Does truth attach to sentences, or to what sentences express? If to sentences, then certainly not to type sentences, such as ‘I am going to London tomorrow’, but only to token sentences, that is, sentences considered as uttered by a particular speaker at a particular time. It would, however, be inconvenient to restrict truth to utterances that are actually made; we may therefore adopt the device and terminology of Davidson, and speak of a ‘statement’ constituted by a triple [s, i, t] of a type sentence s, an individual i and a time t, the existence of which does not depend on whether i in fact uttered s at t. I shall presume that the identification of a type sentence depends on identifying the language to which it belongs. A familiar, irritating obstacle prevents our explaining that, when i did not utter s at t, the ‘statement’ is to be said to be true just in case he would have said something true if he had done so; the obstacle consists in sentences like, ‘I am not now speaking.’ The difficulty is not serious. What more is needed in order to obtain, from a type sentence s, something apt to be characterised as true or false is an assignment of references to indexical and demonstrative expressions occurring in s; we may say that s ‘comes out’ true or false under such an assignment. We may therefore take a ‘statement’ [s, i, t] to be true or false according as s comes out true or false under the assignment of t to the word ‘now’, i to the words ‘I’, and ‘me’, the place where i is at t to the word ‘here’, and so on. There is no reason to be disconcerted by the fact that this will yield as true a ‘statement’ involving the sentence, ‘I have been dead for a hundred years.’ Demonstrative expressions like ‘that house’, ‘this country’, etc., are less easily dealt with. Many should be regarded as devoid of reference unless, at t, i actually makes a pointing gesture or the equivalent; but the question need not be pursued here. For convenience, however, we may continue to speak of the ascription of truth to sentences, as long as we bear in mind that it is to ‘statements’, in our special sense, to which it is really to be ascribed.
Michael Dummett was the Wykeham Professor of Logic at Oxford University from 1979 to 1992, and is Honorary Fellow of New College, Oxford. He is the author of books and articles on philosophy, the theory of voting, and the history of card games. He was knighted in January 1999.