2.9.2 Searle's objection

In ‘What is a speech act?’, John Searle introduces a memorable example of an utterance in which Grice's conditions are all met for it to mean one thing, but where the words used suggest that the utterance means something quite different, if it means anything at all. The conclusion Searle invites us to draw is that what our utterances mean is not exhausted by the speaker's intentions alone. An additional consideration is the meaning of the expressions used. If they don't match the intention, then nothing is meant.

What should we make of Searle's American soldier example? Grice could just dig in his heels and insist that what the American soldier's utterance means really is that he is a German officer, notwithstanding any intuitions we may have to the contrary over this interpretation. In philosophy we are often called on to reject the pre-theoretical intuitions we have about situations. But it is usually better to avoid doing so unless forced.

Searle's own response is to suggest that Grice has overlooked the importance of community-wide linguistic conventions, the ‘rules of language’, to the meaning of an utterance. Searle proposes that the speaker's intentions are the main source of what our utterances mean, but he adds a further condition: these intentions must accord with the meanings of the expressions we use, where this is fixed by community-wide conventions, the rules of the language from which the expressions are drawn.

Searle thinks of himself merely as extending the Gricean analysis, correcting it for an oversight. He does not think he is overthrowing the Gricean project. But it may be that adding this new constraint – that what the speaker means depends on community-wide conventions – would in fact leave Grice's project deeply damaged. This would depend on whether the meanings of expressions, fixed as they are by community-wide linguistic conventions, can be reduced to the mental states (the beliefs, intentions, etc.) of those who make up that community. If they can, then the reduction of linguistic meaning to mental content will have succeeded, albeit not quite in the way Grice envisaged. On this new picture, but not on Grice's original picture, the meaning of an individual utterance would depend on more than the mental states of the speaker alone. It would depend also on the mental states of those in the speaker's community, since these fix the meanings of the expressions used, and these meanings in turn are a factor in the meaning of the individual utterance.

Two questions arise at this point. It is clear that following Searle's suggestion or persisting with Grice's original theory will require a viable theory of how the meaning of expressions reduces to the content of mental states. So what is that theory? A different question is whether the relevant mental states are those of the speaker alone, or whether they also include those of members of the wider linguistic community as Searle implies. Let us start with this second question.

Grice could endorse a slightly weakened version of Searle's additional requirement, and in doing so persist with his claim that only the individual speaker's mental states are relevant to what an utterance means. Searle requires that the speaker's intentions actually accord with the community-wide meaning of the sentence used. But Grice could require merely that the speaker's intentions accord with what the speaker thinks is the community-wide meaning of the sentence used – or even with what the speaker thinks the specific audience takes the sentence's meaning to be. Adding such a weak requirement is enough to explain our intuitions about the American soldier example. The American soldier does not think that the established meaning of ‘Kennst du das Land, wo die Zitronen blühen?’ is that the speaker is a German officer, so Grice would not be committed to saying that this is what the utterance means. Moreover, Grice's attempt to define the meaning of utterances in terms of the individual speaker's psychological states would still be on track. The weakened condition, unlike Searle's condition, is couched in terms of the beliefs of the individual utterer alone.

We might wonder whether this weaker condition deals with counterexamples as well as Searle's condition does. For example, suppose the American soldier really did believe that ‘Kennst du das Land, wo die Zitronen blühen?’ meant the same as ‘I am a German officer’. Perhaps he was told this in military training. Would it now be fair to say that the American soldier's utterance meant that he was a German officer? Grice's new condition (and all the old ones) would be met but Searle's condition would not, so an affirmative answer would favour Grice's weaker condition and a negative answer would favour Searle's stronger one. I will not try to settle this issue here, except to note that people's intuitions about this new version of the scenario differ. Some think that the American soldier's utterance does now mean the same as ‘I am a German officer’. They are likely to be happy with the weaker condition. Others deny that it means this. They are likely to be sympathetic to Searle's stronger condition.

Now consider the challenge, facing both Searle and Grice, of how to reduce the meaning of expressions to the meaning of mental states. Even adopting the weaker condition leaves this challenge in place. Grice could not treat beliefs about what expressions mean as fundamental, because ‘what expressions mean’ is itself a kind of non-natural meaning, and as such needs to be reduced to the content of mental states.