2.9.1 The Humpty objection

The Humpty objection to the Gricean position can be formulated as the charge that Grice is too much like Humpty for his own good. Humpty, we saw, seems to be advocating the following thesis:

The charge against Grice can be formulated in a two-premise argument:

• Premise One: Humpty's thesis is false.

• Premise Two: Grice's position entails Humpty's thesis.

• Conclusion: Grice's position is false.

To assess this charge, let us first consider Premise One. There does seem to be something terribly wrong with Humpty's thesis, but before getting to what is wrong with it, it is as well to appreciate why it could be attractive, to us and not merely to a self-important egg. Words do not have their meaning intrinsically. What they mean is arbitrary. The word ‘dog’ might have meant what the word ‘cat’ in fact means, and vice versa. Equally, nothing about the letters ‘i’, ‘m’, ‘p’, ‘e’, ‘n’, ‘e’, ‘t’, ‘r’, ‘a’, ‘b’, ‘i’, ‘l’, ‘i’, ‘t’, and ‘y’ requires that, taken together, they must mean what they do – roughly, a lack of susceptibility to physical infiltration. They could easily have been combined to mean something quite different. What is wrong with assuming that this arbitrariness is resolved by a decision of the utterer to mean one thing rather than something else? Nothing at all, Humpty would no doubt insist (see Alice and Humpty's conversation).

He could support his case by pointing to the benign practice, common in philosophy as in any sphere of life, of deciding to introduce a new term with a specific meaning laid down in a definition, or else of taking an old term and stipulating that it will be used in a novel way, or with one from among several established usages. We have an example of this in Grice's potentially insightful departure from the normal usage of ‘utterance’. Is this practice of stipulation not proof that we can choose to use a term however we want and to bestow on it any meaning whatsoever?

Against this, there are strong reasons for thinking that Humpty's thesis is mistaken, as Premise One asserts. A quick reason for thinking this is to follow Grice's own methodology of appealing to intuitions. Suppose Humpty were to utter: ‘There's glory for you.’ We would judge that his utterance means that some salient object or event is glorious. Humpty, following his own thesis, regards it as meaning that he has just given a nice knockdown argument. So, our considered intuitions do not accord with Humpty's thesis.

A second and more sophisticated reason for denying that meanings are entirely in our control depends not on any appeal to brute intuitions but on consideration of the role of meaning in the communication of knowledge. Utterer-controlled meaning, unless it is accompanied by uptake on the part of the audience, would not lead to communicative success. Evidence for this is provided by Alice's confusion when Humpty uses a word in a particular way without announcing to Alice beforehand that he is going to use it this way. What our utterances mean must be available to the audience if this meaning is to be any use. To that extent, what words mean is not entirely up to the speaker. Grice himself seems to be working towards this point when he writes, ‘the intended effect must be something which in some sense is within the control of the audience’ (Reading 1, p. 186).

Moreover, Humpty's thesis is not clearly supported by the phenomenon of stipulation. When we depart from common usage we generally have to announce this beforehand, using terms whose meaning is already familiar. Humpty fails to make a prior announcement, which is why Alice is left confused. It is only because he falsely believes his thesis that he is led to suppose that he can mean whatever he wishes with no prior announcement.

The falsity of Humpty's thesis can also be reconciled with the arbitrariness of language. There are other ways than individual legislation for words to come to mean what they do. Perhaps they mean what they do because we have beliefs about what they mean, not because we decide what they mean. So let us accept Premise One – that Humpty's thesis is false – and turn to the second premise, which is that Grice's theory entails Humpty's thesis. This is where the Humpty objection looks vulnerable.

Humpty's thesis asserts that our utterances and the words in them mean whatever we want or choose them to mean. Grice's theory is that what our utterances mean is a matter of what we intend to communicate. Intending, wanting and choosing may sound as though they come to much the same thing, but reflection suggests that they may be quite different.

If the subtleties noted in the previous activity are taken seriously, the second premise in the argument against Grice looks weak. Humpty's thesis is about choosing and wanting, neither of which is the equivalent of intending.

The main difference between Grice and Humpty seems to be this. Intention formation, the notion at the heart of Grice's theory but not Humpty's, requires an expectation of success. This was a lesson of the bank vault example: in general, we cannot form intentions to act unless we expect there is some chance that acting on the intention will lead to its fulfilment. In Grice's theory, the intention is to bring about a change of belief in an audience, and to bring it about through having them recognise this very intention. If we selected words according to personal whim, there would be no reasonable expectation of success. When Humpty used ‘impenetrability’, he had no reason to think that he would be successful in getting Alice to recognise him as meaning what, after the fact, he insists he did mean. After all, he had not warned her beforehand that he would be ascribing this unusual meaning to the word. With no such reason, he could not even have formed the intention to communicate using those words with that meaning. So Grice would not allow that Humpty meant what he later claims he meant.

Premise Two, then, seems to be false. Humpty's thesis may be mistaken, but Grice's theory looks like it is consistent with this falsity. In fact, once the distinctions between choosing and intending and between wanting and intending are granted, Grice's theory even seems to explain why Humpty's thesis is false. Because we cannot expect to be interpreted however we wish, we cannot intend to be interpreted however we wish, and so cannot mean whatever we wish.

These considerations may see off the Humpty objection, but they should also remind us that Grice still owes us a viable theory of what expressions mean. Grice, unlike Humpty, does not hold that what our utterances mean is entirely up to us. This is because what our utterances mean is a matter of what we intend to communicate, and we cannot intend to communicate that p unless we expect to succeed in communicating that p. This expectation of success is clearly tied to our choice of words. A speaker will have to feel that she has chosen appropriate words, words that her audience will interpret accordingly. For example, she will choose ‘tiger’ rather than ‘apricot’ if she intends to communicate information about a tiger (unless she and her audience have a special code). But to talk of ‘appropriate’ words, here, is to talk of words that have a particular meaning, i.e. ones that accord with what the speaker intends the audience to come to believe, etc. Grice's theory of the meaning of utterances therefore needs to be supplemented with a theory of the meaning of expressions that is more plausible than the one he provides in Reading 1. We saw earlier (section 2.7) how that theory does not extend to words or novel sentences.

The meaning of words, and of the sentences built out of words, also figures in the next objection, due to John Searle. In the subsequent discussion of Searle's objection I will sketch a theory of word meaning that seems to be compatible with Grice's project. This theory has in fact been around since long before Grice wrote his paper.