Language and thought are related in at least this way: language is a means for the expression of one’s thoughts and a vehicle for their communication to others. A speaker uses the words ‘The sky is blue’ (say), he thinks that the sky is blue, intends to say that the sky is blue, and he has expressed his thought; an audience hears the words, and if he understands them and fulfils the speaker’s intention by realising that the speaker said that the sky is blue, there has been communication. This description gives us an idea of what it is for an overt act of speaking and the states of mind of speaker and hearer to have some shared content: the speaker says, and the speaker thinks, and the audience grasps that the sky is blue.
Pollock thinks that such a simple description is seldom applicable: he thinks that it is incorrect when the words a speaker uses are dependent on context for their sense, and he thinks that nearly all the sentences we use contain context-dependent words. He puts his central thesis by saying that statements (products of acts of saying) are not the same things as propositions (objects of belief). To put it another way: for the most part, that which we believe we cannot say.
The context-dependence of bits of language has long been recognised. If I say ‘I’m tired,’ and you say ‘I’m tired,’ then I may have said something true and you something false, so that we can’t have said ‘exactly the same thing’. But we can attribute a constant conventional significance to words like ‘I’ (and ‘now’ and ‘that cat’) even though they are used to different effect by different speakers: we can treat them in such a way that knowing what they mean is knowing what determines what is said by using them from occasion to occasion. What we all know in knowing what ‘I’ means, for example, is that, in any particular instance, it is used to refer to the person who gives voice to it. This is simply a way to allow that we can make more distinctions when we say what is said by people using the word ‘I’ than we can make among the sentences people use which contain the word ‘I’.
Pollock for his part would register this by saying that there are more distinctions among statements people make than among the sentences they use. His claim is then that we need also, and further, to allow that there are more distinctions among the beliefs that speakers express than among the statements they make. A belief which I have about myself, and which I convey in saying ‘I’m tired’ is something in believing which I think of myself in a certain way – in that way in which a person typically thinks about herself and in which no one else can think about her. But then, Pollock claims, the proposition I believe is idiosyncratic: you cannot believe what I believe when I believe I’m tired; and if per impossibile I said what I believed, you wouldn’t understand me.
Pollock is not alone in holding that a proper treatment of context-dependence will prevent us from thinking of our thoughts as characterisable in the same terms as those we use to characterise the acts of speech which express those thoughts. The literature in the field is expanding, and much of it is too recently published for Pollock to have taken it into account. But nearly all who contribute to it commit themselves to rejecting some component of the cluster of ideas presented in my first paragraph. Pollock’s rejection of those ideas is more explicit than many people’s, but he has no more than anyone else to say about a substitute for them. Pollock’s way is to put the phrase ‘says what he believes’ into scare quotes, as if we had to abandon our natural ways of thinking by sneering at them (or perhaps Pollock thinks we can retain them provided we put our tongues in our cheeks).
It can seem to be a devastating objection to Pollock that he apparently admits that, strictly speaking, we very seldom say what we believe. But it is actually not clear (though Pollock must suppose it is clear) that he is forced into such an admission: he may not need his scare quotes. It is true that Pollock denies that that which we believe is what we say: but he might try saying that even so we can say what we believe: we can say what we believe in the sense in which ‘what’ is not a relative pronoun but reports the answer to a question (the question ‘What does x believe?’). And certainly Pollock’s point that uses of the word ‘I’ are connected ineluctably with a certain mode of thought shows that one doesn’t say everything about the meaning of ‘I’ when one accepts the orthodoxy that it is used to refer to whoever speaks it. So one might think that we are bound to resort to the apparatus of idiosyncratic propositions which Pollock introduces in order to say some more about ‘I’ and about other context-dependent words.
But are we so bound? Those who think that we cannot get any more out of the notion of a proposition or a statement or a thought than can be got from considerations about how we report beliefs and sayings and thinkings will be suspicious from the start. They hold that the idea of a content shared by an overt act of speech and by the states of mind of speaker and hearer comes down to no more than this: that one can, as reporter, make an utterance which serves to record what was said by a speaker and what was thought by the speaker and what was grasped by the hearer. (I put the last four words of my opening paragraph to this end.) These people think that we have no understanding of that which the speaker thinks excepting what is given by saying what (indirect question) he thinks. And they will see a point in protesting ‘Of course you can believe what I do when I believe that I’m tired; you show yourself as believing that when you find me a place to sleep, or when you say I’m tired, by directing the words “You’re tired” towards me.’ To see the sense in their response to Pollock, one has to realise that, by accepting (for example) the obvious fact that we can make more distinctions in saying what is said by people using the word ‘I’ than we can make among the sentences they use, we are not compelled to take the view that there are objects (‘things said’) among which we then distinguish. There is no need to reify statements or propositions. And perhaps there is a need not to reify them.
But philosophers who find the notion of a proposition in no way problematic will be ready to grant Pollock his apparatus. And they will probably see no especial difficulty for the idiosyncratic propositions that he introduces. For they are likely to take the view that we can admit fine-grained propositions so long as there are genuine fine-grained distinctions to be drawn. Pollock draws distinctions among modes of thought which are genuine enough, so that (on this view) he has all the objects that he thinks he needs – including the objects that one person alone can think. Still, even these philosophers may come to have their doubts about the uses to which Pollock can put idiosyncratic propositions. It seems plain wrong to suppose that when we express our thoughts in words we are trying to do something impossible – to say that which cannot be said. But then it looks as though we shall distort the phenomenology of speech unless we insist that Pollock’s idiosyncratic and ineffable propositions are things which our communicative intentions never concern. One wonders then, though, whether they can play any fundamental role in an account of language use.
Questions about the connections between acts of speech and states of mind are among the most fundamental raised by Pollock’s book. But Pollock’s chief concern is not with these questions but with the repercussions of his own answers to them for the project of constructing a philosophical theory of language. He seeks to say in systematic fashion what contributions words make to statements on occasions where a statement is determined both by the propositions that can be sent by a speaker making the statement and by the propositions that can acceptably be received by an audience to it. The result is a new-style ‘statemental’ semantics; great philosophical ingenuity and mathematical competence are exercised in its construction. Pollock does not strive for ‘horizontal completeness’ – which would require an account of all the parts of language; nevertheless he gives formal accounts, with detailed informal justifications, of subject-predicate sentences, of sentences about what is necessary, of sentences about people’s knowledge and belief, and of questions and commands.
Statemental semantics is applied across the board. For we are told that, just as I can use ‘I’ in communication with you although you cannot believe what I believe about myself, so you and I can communicate with one another using a proper name even where I think of the object it stands for differently from you. Again, since you and I may have different ideas about lemons, for instance, one needs (in Pollock’s view) to distinguish the attribute of lemonhood attributed in statements from more private concepts predicated in the different propositions about lemons that we may each entertain.
What goes for ‘lemon’ goes for all predicates that are not simply definable. In giving this account of predicates, Pollock takes a middle course. He retains something of the old empiricist view that two people’s capacity to communicate about lemons consists in their each attaching to the word ‘lemon’ some stock of simple ideas: but he also retains something of the presently more widely accepted (if just as old) view that lemons themselves and people’s causal connections with lemons are crucial to that capacity. This is just one of the many places at which Pollock engages in current debates about reference, about the relations between words and minds, and between words and the world. His new approach to semantic questions is most thoroughly worked out.
And Pollock is not content to analyse the statements we make. He wants also to say what it is to make them. There is an account of speech acts just as novel as the account of their contents. Making statements is characterised in part by the moral obligations incurred in speaking. Languages are institutions, comprised of constitutive rules; and speaking is participating in such an institution.
The moral analysis of participation (whether in a particular language or in a football game) completes the analysis of language in general. It has all been done in terms ultimately of non-linguistic and non-semantical notions. This is vertical completeness, to which Pollock does aspire.
To be persuaded of the worth of Pollock’s undertaking, one must presumably think of the ingredients in his analysis as better grounded than the notions analysed. It might (conceivably) be maintained that notions of the contents of individuals’ consciousnesses, and of moral obligation, are more easily understood, or grasped, or applied, than the linguistic notions analysed. Or (more likely) that these mental and moral notions are more fundamental metaphysically speaking than the linguistic ones. To me it seems more obvious that Pollock needs to say one or other of these things than that either can plausibly be said.