H. P. Grice's "objector" in "Causal Theory" is not Austin!

 Grice’s objector isn’t Austin

I will not defend Grice’s objector, and insist the doubt-or-denial condition attaches itself to L-statements more strongly than Grice supposes.

 Rather, I hope to show that Grice’s objector is not Austin, and in fact shares certain basic assumptions about language that bring the objector closer to Grice than to Austin.

In particular, the objector has a weakness for generalization and a fixation on truth-value that Austin does not share.

Grice never mentions Austin by name, but Sense and Sensibilia was the most prominent contemporary attack on sense-datum theory, so Austin is naturally read as the prime target in this attempt to rehabilitate a causal theory of perception.

Grice also criticizes Austin explicitly elsewhere, challenging the Austinian maxim of ‘No modification without aberration’ (1979, 189), which is at work in ‘looks to me’ cases (Grice 1989, 8).

As it happens, Austin explicitly examines statements of the form ‘X looks p to me,’ and he does not say that a condition of doubt or denial must always attach to them.

In ‘Other Minds,’ he claims that a sentence like

Here is something that looks to me red.

could be variously interpreted: Contrast

Here is something that (definitely) looks to me (anyhow) red.

with

Here is something that looks to me (something like) red (I should say).



In the former case I am quite confident that, however it may look to others, whatever it may ‘really be,’ &c., it certainly does look red to me at the moment.

cf. Grice on

The tie is blue.
The tie seems blue.

LOOSE USE -- DISIMPLICATURE

In the other case I’m not confident at all: it looks reddish, but I’ve never seen anything quite like it before, I can't quite describe it—or, I’m not very good at recognizing colours, I never feel quite happy about them, I’ve constantly been caught out about them. (1979, 91)


Neither Austin nor Grice holds the view of Grice’s imagined objector, that certain implications must always attach to any utterance of a ‘looks to me’ locution.

But the lesson 16 here is not simply that Austin does not claim something like a doubt-or-denial condition always attaches to ‘looks to me’ locutions.

More to the point, Austin wants to spell out the context of an utterance before he can make sense of it.

To say of a certain class of statements that a certain condition always attaches to them is as gross a generalization as the assertion that every statement has a fixed literal meaning regardless of what further implicatures might be conveyed through contextual or other cues in its utterance.

Besides this tendency toward generalization, another point where Grice’s objector stands with Grice, and in opposition to Austin, is a conception of meaning that privileges truth-value.

Grice’s objector takes the non-fulfilment of the doubt-or-denial condition as a case for a third truth-value, N, which attaches to statements when neither T nor F apply (Grice 1965, 455).

 Grice seems to read Austin’s disparagement of the ‘true/false fetish’ as an objection to the Law of the Excluded Middle, insisting that we need a third truth-value to account for utterances where we cannot say that something either true or false has been asserted.16

But Austin’s concern with the ‘true/false fetish’ isn’t that this pair of truth-values must be supplemented by a further truth-value in order to capture the semantic values of all statements.

Rather, he is concerned that a too-narrow focus on truth-value risks blinding us to other ways of evaluating speech acts, which often better capture what is expressed.

A promise is neither true nor false, but Austin does not claim that statements of promise thus take a third truth-value, N, which they share with various forms of nonsense.

His point is not that there are more than two values in the domain of truth-value, but rather that there are more dimensions to meaning than the one defined by the true-false axis.

Falsehood is one kind of infelicity—and truth one kind of felicity—that can befall our utterances, and How To Do Things With Words shows, among other things, that truth and falsity occupy only a small place in a much larger picture. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 16

Even defenders of Austin occasionally take this route.

Charles Travis (1991) talks about a third truth-value, N, in his defence of Austin against Grice. 17

 In responding to the claim that statements like ‘The sky looks blue to me’ have invariant truth-conditions, one route we could take is to generate examples of such statements where the truth-conditions actually do turn out to vary depending on the contexts in which they are uttered.17

In other words, we could try to prove that Grice is wrong.

Charles Travis takes this route, suggesting that it is one of two strategies one might adopt in challenging what he calls the ‘classical picture’ of semantics:

‘One might try to show that the picture is senseless: that somehow or other, in stating it, we have appealed to some (putative) notion that is not, and could not be well defined.

Or one might accept the picture as sensible and coherent, and argue that it is false’ (Travis 1989, 14 – 15). While Travis opts for the second strategy,18 I am more inclined to pursue the first.19 3.3. An Austinian Response

One problem with ‘proving Grice wrong,’ if such a thing is possible, is that it accepts implicitly a privileging of the true/false dimension of assessment:

Grice’s argument consists of a series of assertions, some of these assertions are false, therefore Grice’s argument does not hold up.

Just as there are more ways of appraising a sentence than just its truth or falsity, there are more ways of appraising an argument than just its soundness or unsoundness.

Instead of proving Grice wrong, I will use an Austinian term of criticism and argue that Grice cannot seriously mean what he thinks he means.

Grice claims that the cancellability of conversational implicature shows that we can separate what is said (which cannot be cancelled without altering the sentence’s truthconditions) from what is conversationally implicated (the part that is cancelled), so that the !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 17

Travis is particularly creative in this regard, with sentences like ‘The table is covered in butter’ (1991, 240) or ‘There’s milk in the refrigerator’ (1989, 19). 18

Travis has engaged with Herman Cappelen and Ernie Lepore on this topic, who defend a ‘semantic minimalism’ that would be more sympathetic to Grice.

See Cappelen and Lepore (2005, 2006) and Travis (2006) for some of the moves in this debate. 19 Baz (2012), especially chap. 4, explores the divergence between Travis and ordinary language philosophers like Austin in some detail. 18 truth-conditions of a statement rely only on the former, and the meaning of the statement does not essentially include the latter.

 For instance, ‘The sky looks blue to me’ carries the implicature (according to Grice) that the speaker has some cause for doubt or denial because it violates the conversational maxim of Manner, being more prolix than the straightforward ‘The sky is blue.’

RATHER it's informativeness -- he omitted the section in the WoW reprint.

He is more concerned with a principle of conversational helpfulness, alla Butler, than manner-quality-quantity-relation alla Kant.



But, says Grice, we can cancel this implicature by adding further prolixity:

‘The sky looks blue to me, but I don’t mean to suggest that I have any reason to doubt or deny that the sky is in fact blue.’

Can Grice seriously mean this?

Well, he says that the cancellation can be unhappy.

"He has beautiful handwriting; I don't mean to suggest he's hopeless at philosophy."

"I cannot be claimed to have IMPLIED that, even if that's all what my co-conversationalist ends up thinking."

Would hearing that last sentence really cancel any implicature of doubt or denial that a hearer might draw from ‘The sky looks blue to me’?

If ‘The sky looks blue to me’ carries an implicature of doubt or denial because its prolixity flouts the Gricean Maxim of Manner, this much longer sentence flouts it even more so, and its utterance must have implicatures of its own.

 Rather than cancelling an implicature, it more likely suggests the speaker is trying to backpedal from words she wishes she had not uttered, and if anything, heighten the suspicion of doubt or denial.

Compare: ‘I can’t believe she doesn't like tea… not that I think there’s anything wrong with that!’

The latter half of that sentence, far from being accepted as a standard cancellation of the implicated sentiment of the first half, has become a cliché of ham-handed attempts to deny a sentiment that instead only highlights it.


Far from cancelling the first implicature, the added words simply add a further implicature. The lady doth protest too much.20

Two separate points are at play here. The first, which I raised in the above paragraph, is that the so-called cancellation of an implicature is as much a speech act as the original utterance, and should be just as subject to implicatures and infelicities as the original utterance on a Gricean reading.

The second, which I alluded to above and will argue for below, is that, if we treat cancellations as the speech acts that they must surely be, they can no longer do the work that Grice wants them to do. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 20

After developing this argument, I discovered a similar line of criticism against conversational implicature in Weiner 2006, although Weiner doesn’t make this argument part of a broader methodological point. 19

Let us consider how we use utterances of the form ‘… but by that I don’t mean to suggest …’ which is the form of the Gricean cancellation.

We often use these clauses to clarify an utterance that might have been misunderstood.

 If my words could be taken in more than one way, I can clarify my meaning by denying a possible understanding that I don’t intend.

To illustrate this point, we can adapt Grice’s letter of recommendation: the professor is speaking to a colleague about a student and says, ‘Smith is a fine thespian.’ T

he professor says that, knowing and admiring his student’s theatrical accomplishments. But then he realizes that, in the context of this conversation, and what with Smith on the job market, his praise might be mistaken for a dismissal of the student’s philosophical ability, and so hastily adds,

‘but I don’t mean to suggest by that anything to the detriment of Smith’s work in philosophy.’

The professor can reasonably hope to make clear that he does not wish to denigrate his student’s philosophical abilities, but only because the original utterance could plausibly have been interpreted in the sense that the added clause later makes explicit.

We do better, then, to interpret the added clause not as a ‘cancellation,’ which simply removes something from the original speech act, but rather as a further speech act that resolves a possible misunderstanding of the original speech act.21

In other words, ‘… but by that I don’t mean to suggest …’ locutions do not deal with implicatures at all, but rather with ambiguities.

Grice considers cases where only one natural interpretation is available.

His purported cancellation of the doubt-or-denial condition attaching to L-statements is meant to remove an implicature even in cases where there is no ambiguity.22

Gricean cancellation !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 21 T

his reading helps us to imagine a felicitous utterance of

‘I can’t believe she doesn't like tea -- not that I think that there’s anything wrong with that.’

Suppose the woman in question has often made anti-tea remarks in my hearing, or she is the daughter of a well-known anti-tea person .

In these cases, however, the ‘not that I think there’s anything wrong with that’ clears up a possible misunderstanding, which is precisely why it is acceptable. 22

One could, of course, imagine scenarios where ‘The sky looks blue to me’ could be taken in more than one way. Suppose I’m sitting outside with a child who’s drawing a picture of the landscape and colours the sky purple. As a gentle corrective, I say, ‘The sky looks blue to me,’ just as a third party approaches.

I might clarify my meaning to this third party by saying, ‘I wasn’t suggesting that there’s any reason to doubt or deny that the 20

does not try to clarify a possible misunderstanding, but rather to insist that we understand an utterance otherwise than the one way in which it is typically understood.

As I suggested above, the words Grice wants to use in that role do something else.

 I might add that it seems highly unlikely that any words could play that role: Humpty Dumpty might claim that his utterances can be understood otherwise than the one way in which they are typically understood, but the rest of us do not have this luxury.

Cancellation is not so much unsuccessful as incoherent.

Ultimately, the whole matter of cancellation merely postpones the problem it is supposed to solve.

Utterances often seem to mean something more or something else than what they strictly seem to say, and conversational implicature is supposed to account for that something more or something else. It also makes room for an analysis of what utterances strictly say by isolating this something more or something else and providing means for cancelling it so that analysis can focus exclusively on what is said.

Grice claims we can cancel the implicature through a further utterance, which he calls a ‘cancellation.’

He analyzes the original utterance in considerable detail, but he devotes no attention to the utterance of the cancellation.

He does not countenance the possibility that this utterance might also mean something more or something else than what it strictly says.

Cancellation does not effect the separation of what is said from what is implicated, but rather, at best, diverts the question of their separation to a second, more elaborate utterance.

On this analysis, Grice is guilty of a subtle form of question begging.

He wants to insist on the sharp separability of what is said from what is implicated by virtue of the cancellability of implicatures (in the case of conversational implicature, which is what interests us here). But the very intelligibility of the notion of cancellability already assumes this separation: we are asked to consider only the semantic impact of the cancellation without considering what implicature might attach to its utterance.

In other words, Grice’s !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! sky is blue.’

But this utterance makes sense precisely because it resolves an ambiguity. Grice wants cancellation to be able to apply generally, even when there is no ambiguity. 21 cancellation is an idealization, a piece of semantic fiction that only does the work it’s intended to do if we turn a blind eye to how it might actually be used in an utterance.

In begging the question on this point, Grice blinds himself to the fact that cancellation cannot neatly isolate the semantic component of a statement.

Grice’s conception of cancellation is meant to show, among other things, that Austin’s fine-grained examination of the details of ‘what we should say when’ (Austin 1979, 182) cannot disrupt the generalizing thrust that much philosophical theorizing requires.

However odd it might sound, Grice wants to say, we can still isolate the literal meaning of ‘looks to me’ locutions in a way that allows us to build a causal theory of perception—or, more generally, we can still isolate the literal meanings of odd-sounding expressions in a way that allows us to build philosophical theories.

But if the notion of cancellation is incoherent, so is the notion of literal meaning that it supports: both of them are idealizations of the sort Austin criticizes.

Grice’s argument attempts to isolate the literal meaning of an expression, and I have claimed that his attempt fails. Searle also attempts to supersede Austin by isolating the literal meaning of expressions. But where Grice attempts to move against Austin’s method, Searle attempts to appropriate and improve on Austin’s work.