For Truth in Semantics

For Truth in Semantics

Description

It is odd that someone with a memory as poor as mine should remember so pricesely when he first heard mention of assertibility conditions. But I think I do. For I can remember going into Philip Pettit’s rooms in Trinity Hall when I was an undergraduate and seeing on the tiny blackboard he had there the three expressions t-c’s, a-c’s and j-c’s. Never one to miss an opportunity to avoid discussing my essay, I asked what these expressions stood for: the answer, as of course you have already worded out, is ‘truth conditions’, ‘assertibility conditions’ and ‘justification conditions’. Because I can remember this I know that from the very beginning I was taught not to assume that when someone said ‘assertibility conditions’ they were talking about evidence that warranted an assertion. Of course, for Dummett, a-c’s are j-c’s; but if I had not started off thinking of this as a substantial claim, I might never have become interested in theories that use the notion of a condition for assertion that is not that of having evidence for the belief that the assertion expresses. I think that would have been a pity, as I wrote a doctoral dissertation (1981) and then a book (1985b) that showed how such a notion solved the central questions of the semantics of indicative conditionals. On the way I became interested in Dummett’s use of the term ‘assertibility’, and this book is the product. ‘Interested’ may be too mild a word. For this book is a polemic. Though I shall make positive suggestions from time to time, my main intention is to attack the cluster of views that Michael Dummett has developed under the banner of ‘anti-realism’. I think anti-realism should be given up: given up because many of its claims are false, and the conjunction of its central claims is inconsistent. At the heart of anti-realism are three claims: (a) that standard truth-conditional theories of meaning—which Dummett has styled ‘realist’—are objectionable because such truth conditionals transcend verification, (b) that such theories are committed to bivalence, and that this entails verification-transcendence and (c) that we should therefore develop an alternative style of semantic theory, whose basic notion is not truth but assertibility. I call (a) the negative programme, and (c) the positive programme of anti-realism. This collection of ideas is of quite general interest. For (a) is related to verificationism, which is in the background of all modern analytic philosophy, and (c) derives from pragmatism, the other main tradition of modern empiricist thought. Because of this, I think, we can take a look through the lens of anti-realism at how more general issues in epistemology impinge on the philosophy of language. I think I am an empiricist: ‘empiricism’ is a broadly encompassing term. But I have three major objections to these claims. I deny (a) because it rests on a misunderstanding both of what is involved in a theory’s being verifiable and of the way that truth conditions operate in realist theories. I deny (a), that is, despite the fact that I am sympathetic to verificationism. That is my first objection. I deny (c) because assertibility-based theories entail falsehoods about meaning. That is my second. I also think (a) and (c) are inconsistent, since assertibility-based theories are verification-transcendent if truth-based theories are; so you might say my third objection is that you cannot have verificationist grounds for pragmatism. You will notice that I say nothing about (b), which Dummett has made central to his position. Dummett’s view is that bivalence entails verification-transcendence; I deny this also, because it rests on the same misunderstanding that vitiates (a). So I do not need to discuss the connection between bivalence and realism, since, as Dummett would concede, it is only if realism is verification-transcendent that bivalence is an issue. Neil Tennant has kindly shown me a manuscript of his book, Anti-Realism and Logic, where he defends the theses that in the domain of formal objects bivalence fails, and I am sympathetic to the view that anti-realism of some sort is indeed plausible there. But my concern is with empirical reality, and here, as I say, I shall argue that anti-realism is wrong. Because I think Dummett misunderstand the role of truth conditions in realist theory, I begin, in chapter 1, by giving a brief outline of what I think is the correct account. Polemic, though addressed against one position, is usually intended as an indirect defence of another. The full theory, of which chapter 1 is an adumbration, I give in Assertion and Conditionals (1985b). Chapter 2 outlines both the positive and negative programmes of anti-realism. That much is given, the stage is set for part II, which offers the arguments against Dummett’s understanding of verifiability; and part III, which argues against his pragmatist leanings. So perhaps I should say this book is a polemic preceded by a sketch of its background. In part II I try to clarify the relevant facts about verification; this is the attack on (a), the negative programme of anti-realism. Chapters 6-8, which make up the bulk of part III, are the attack on (c), the positive programme. Chapter 6 shows that anti-realism is verification-transcendent if realism is. Chapter 7 shows that a particular version of assertibility condition semantics, suggested by Crispin Wright, Dummett’s stoutest lieutenant, fails. Chapter 8 explains why no assertibility condition theory of the kind Dummett envisages could succeed. In chapter 6 I suspend my disbelief in Dummett’s notion of verification-transcendence. I do not think that the arguments there are an objection to the positive programme: I do not think that the fact that a theory transcends verification in Dummett’s sense is an objection to it at all. The force of the arguments is to prise apart the positive and negative programmes by showing that the negative programme undermines the positive one. That is why the further arguments of chapters 7 and 8 are needed to show, independently, that the positive programme fails. Dummett’s views are usually seen as a generalization of intuitionist criticism of Platonism in mathematics. I shall say little about intuitionism in se but I do argue, at various points, that mathematics is a misleading model. If that is right, my reticence about intuitionism is in order. My own view is that a realist—that is, a truth-conditional—semantics need not be objectionable on verificationist grounds, if verifiability is properly construed. Unlike most realists, my position is thus not that the truth of realism requires that we give up demanding verifiability of our theories. Whether we should demand that our theories be verifiable is another question: for the moment, I am disposed to do so—subject to caveats entered at the end of 5.5—largely because I would rather be epistemologically safe than sorry. But if we should not, that is another reason for giving up anti-realism. Conceding verificationism is in part, then, a rhetorical strategy. It is widely agreed that anti-realism is verificationist and some people have objected to it on that ground alone. If we can show that anti-realism is objectionable even if verificationist, which anti-realism clasps to its bosom, is true, that strengthens the case against it. I mentioned just now some caveats that I enter at the end of 5.5 about the possibility of a realist’s accepting some form of verificationism. These caveats have the effect of limiting verificationism to the claim that sentences ascribing certain properties to objects ought to be verifiable. I do not say there which properties these should be: but I do think it is consistent with realism to insist that this condition be satisfied for some objects by every property that is expressed by a predicate to which our semantics assigns an interpretive axiom. This, however, is a mere expression of opinion. For reasons that I shall give in 5.5, I do not think it need be part of the project of this book to argue for such a claim. In fact, the form of verificationism I assume in part I has been identified by Wright as a (weak) kind of anti-realism, and it may be as well to bear this in mind throughout part I, until I defend my doubts about this identification. But if Wright is correct, that does not weaken the force of my argument. For what I show is that, whether or not we call this form of verificationism ‘anti-realist’, it does not threaten realist semantics in the way anti-realists have argued. If I were asked what the importance of these questions is, I would say, first: the negative programme of anti-realism is verificationist, the positive programmed is pragmatist; to show that the former is consistent with realism and the latter cannot be carried out, should be of interest to many with no established interest in anti-realism in semantics. I would go on to add, however, that the persistence of the anti-realist view is one of the most serious obstacles to the acceptance of the realist programme, and it diverts much intelligence away from developing realist theories. I hold the spread of true belief and the dissipation of false belief, especially where justified by sound reasoning, to be, ceteris paribus, desirable. This book is meant as a contribution to that end. To borrow a familiar Wittgensteinian image: there is a ladder of argument here up which to climb, but when you get to the top, the ladder is only for throwing away. This book is intended to make itself unnecessary.

Publication Date

1986

For Truth in Semantics

Find @ NYU

Share

COinS