Journal of Social Ontology Article Discussion: Garcia-Carpintero, “Conventions and Constitutive Norms”

Welcome to the first of what will hopefully be many productive discussions of articles in the Journal of Social Ontology at PEA Soup. The target article is Manuel Garcia-Carpintero’s “Conventions and Constitutive Norms”, available open access, with a brief consideration of the main points of Garcia-Carpintero’s article by Brian Ball (New College of the Humanities, London). Although the focus of Garcia-Carpintero’s article, more narrowly construed, is a debate in the philosophy of language about the nature of the norms governing assertion, the ultimate goal of Garcia-Carpintero’s (and Ball’s) discussion is “to get a better understanding of what social constructs conceived as defined by constitutive norms are”.

Garcia-Carpintero on ‘Conventions and Constitutive Norms’

Brian Ball, New College of the Humanities, London

What is the nature of assertion? A long-standing debate in the philosophy of language pits those who, following Austin (1962), wish to characterize this and other speech acts in terms of rules governing their performance against theorists who follow Grice (1957) in appealing instead to the psychological states of speakers when engaging in communication. As Manuel Garcia-Carpintero notes in his rich and interesting paper ‘Conventions and Constitutive Norms’, after Gricean views dominated in the late 20th century, the debate was reinvigorated for the 21st century by Timothy Williamson’s influential (2000) view that assertion is individuated by the knowledge rule, according to which one must assert that p only if one knows that p. Yet whereas Austin held that speech acts are governed by conventions, Williamson gave no immediate succour to conventionalists, holding that the knowledge rule is a constitutive (not conventional) norm of assertion. Garcia-Carpintero quotes from Williamson:

Constitutive rules are not conventions. If it is a convention that one must φ, then it is contingent that one must φ….In contrast, if it is a constitutive rule that one must φ, then it is necessary that one must φ. (Williamson, 2000: 239)

And he comments: ‘Williamson’s argument is either misleading, or a straightforward non-sequitur’ (p.37). Indeed, the central aim of Garcia-Carpintero’s paper is to argue that ‘the modal difference between conventions and constitutive norms… is inapt to cast doubt on forms of conventionalism [about assertion] worth considering.’ (p.37) I agree with this central thesis. But my reasons for doing so are different than Garcia-Carpintero’s – and it will, I hope, be valuable to consider why.

Garcia-Carpintero gives ‘a reconstruction of [Williamson’s] argument’ (p.36) above, which he calls the ‘Argument of Modal Disparity (AMD)’ (slightly abridged here):

1. Assertion is a kind defined by constitutive norms.
2. Constitutive norms impose their obligations in all possible worlds.
3. Conventions… do not….
∴ The assertoric practice is not conventional…. (p.36)

In my view, however, AMD does not capture the argument given in the passage from Williamson quoted above. Williamson opens that passage with his conclusion; and he then gives his premises. Let’s begin with the latter. These have the form of conditionals: we might formalize the first as if p then r and the second as if q then s. The conclusion is the claim that constitutive rules are not conventions. But to connect this with the premises, we might reformulate it as the claim that if it is a constitutive rule that one must φ, then it is not a convention that one must φ – or in our scheme, if q then not p. This does not follow from the stated premises alone, but there is an obvious missing premise: it is not both contingent and necessary that one must φ; that is, not both r and s. Construed in this (supplemented) way, Williamson’s argument is logically valid.

Garcia-Carpintero concedes that ‘[t]here is an interpretation of the conclusion of Williamson’s argument… on which nobody should reject it’ (p.36). Perhaps the above captures this interpretation, and the (sound) reasoning behind it. The argument might still mislead people into supposing that the conclusion conflicts with conventionalism about assertion, as in the reconstructed argument AMD. Now, as Garcia-Carpintero says, ‘there is no point in speculating about the actual intentions of philosophers deploying AMD’ (p.48), and in particular, about whether Williamson was attempting to show that Austinian conventionalism about assertion is mistaken. But Garcia-Carpintero does say that I ‘produce[d] a straightforward version of AMD’ (p.38) in my (2014) Mind and Language paper, and ‘argu[ed] that assertion is not conventional on that basis’ (p.38). This is something I can speak to. In fact, I only wanted to show that the knowledge rule is not itself a convention and was reticent about claiming more. John Searle had suggested that ‘[d]ifferent human languages… can be regarded as different conventional realizations of the same underlying rules’ (1969: 39) governing speech acts. I accordingly wondered whether Williamson’s knowledge rule might be realized in a language L1 as the rule that one must utter s1 in manner m1 only if one knows that p, whereas it was realized in L2 as the rule that one must utter s2 in manner m2 only if one knows that p. I thought this would be so if there were a convention, in communities using L1, that uttering s1 in manner m1 sufficed for asserting that p; and similarly for L2. Being uncertain whether there were such conventional realizations of the knowledge rule (in which the manner of uttering is specified in formal rather than intentional terms), I thought it best to avoid pronouncing on the question of whether speech acts are ‘essentially conventional’ (Ball, 2014: 341) in this sense. Thus, as noted, I agree with the central thesis of Garcia-Carpintero’s paper: for all the modal considerations adduced by Williamson show, some form of conventionalism about assertion might be correct.

However, Garcia-Carpintero’s reasons for denying that considerations of modal disparity give grounds for rejecting conventionalism about assertion are not those I adduced above. For one thing, he is concerned with AMD – and although I have suggested it is not a faithful rendering of Williamson’s argument, it is nonetheless an interesting argument worthy of consideration in its own right. For another, he gives a different characterization of conventionalism than I have done. Starting with the second of these points, here is what Garcia-Carpintero takes to be ‘a sensible form of C(onventionalism) about A(ssertion)’ (p.37):

CA The practice of assertion exists (so that speakers are in fact bound by its constitutive norms) only because a convention instituting and preserving it is in place. (p.37)

Two things are noteworthy. First, CA concerns whether conventions are necessary for performing assertions, whereas I was concerned above with the question of their sufficiency for doing so. For what it is worth, CA aligns more neatly with Searle’s (1969: 38) question, ‘Must there be some conventions or other… in order that one can perform illocutionary acts?’ But questions of necessity and of sufficiency both strike me as worthy of further consideration.

Second, there is talk in CA of the practice – and not just the speech act – of assertion. This is key. Central to Garcia-Carpintero’s paper is ‘the distinction between… blueprints, mere putative normative kinds, and truly normative kinds, whose defining norms have been enforced’ (p.37). Applied to the case at hand, this amounts to a distinction between (as I will put it) the act of assertion, and the practice of performing that act. Garcia-Carpintero regards the former ‘blueprint’ as defined by a constitutive norm such as the knowledge rule; but, he says, ‘constitutive norms thus understood [as defining blueprints] are not really normative’ (p.37).
The distinction may shed light on Garcia-Carpintero’s assessment of AMD. He says:

[w]hen taken as addressed to reject a sensible conventionalism about assertion worth taking seriously [such as CA], the second premise in AMD is false. (p.37)

That is, on Garcia-Carpintero’s view, constitutive norms do not impose their obligations in all possible worlds. In fact, the rules specifying blueprints for speech acts do not themselves impose norms at all. Only if certain further conditions are in place – conditions which may or may not obtain, as a matter of contingency – do the rules in question get any normative bite. Thus, the fact that conventions admit of alternatives, and therefore only impose obligations contingently, does not show that CA is false. Nevertheless, we can now see one reason why Garcia-Carpintero might have spoken of a non-sequitur in his reconstruction of Williamson’s argument, AMD. For arguably there is a simple equivocation involved: we can get true readings of all the premises of AMD, one might think, but only if we interpret premise 2 as concerned with the blueprint for the act of assertion, while premise 3 deals with the practice of performing assertions that is mentioned in the conclusion. Once we focus unequivocally on the practices rather than the blueprints, however, we see that premise 2 is false.

If this interpretation is correct, Garcia-Carpintero thinks there are two entities involved when people make assertions: a blueprint act type; and a practice which involves acting in a way that is sensitive to that blueprint. The former strictly speaking imposes no obligations; the latter imposes obligations contingently – and this is consistent with its doing so only because certain conventions are in place (as per CA). This is a very interesting suggestion. Williamson, of course, holds that, just as a population might speak slightly different languages at different times (cf. Lewis, 1975), so too they might play slightly different games, or engage in slightly different speech acts at different times. But Garcia-Carpintero draws an analogy with the case of soccer, pointing out that there have been changes in the official rules of over the history of Association Football, and suggesting that there is something that has persisted through these changes. Similarly, then, we might expect that, even if we conformed to different blueprints at different times, one and the same practice might persist through these changes.

Let me develop the analogy a little further. David Papineau (2017) has claimed that sports and games are not the same, since some sports (e.g. running) involve no game, and some games (e.g. chess) involve no sport. If that’s right, one might take games to be defined by their rules, and sports to be, as Papineau suggests, more intimately related to excellence in the exercise of physical skills – in which case sports might be individuated by their histories. Thus, engaging in one and the same sport – soccer – might involve playing different games at different times. Suppose, then, that the practice of assertion is (similarly) defined by its history. And suppose that it has hitherto involved performing acts conforming to the blueprint specified by the knowledge rule. Could it not evolve in such a way as to involve performing acts conforming to some other blueprint? If it could, then it is not essential to that practice that participants in it are bound by the knowledge rule, and the obligation to assert only if one knows would be contingent, consistent with its being conventional. Still, it is not obvious that such a change would not amount to a kind of speciation: perhaps a new practice would come into being were such a shift to occur. If that’s right, its not clear we need two entities, blueprint and practice (or if there are two things, that they can come apart in the required manner): a practice requiring, say, that we assert only if we reasonably believe ourselves to know would be distinct from ours; in which case, it is essential to our practice that it involves conforming to the blueprint specified by the knowledge rule after all.

But perhaps this goes a little too far. Some of what Garcia-Carpintero says suggests not that he thinks there are two distinct entities, blueprint and practice, but only that blueprints as such impose no obligations. Thus, arguably, there is just one entity – the blueprint; but there are two kinds of blueprints – those whose rules have normative force, because some group imposes the requirement to conform, and those whose rules do not. This, I think, raises the following substantive question in social ontology: Is any blueprint (speech act type) a genuinely normative kind (or practice)?

A related question is why exactly blueprints don’t themselves have normative force (as such). Take the speech act of assertion as individuated by Williamson’s knowledge rule. Obviously, there are possible worlds in which no one makes an assertion (and there is no practice of doing so). But is it not true, even in such worlds, that subjects must assert only if they know? (Put another way: is it false? If so, what falsifies it?) On the face of it, the knowledge rule involves a deontic modal taking wide scope over a conditional: what is obligatory is that one asserts only if one knows; or in other words, that if one asserts then one knows. This condition is trivially fulfilled by anyone who does not assert. But why, according to Garcia-Carpintero, is there no genuine obligation here – no normative force? Is it that no obligation to know can be detached? That seems insufficient: after all, in the actual world I might refrain from asserting precisely because (I know that) I don’t know; and in such a case, the knowledge rule has normative force, though I incur no obligation to know.

It is time to wrap up. As we have seen, Garcia-Carpintero’s central thesis is that some kind of conventionalism about assertion is compatible with the fact that constitutive rules are necessary while conventional norms are contingent. I agree. But Garcia-Carpintero is keen to stress that none of this means that he endorses conventionalism: in fact, he says, CA is ‘wrong… in my view’ (p.37); but ‘the true reason has nothing to do with modal disparities’ (p.49). Instead, he finds it hard to see how the fact that a norm is in force ‘whose point is ultimately to allow for testimonial knowledge’ (p.49) could be a matter of arbitrary convention. Here too, I agree. And Garcia-Carpintero’s careful exploration of the issues here will no doubt ensure, just as he hopes, that we will be able ‘to get a better understanding of what social constructs conceived as defined by constitutive norms are’ (p.37).

References
Austin (1962). How to Do Things with Words, Clarendon Press: Oxford.
Ball (2014). ‘Speech Acts: Natural or Normative Kinds?’, Mind and Language 29(3): 336-350.
Garcia-Carpintero (2019). ‘Conventions and Constitutive Norms’, Journal of Social Ontology, 5(1): 35-52.
Grice (1957). ‘Meaning’, Philosophical Review 66(3): 377-388.
Lewis (1975). ‘Languages and Language’, in Gunderson (ed.), Minnesota Studies in the Philosophy of Science, University of Minnesota Press, pp.3-35.
Papineau (2017). Knowing the Score: How Sport Teaches Us About Philosophy (and Philosophy About Sport), Constable: London.
Searle (1969). Speech Acts: an Essay in the Philosophy of Language, CUP: Cambridge.
Williamson (2000). Knowledge and Its Limits, OUP: Oxford.