

This omission infects PD’s conception of argumentation itself, and not just its conception of reasonableness: “Argumentation is by definition aimed at reaching the interactional effect of having a standpoint accepted by a rational judge who judges reasonably.” (p. 24, note 97) This is PD’s definition, not the definition, of ‘argumentation’. Argumentation also includes inquirers and individuals trying to figure things out non-dialectically; on the PD view these do not count as argumentation at all, though they manifestly do. What makes an interaction an instance of argumentation is not that it has the aim of having a standpoint accepted, by a rational judge or anyone else, but rather that it aims to support a conclusion, whether accepted or not. This is accomplished, when it is, by means of adducing arguments, and the quality of such arguments is a function of their ability to provide epistemic support for their conclusions, not their ability to achieve effective convincingness. (Biro and Siegel 1992, 1997; Siegel 2023)

van Eemeren and van Haaften repeat PD’s understanding and use of the terms ‘rationality’ and ‘reasonableness’, also stated in van Eemeren and Grootendorst 2004, p. 124:

Like in ordinary language, in our terminology rationality refers to using one’s faculty of reason and reasonableness to doing so appropriately, i.e. in a way that is suitable for dealing with the problem at issue. “Acting reasonably” therefore has a more specific meaning in pragma-dialectical parlance than “acting rationally”: it means acting rationally in a way suitable for resolving a difference of opinion. (p. 18, emphases in original, notes deleted)

Leaving aside the questionable remark concerning ordinary language – if I use my ‘faculty of reason’ inappropriately, I have not acted rationally – their usage of these terms makes clear, again, that acts such as speech acts in a critical discussion qualify as reasonable just insofar as they conduce to the resolution of differences of opinion, thus emphasizing again PD’s limiting commitment to understanding ‘argumentation’ strictly dialectically. We have already noted that ‘on the merits’ does not help, since those merits involve problem- (and conventional-) validity, which may well fail to secure epistemic rationality or reasonableness in cases in which the reasons offered in the moves made (acts performed) fail to afford justification to the thesis, conclusion or standpoint at issue, despite their problem- (and conventional-) validity. As we have repeatedly noted, it is not uncommon for good arguments – in the epistemic sense we have emphasized – to fail to convince. Whatever other account of reasonableness one adopts, one must recognize that arguments can be bad in that sense. Alas, it is just as common for people to be convinced by bad arguments. What should we say in the latter case? To return to the speech act framework: we have the perlocutionary sequel, thus the perlocutionary object as understood by PD has been achieved. The argumentation has thus been effective. But it is nonetheless defective, since it consists of arguments that are bad, ex hypothesi. But PD cannot say this, since the argumentation is problem- and conventional-valid. Because it has these properties, it is, by PD’s lights, reasonable. But it is not. If using bad arguments to convince people is sufficient to render an act of arguing defective (as it surely is!), it makes no difference whether those bad arguments are effective or not. And if the arguments used in arguing are good ones, why say that the acts of arguing are defective if they fail to convince?

Being effective is neither necessary nor sufficient for an argument to be reasonable – even in the ‘dialectical’ sense – and its being reasonable is neither necessary nor sufficient for it to be effective. PD must be understood as claiming that an act of arguing is non-defective only if the arguments deployed are in fact reasonable (in whatever sense) and they are in fact effective. But this is plainly counterintuitive. How can my act of arguing be deemed defective because of your obtuseness?

To be clear: we applaud PD’s quest for ‘reasonable convincingness’, that is, argumentation that is effective at convincing one’s interlocutor because it is reasonable. If ‘reasonable’ here means ‘on the basis of strong, epistemically forceful reasons’, we agree. Our objection concerns not the quest for reasonable convincingness, but rather PD’s understanding of ‘reasonable’, and the account of argumentation in which it is embedded. The pragmatic, instrumental, strictly dialectical account, according to which a move is reasonable because it is problem- and conventional-valid, simply leaves out of consideration the epistemic quality of the reasons and arguments advanced. Following the rules of a regimented critical discussion is no guarantee of epistemic quality.

We hope we have made clear how PD’s mistaken identification of the perlocutionary object of arguments infects its conception of reasonableness. Whether an argument supports its conclusion is not determined by whether the arguing parties come to agree whether it does. My aim in giving you an argument that I think does provide support for some proposition is to get you to see that it does. I may, of course, be mistaken in thinking that it does, and you may try to show me that I am. But which of us is right does not depend on our coming to agree one way or the other. Mis-identifying the perlocutionary object as achieving agreement leads inexorably to misconstruing ‘reasonableness’ as involving agreement rather than epistemic support.