By In Featured Philosophers, Metaethics, Value Theory Comments (23)

Expressivism without Minimalism

By Tristram McPherson

It is a striking fact that many of the most recently influential expressivists (e.g. Simon Blackburn, Allan Gibbard, Mark Timmons) have embraced minimalist accounts of words such as ‘truth,’ ‘fact,’ and ‘property.’ And others have argued that embracing minimalism is indispensable for the expressivist. In this post, I argue that expressivists can and should resist the idea that they are forced to embrace minimalism.

On first approximation, expressivism claims that indicative normative sentences semantically express psychological states that are broadly desire-like, where indicative non-normative sentences semantically express beliefs. Roughly, minimalism is the thesis that the meaning of certain words is exhausted by certain equivalences, which deprive these words of the distinctive metaphysical significance they are sometimes taken to have.

Some expressivists embrace minimalism simply because they are independently convinced of its truth (e.g. Gibbard on truth in Thinking How to Live). However, others have suggested that – if we are conceiving of expressivism as a theory of our actual thought and talk – embracing minimalism is crucial to the viability of the expressivist program. The most prominent reason for thinking this is that it is seemingly obviously felicitous to (e.g.) reply to a normative claim by saying “That’s true,” or “That’s a fact.” And it may seem that embracing minimalism is the only plausible way for the expressivist to vindicate this apparent felicity. (It is worth flagging here that some hybrid expressivists appear capable of vindicating the felicity of such claims without committing to minimalism, in virtue of the belief-like part of the hybrid state they posit. I focus on non-hybrid views here.)

The expressivist’s alleged need to embrace minimalism has been strongly endorsed. Here’s James Dreier: “To save the phenomena, then, expressivism needs a conception of truth that doesn’t commit to anything metaphysically heavy-duty.” (“When do Goals Explain the Norms that Advance Them?” 170). And Huw Price argues on very similar grounds that an anti-realist alternative to adopting minimalistic resources would be a reductio of expressivism as a hermeneutic thesis (“From Quasi-realism to Global Expressivism – and Back Again?” §1.2). And many other philosophers now appear to take for granted that any plausible expressivism must be wedded to minimalism.

If this were so, it would be Bad News for (non-hybrid) expressivism. The problems arising are diverse; here I gesture at the simplest one to state. Embracing minimalism saddles the expressivist with controversial commitments about a topic other than the one she hopes to theorize, and the philosophy of truth is hair-raisingly tricky. Crudely, then, if I have to multiply my theory- of-truth-independent credences in expressivism by my credences in a controversial claim about truth, the resulting credence in expressivism will be tiny. This gives the expressivist reason to resist the idea that felicity data forces them to embrace minimalism. I propose a means of resisting in two stages.

The first stage begins by observing that normative thought and talk present philosophers with a hard interpretive problem. What do I mean by this? It is not difficult to get competent users of normative concepts puzzled by the problem of how to interpret normative thought and talk. And once puzzled, they reach for a wide variety of different solutions. The same is true of the experts: for several generations, philosophers have been systematically investigating normative thought and talk, without convergence.

This observation is compatible with the possibility that the correct interpretation of normative thought and talk vindicates everything we are ordinarily inclined to accept about the topic, in some highly non-obvious way. However, the difficulty of the interpretive problem makes it more plausible that there is no such interpretation. It is thus plausible that some of what we are inclined to accept about normative thought and talk is in error. And this motivates the search for principled theories of error concerning initially plausible features of normative thought and talk.

The second stage of my proposal sketches such a theory of error for the apparent felicity data for ‘true’. To begin, consider:

T-Schema     For every well-formed indicative sentence ‘S,’ ‘S’ is true just in case S.

The standard minimalist idea applied to truth is that T-Schema tells us everything that we need to know about both the meaning of ‘true’ and the nature of truth (cf. Terence Cuneo, “Properties for Nothing, Facts for Free?”, 231). One problem for the most straightforward forms of minimalism is that T-Schema is probably false. For example, there are paradoxes associated with ‘Liar’ sentences (e.g. “This sentence is not true”). And many philosophers have proposed that the false premises that generate the paradoxes are instances of T-Schema.

This diagnosis of the Liar paradoxes invites a plausible theory of error. This theory suggests that what Matti Eklund (“Inconsistent Languages”) calls the “pull” of the Liar paradoxes is explained by the fact that a disposition to accept any instance of T-Schema is – non-accidentally – correlated with competence with ‘true.’ Consider a (sketchy) conjecture about how that disposition is acquired. We typically learn to use terms felicitously by encountering others’ presumptively felicitous uses of those terms, and by getting felicity feedback on our own utterances. While doing so, we form something like an implicit model of how to use the relevant terms felicitously (perhaps aided by an innate interpretive framework that narrows the range of candidate hypotheses). For those of us who do not wrestle with how to interpret Liar sentences at an early age, T-Schema is plausibly part of the most elegant model to acquire for ‘true.’ If this is right, the process by which almost all of us become competent with ‘true’ likely also disposes us to accept arbitrary instances of T-Schema, even when they are false.

This discussion of Liar sentences is useful in two ways. First, it shows that we can offer plausible theories of error for felicity judgments: we should not simply assume that these are the least vulnerable of our apparent metaethical evidence. Second, it provides a model for the theory of error I propose on behalf of the expressivist.

To begin developing this theory, consider a familiar and compelling motivation for expressivism: normative thought and talk is functionally powerful as a tool for (intrapersonal) planning and (interpersonal) coordination. The expressivist suggests that non-cognitive states are the crucial psychological mechanisms for such a tool: we want our planning and coordination to issue smoothly in action, and non-cognitive states are functionally crucial for the production of action.

In order for normative thought and talk to play these roles in planning and coordination, there arguably must be structurally rational relations among normative thoughts, and implication relations among normative sentences. If thinking I ought to resist racism were not in rational tension with thinking It is not the case that I ought to resist racism, then it would be hard to see how I could reason my way to conclusions about what I ought to do. Similar points apply to implication and discourse. (It is controversial whether we can explain inferential and implication relations among the relevant non-cognitive states. But if we cannot, expressivism is hopeless for reasons that have nothing to do with minimalism, so I set this aside here.)

In non-normative discourse, aptness for inferential and implication relations is tightly correlated with indicative syntactic form (where this form differs from, e.g., the typical syntax of interrogatives or imperatives). If normative talk – as the expressivist understands it – in fact arose as a linguistic vehicle to enable expression of inferential and implication relations, it did so by adopting the same indicative syntactic form. In non-normative sentences, indicative form is highly correlated with truth-aptness. A flat-footed explanation of this is that non-normative indicative sentences function to describe the world. But a universal generalization on the correlation between indicative form and truth-aptness would be part of the most elegant implicit model for the felicity of ‘true’ that we would expect language-learners to develop as they acquire English. The expressivist can propose that, just like the intuitive applicability of the T-Schema to Liar sentences, the applicability of the T-Schema to normative sentences is an elegant overgeneralization that we would expect users to come to implicitly accept as they become competent with a natural language. Even if normative sentences are not truth-apt, then, this conjectured account can explain why being disposed to take them to be truth-apt is non-accidentally correlated with having become competent with ‘true’ in the ordinary way.

It may seem that, in presenting this theory of error, I have exhibited the very vice that I criticized above: I seem to have committed the expressivist to a contentious theory about some ‘truth’ talk. However, my aim here is not to propose that the expressivist should adopt the theory of error sketched here. Rather, the sketch illustrates the credibility of the broad strategy of challenging the probative force of the relevant felicity judgments. This in turn motivates the idea that minimalism is not the only game in town for expressivists: they can afford to be uncommitted about the theory of truth, which is the attitude they should have hoped to be able to adopt in the first place.

Note: this post contains elements of a paper-in-progress with the same name; feel very welcome to email me for a draft, but let’s keep discussion here focused on what’s in the post

Tristram McPherson, Ohio State, dr.tristram@gmail.com

 

23 Responses to Expressivism without Minimalism

  1. Jack Woods says:

    I like this quite a bit; it reminds me of discussions in Spreading the Word about why we favor cognitivist interpretations. I also like the emphasis on hermeneutic interpretations of ordinary normative discourse (as I better given that I’m in print endorsing this kind of approach.)

    My main worry would be explaining why it’s only in certain normative domains that we’re inclined to make this mistake. Ordinary folk are reluctant to ascribe truth to judgments of opinion, judgments of taste, etc, except in the flat ”reiteration” sense of truth. It would be nice to have an explanation of why this is given that these attitudes play a similar role in guiding behavior, etc.

  2. Tristram McPherson says:

    Nice point, Jack.

    It would be nice to explain that contrast.

    Here’s a really crude go at it: you might think that judgments of taste e.g. do not present the same sort of a “hard interpretive problem:” e.g. almost no one is advocating a strongly invariantist descriptivist etc. treatment of ‘tasty.’ Now suppose the elegant model is associated with some proto-realist assumptions. The contrast between taste discourse and the paradigms of truth might be striking enough to lead the folk to caution about truth-ascription there, while the muddier metaethical waters might do less to defeat the elegant generalization. (But there again, I think that if you ask some of the folk point blank: do you think there are moral truths? You will get mixed answers and lots of hesitation. So I think we can imagine a certain amount of the defeat going on there as well.

    On the other hand, it is worth noting that this would be an awkward move for the minimalist expressivist: hesitation about truth-acriptions in the case of taste looks like evidence against the idea that we are just using ‘true’ to track indicative form, e.g., as a means of expressive convenience.

  3. Jack Woods says:

    Right, yeah, I agree about the muddiness of the contrast and that the sort of sketch you just gave is one way to go; it might be aided by noting that we folk-slogan claims about the non-objectivity of taste, etc. whereas it’s far less common to do so for moral judgments, etc. It might also be aided by noting that criticism of others, which correlates with the common-concept non-relativity of truth, is a stronger piece of the central role of moral judgment than it is judgments of opinion or taste.

    I would worry a bit about saddling the minimalist picture with the idea that truth is a device of generalization of sentences with indicative grammatical form (as your second point suggests). But that’s a bit afield of the general discussion.

  4. Tristram McPherson says:

    Jack –

    I don’t actually think your second point is far afield: I think it is worth sharply distinguishing three things:

    (a) theories of truth that are compatible with there being normative truths, given pure expressivism (this is what Dreier is after in the quote in the post)
    (b) theories of truth that are in some sense ‘deflationary’
    (c) theories of truth which take the t-schema to in some relevant sense tell us everything we need to know about truth (these are the ones I am calling ‘minimalist’, and here there is a tie to indicative form).

    I think these come apart in all kinds of ways: for example, a coherence theory of truth seems not at all deflationary to me, but at least some coherence theories might play well with expressivism. And we can imagine theories that are ‘deflationary’ in a natural sense, but do not play well with expressivism.

    This is relevant, because I think that, given the broader dialectical campaign I an suggesting for the expressivist, it would be healthy to carefully map just which theories of truth fall into (a).

    Once we have that in hand, then even setting aside the sort of theory of error strategy I suggest here, no one should think that e.g. the viability of expressivism rests on ‘solving’ the creeping minimalism problem.

    And more broadly, the more ways the chips can fall truth-wise, compatible with expressivism, the better for the credibility of expressivism.

  5. Jack Woods says:

    I see. I suspect (c) subdivides, but that’s fine; you can just restrict to [the t-schema to in some relevant sense making use only of the grammatical form of a sentence tell us everything we need to know about truth of that sentence] or something like that.

  6. Jamie Dreier says:

    I find this interesting, and kind of plausible, but there is something weird about the dialectic.

    I think you are suggesting, Tristram, that ordinary speakers in effect think that ‘true’ is governed by minimalist principles, that the minimalist theory is, uh, true of their language (without being able to articulate this, of course). But, probably, you think, they are wrong, because minimalism has some serious problems.
    But then don’t you have an error theory of truth? And shouldn’t someone who likes expressionism and doesn’t want to have an error theory of normative language feel pretty much the same way about truth talk?
    In a way my point is this. Minimalism about truth and expressionism about normative language have a very close affinity; it isn’t just that the latter needs the former. They are, I think, going to be separated only painfully.

  7. Tristram McPherson says:

    Hi Jamie!

    So, first, I want to cleanly distinguish a theory of error – the sort of thing that purports to explain some systematic pattern of error in belief or inference – from an error theory of a term T, which I am thinking of as imputing a false belief to anyone who is a competent speaker with a term T. I am attempting to develop a theory of error that is not an error theory.

    As I am thinking of the picture I am sketching
    (i) when ordinary speakers use ‘true’ in sentences like “It is true that water is wet” they make no error at all
    (ii) a theorist of truth (or an expressivist) could raise their children not to be disposed to make the incorrect attributions of ‘true’ to e.g. liar sentences or normative sentences. And these children could thereby be completely competent users of ‘true’

    There is no error theory here, because the rules for correct use of ‘true’ do not require accepting the incorrect claims. That’s the point dramatized by the truth-theorist’s child.

    I think that the picture I am sketching is compatible with a lack of an error theory on a variety of the most plausible metasemantic theories. For example, it seems true on a radical interpretation picture, where, presumably, there are going to be strong reasons not to attribute paradox-generating meanings. It also seems true on externalist pictures; e.g. a teleosemantics might take correspondence to the facts as being the thing which is the sustaining explanation for truth-talk, even if the practice generates lots of ‘false positives.’ I am inclined to think, more strongly, that it is a plausibility constraint on a metasemantic theory that it allows the sort of possibility that i am imagining here.

  8. Tristram McPherson says:

    More later; for now I need to stack some firewood and then go to dinner!

  9. Frank Jackson says:

    Tristram, you mention liar sentences. You might also mention indicative conditionals. A perfectly respectable view about them is that they lack truth conditions. As it happens it isn’t the view I most like about them, but plausibly a supporter of this view doesn’t have to embrace minimalism about truth to make it fly. They can say “Sometimes philosophy finds out something surprising. An example is that despite the fact we talk of ‘If it rains, the match will be cancelled’ as being true or false, this is a mistake.” Of course they should then go to explain why the result isn’t as damaging as one might have thought. Mutatis mutandis for expressionism in ethics one might say. Indeed maybe this is just the sort of thing you are saying.

  10. Jamie Dreier says:

    I appear to have posted my first blog comment ever on Expressionism. Either I have an overzealous autocorrect or a subconscious interest in Schoenberg; I see that the same is true of Frank Jackson.
    Well, just in case: I foreswear Tristram’s salvation of substantial truth, his dangerous romanticism, and I embrace the New Atonal Pragmatism!

  11. Jamie Dreier says:

    (I do get your distinction, anyway — I’ll say more later.)

  12. Derek Baker says:

    Hey Tristram,

    I like this a lot. I also think you might be right that truth minimalism shouldn’t be essential to expressivism, and there might be good reason to deny it. At the same time, I feel like one of the particular solutions Jack and I offered for the expressivist does depend on normative claims being apt for being called true or false. So I want to push back a little, or at least indicate a worry.

    One concern with expressivism is that it can’t distinguish narrowly semantic inconsistency from pragmatic inconsistency. That this is a worry shouldn’t be surprising, either, given that contemporary expressivism seems motivated by a “meaning is use” kind of picture. If that’s your starting point, separating out semantic properties from what’s communicated will be tricky.

    The standard example, from Mark van Roojen, is the Moore-paradoxical sentence, “It’s raining but I don’t believe it’s raining.” So there is obviously rational tension between the two conjuncts, but we don’t want to say that this is a literal contradiction. Moore’s paradox just is that the sentence sounds self-contradictory, even though it’s perfectly possible that both conjuncts are true.

    Putting aside how you might solve this for the expressivist, my worry is more, how do you even state the problem? Obviously you state it, “How does the expressivist distinguish pragmatic from semantic inconsistency?” But “semantic” and “pragmatic” (in this context) are very learned, technical terms. How do we state the problem to someone who isn’t sure exactly what these terms mean (like me, for example)? The most natural way of stating the problem is that some claims are in rational conflict with each other even though they could both be true at the same time, and so how do we avoid conflating that kind of inconsistency with the kind that holds between sentences that can’t both be true at the same time? There might be other ways to do this, but they aren’t obvious.

    Anyway, my main worry is that if the expressivist denies that normative claims can be true or false, it’s hard to identify in general terms the types of conflicts between normative sentences that the expressivist needs to avoid conflating.

  13. Tristram McPherson says:

    Thanks Frank! This is really helpful.

    I am a rank amateur on conditionals, so I am hesitant to appeal to claims about them without doing a bunch more work, but if I understand the literature correctly, this may be an even better illustration of the point that widespread folk intuitions regarding whether certain linguistic forms are truth-apt are not sacrosanct. As I understand it, the core of the case against truth conditions for indicative conditionals is that there is *no consistent set of truth conditions that can be assigned to such sentences that capture the data about which such conditionals are assertable etc. Of course, this argument is controversial. But the form of the argument seems to me wholly legitimate.

  14. Tristram McPherson says:

    Derek: this is interesting; I worry I am missing what you are driving at, though.

    Let me restate the problem you state, to make if I understand it. The task is to provide an account of something like the *implication relation, right? And then we want to provide an account of both what we might call ‘logical’ and ‘analytic’ implications. But the challenge is how to give a naive gloss on this.

    What is wrong with the following super-naive idea: the logical implications of a set of sentences are are implications of those sentences that can be derived by a speaker (considering the sentences themselves independently of anyone accepting them) who has perfectly mastered the rules governing the logical vocabulary of the language.

    For example, I can notice that “P,” independently of my accepting P, is consistent with “I disbelieve that P”, so it is pretty easy for speakers to consider what follows from sentences independent of one’s accepting them; we do it all the time.

    And then the for analytic entailments, just the same idea with ‘semantic rules’ added.

    Now, I want to emphasize that I am being deliberately crude here. I think the overall story of how to explain semantic entailments, if there are any, much less how to do that for the expressivist, is hard. But so far I am not seeing what special role truth might play in helping to explain. And indeed, if truth is a super-minimal notion, it is hard to see how truth could play an ineliminable explanatory role in this context. But maybe I am missing the force of the worry somehow.

  15. Tristram McPherson says:

    In case it is not clear, the bit in the last post about considering P independently of anyone accepting it is what is supposed to do the work of distinguishing the relevant implications from Moore-inconsistencies

  16. Derek Baker says:

    Hi Tristram,

    I agree that showing that the Moore-paradoxical sentence isn’t a logical contradiction is easy. But I don’t see how you have a principled way of saying that it is a pragmatic rather than literal contradiction on the basis of “semantic rules.” What makes rules semantic? Maybe that these are entailments that one needs to accept as part of understanding the meanings of the relevant words. But I’d think that someone who didn’t realize that “It’s raining, but I don’t believe it’s raining” is somehow self-undermining would have betrayed linguistic incompetence. You can’t just say that certain rules are *semantic* without begging the question, because the whole problem is shy some rules are semantic and others merely pragmatic. We need to give a principled reason for putting a given rational tension on one side or another.

    I don’t think you necessarily want to appeal to truth to *explain* entailments for the expressivist. But consider what I’d take to be the normal statement of the philosophical problem Moore’s paradox posed: it sounds like a contradiction even though both conjuncts can be true at the same time. Is that description of the problem inconsistent with minimalism about truth?

  17. Eric Hubble says:

    I also like this a lot! I’d like to bolster your case. Here’s another (complementary, I think) explanation of why people would erroneously label normative sentences true and false.

    “True” and “false” are useful, in the sorts of ways that deflationists of various stripes like to point out. They let us express agreement and disagreement even if we don’t know what was said, etc.

    People know about these useful functions of “true” and “false”; that’s why they use them! Since the words serve these functions in normative discourse just as well as non-normative discourse, people use them there too.

    What the people don’t know is that “true” and “false” pick out the metaphysically robust properties, Truth and Falsity. And they also don’t know that normative sentences can’t be True or False. That’s all complicated philosophy. And since they don’t know this, nothing stops them from using “true” and “false” in normative discourse. And so they mistakenly do, since it’s so useful.

    Now we have two complementary theories of error! Even better.

  18. Frank Jackson says:

    Yes that’s the thought with this to add. A feature of no truth accounts of indicative conditionals (or anyway the version that seems to me the most appealing) is that they have a plausible story to tell about why indicative conditionals are useful despite not having truth conditions. When the indicative conditional ‘if p then q’ is assertable, the corresponding material conditional is i) probable and ii) its probability is robust wrt to p, and that’s enough to explain the utility of the indicative conditional. I took you to be saying something similar about ethical sentences: a no truth expressivist about them has a plausible story to tell about the utility of ethical sentences. No need to give them TCs.

  19. Jussi Suikkanen says:

    Hi Tristram

    this is very interesting (and my apologies about being late). I just had one concern about where the dialectic would go after this. It’s based on what you say here:

    “The expressivist can propose that, just like the intuitive applicability of the T-Schema to Liar sentences, the applicability of the T-Schema to normative sentences is an elegant overgeneralization that we would expect users to come to implicitly accept as they become competent with a natural language. Even if normative sentences are not truth-apt, then, this conjectured account can explain why being disposed to take them to be truth-apt is non-accidentally correlated with having become competent with ‘true’ in the ordinary way.”

    I worry that this will lead the expressivist to a dilemma. In order to make the case of overgeneralisation, the expressivist will need a substantial account of truth-aptness. Assume first that the expressivist makes the account very strict and demanding – maybe in terms of something like traditional correspondence theories of truth. Now, I worry that these accounts will be just as difficult to defend as minimalism generally for the non-normative realm. There are quite a many domains where there are no obvious things to which propositions correspond and yet we want to accept truth-aptness and not claim that ascriptions of truth are overgeneralisations.

    To avoid this horn, the expressivist could opt for a less strict, less demanding account of truth-aptness (even if not as minimal as to say that all well formed indicative sentences are truth-apt) and there are several accounts of truth-aptness and truth that would be appropriate here. However, in this case my worry is that on the standards of these accounts, normative sentences even in the expressivist framework can come out as truth-apt.

    So, I guess the worry is that for this view to work, the expressivist needs a substantial account of truth-aptness and truth such that (i) it carves the truth-aptness joints exactly at the right place between descriptive and normative and (ii) it is also a plausible account of truth for all the non-normative domains generally. And, I worry that (i) and (ii) pull in different directions. The more you want to get (i) the less you get (ii) and the more you get (ii) the less you get (i). What the expressivist thus needs is a Goldilocks account of truth and truth-aptness but it’s not immediately clear to me what such an account could be – finding such an account might be more difficult than defending minimalism against the paradoxes or accepting a substantial account of truth on which normative sentences can come out as true even if we accept expressivism.

  20. Tristram McPherson says:

    First of all, my apologies for being so slow to reply; a lot of time-sensitive demands got in the way.

    Derek: I think that I am better understanding the worry. Let me state how I understand it, to make sure I have it right. The worry is *not that we need truth to *explain the Moore-paradoxicality of certain normative utterances. Rather, the idea is that:

    1. There are Moore-paradoxical normative utterances
    2. The natural way to initially orient one to the idea of Moore-paradoxicality appeals to truth
    So, this is evidence that truth applies to normative sentences

    Is that the idea? If so, I guess I am unpersuaded.

    First, the main idea of my theory of error is that truth-aptness generally walks around with certain inferential and implication properties. *Of course we need to do better than imagining a list that says ‘semantical rules’. But here is the question: what is the best explanation of the inconsistency of “John is a married bachelor”? Either that best explanation appeals to truth or it doesn’t.

    If it does, then that looks awkward to me for the minimalist: it looks like truth is explaining semantic properties, and it is hard to see how a notion whose significance is exhausted by the T-schema etc. could do that. But if the best explanation of semantic inconsistency doesn’t advert to truth, then the best explanation of Moore-paradoxicality will advert in part to the absence of the presumed explanans there, and not to truth. And then I apply a version of the same theory of error to the initial gloss on Moore-paradoxicality: the initial gloss seems natural, because in non-normative cases, truth-aptness walks around so consistently with whatever the explanans is.

  21. Tristram McPherson says:

    Hi Eric!

    I think your proposal would be attractive on certain pictures of language. I’ve been thinking about it a bit, and I guess I am still a bit dissatisfied by it as a standalone story. One example: it seems to me that it would be useful in pretty much just the same way to be able to agree with, e.g. commands. But we don’t think “Open the door!” “That’s true!” is a felicitous exchange. Roughly, felicity judgments about truth ascriptions seem to track something like syntactic form *better than expressions for which is is useful to be able to agree or disagree.

    On the other hand, I think that within the broad sort of error I am proposing, pointing out the utility of truth-talk for expressing agreement etc. could help to explain why the use of sentences that involve normative truth ascriptions persist, and of course that sort of use would play an important role in reinforcing the sense of the apparently obvious felicity of such uses.

  22. Tristram McPherson says:

    Frank –

    Your addition seems exactly right to me!

  23. Tristram McPherson says:

    Hi Jussi!

    You make a bunch of interesting points here. Let me take them in turn.

    First, you say: “In order to make the case of overgeneralisation, the expressivist will need a substantial account of truth-aptness.”

    This is crucial, and I want to resist.

    First, consider the liar sentences. In order to make the theory of error work there, we don’t need a substantial *theory of truth-aptness; just a *constraint on such theories: that the correct theory doesn’t lead to a contradiction.

    Now, admittedly, in the case of normative sentences we do not have such a clean and intuitive constraint so immediately at hand. But it does seem to me that *at most, the proponent of my argument would need to argue for such a constraint, and not for any particular theory of truth-aptness.

    Next, I want to emphasize my understanding of my project: it is *not to argue that minimalism is false (I have provided no such arguments). Rather the point is to argue that: minimalism is not the only game in town for expressivists: they can afford to be uncommitted about the theory of truth.

    So, as I see the dialectic, the task isn’t at all to argue for a non-minimalist theory of truth-aptness: rather the question is just: given a plausible non-minimalist theory of truth-aptness, is this a plausible theory of error for the relevant felicity judgments?

    That said, I want to be concessive about two very good points that arise out of your comment:

    1. I have not in any way surveyed the field of theories of truth-aptness. If there are some such theories that are (a) plausible, (b) inconsistent with there being truths for expressivistically understood fragments of discourse, and (c) in some way unfriendly to the sort of theory of error that I have offered, then the expressivist at least has more work to do, and in some cases may need to take dialectically costly commitments rejecting them.

    But I would want to see the argument for a theory fulfilling (a)-(c) before I worried *too much about that.

    2. As in the fourth comment in the thread (my second reply to Jack), I want to grant your excellent point that the set of theories of truth that are compatible with there being normative truths, given pure expressivism is broader than the set of minimalist theories. Here again, my dialectical aim is to license the expressivist being as uncommitted as possible concerning truth. So if there are some theories of truth that are expressivism-truth compatible, and they are substantially different from minimalism, that is grist for my overall mill.

    3. Your sense is that, given the diversity of *other domains that are apparently truth-apt, the most generally plausible theory of truth will also be compatible with there being normative truths, given pure expressivism. This is an interesting sort of competing high-level case, and I would love to see it fleshed out. I should say, however, that I am not sure exactly which sets of problem cases you find gripping. Myself, I have never been at all persuaded by claims that there are ‘truths without truthmakers’ in various cases.

Leave a Reply

Your email address will not be published.