ABSTRACT
According to Michael Dummett, a truth~onditions theory of sense for a class of sentences is equivalent to realism in respect of their subject matter. 1 Realism is the thesis that truth (falsity) is an epistemically unconstrained property of a sentence; there is nothing in the concept of truth (falsity) to exclude the possibility that a sentence be unknowably true (false). This property of truth reflects the realist conviction, embodied in our customary linguistic practices, that the world, or a given sector of it, is determinately constituted, quite independently of any limitations on our capacity to come to know truths concerning it. Since, for the realist, truth-value may thus transcend our power to determine truth-value, as the world may transcend our power to discover its constitution, the principle of bivalence is regarded as generally assertible, whether or not we are equipped to determine, even in principle, what truth-value a sentence has. If it could be shown that there is some intrinsic difficulty in such a radically non-epistemic notion of truth , then realism, as a piece of commonsense metaphysics, would be undermined: the world, or some sector of it, could no longer be conceived in a realist way, as independent of our knowledge-acquiring capacities, but must rather be seen as somehow constituted or constrained by our cognitive activities, after the pattern, perhaps, of the intuitionists' conception of mathematical reality.2 Now Dummett claims to have an argument showing that there is indeed an intrinsic difficulty in the idea of conditions for truth being possibly knowledge-transcendent.3 The argument issues from an alleged failure of such a notion of truth-conditions to meet certain requirements, compulsory on any putative central semantic concept, deriving from the general theory of meaning. If Dummett's argument were sound, we should be compelled to repudiate realism with respect to any class of sentences, or sector of reality, for which those general requirements on the notion of truth, construed as the central concept of a theory of meaning, could not be met. My main purpose in this paper is to avert such wholesale metaphysical revision
*The rust draft of this paper was written in the spring of 1976; since then I have had the benefit of reading other people's work on the subject and the comments of several friends.
