To: David G.
From: Geoffrey Klempner
Subject: Dummett's argument for an anti-realist theory of meaning
Date: 17 September 2002 12:55

Dear David,

Thank you for your letter of 27 August, with your third essay for the Philosophy of Language program, in response to the question, "Give a critical account of the main features of Michael Dummett's argument for a global, anti-realist theory of meaning.'

This is a very difficult topic. You are correct in your inference that the essays (and units) increase in difficulty as the program proceeds. However, when it comes to understanding Dummett, many professional philosophers feel at a loss. When I started my postgraduate on the philosophy of language at Oxford university, the supervisor who had been assigned to me said that he could not accept any essays on Dummett or anti-Realism because he himself had not mastered the topic! (At the last minute, however, I was able to get John McDowell, the supervisor I had originally asked for.)

You give a good account of the background to Dummett's argument. The main point to make, as you say, is that we are concerned with taking a realist view of classes of statements, rather than of classes of objects?

But what exactly does that entail? There are in fact two ways in which one can take an anti-realist view of statements: First, you can deny the realist view of what it is that makes a statement true ('anti-Realism about truth'). Secondly, you can deny the realist view of a statement's meaning ('anti-Realism about meaning'). Dummett wants us to accept that these two views - about truth and about meaning - necessarily go together. I happen to disagree. In my view, a truth-conditional theory of meaning ('Realist' by Dummett's definition) is fully consistent with an anti-Realist view of truth. This is one of the things I try to show in the program.

You have picked up on one of Dummett's two arguments against a Realist theory of meaning, the so-called 'acquisition' argument. However, there is another argument which Dummett uses, concerning what it takes to *manifest* linguistic knowledge. The argument is what you'd expect: according to Wittgenstein's doctrine of 'meaning is use' a necessary constraint on a theory of meaning is that it gives an adequate account of how a speaker's implicit knowledge of meanings is 'manifested' in their linguistic behaviour. Dummett remarks at one point in his Frege book that it is logically possible that men could have sprung up from dragon's teeth, with the full ability to use a language. Such an alleged possibility would by-pass the acquisition argument, but not the manifestation argument.

You concur with my argument against Dummett, that his demand for a global anti-Realist theory of meaning is seriously weakened in the face of the fact that no such account has ever been successfully been given or (I allege) could be given.

This worries me, though. I don't like arguments which say, 'I can't see how such-and-such can be achieved, therefore such-and-such is unachievable'. However, there is more to say.

A more substantial argument is that Dummett sets far too high a standard for what would count as 'manifestation' of one's knowledge of meaning – effectively the demand for a reduction of meaning into behavioural terms. (This is the point which John McDowell makes in his seminal paper, 'Truth Conditions, Bivalence and Verificationism' in 'Truth and Meaning' McDowell and Evans Eds. OUP 1976.) The sticking point is Dummett's notion that the correlation between statements and 'manifestations' should be one-to-one. The alternative view is that it is the theory of the speaker's linguistic knowledge *as a whole* which is supported by the observed facts of linguistic usage. As in other theories – e.g. theories of physics – it is recognized that the best explanation or best theory can, or almost certainly is one which is *underdetermined* by the observed facts.

In the light of this, it is perfectly proper to remark that theories which take a truth-conditional form are, to date, the only serious contenders for an account of linguistic understanding. We have to go with the best theory.

The essay title did stress that Dummett is arguing for a *global* anti-realist theory of meaning. This is in contrast to arguments which concern anti-realism regarding a specific area of discourse. If there had been room, you might have looked at the way Dummett's uses specific areas where an account of meaning might be in dispute - e.g. the reality of the past, which you mention - as a springboard for his argument for global anti-realism. In Dummett's view, however, the only coherent form of anti-realism is a global theory: this is because the manifestation and acquisition arguments apply to any area of discourse.

Overall, it is remarkable that you have managed to squeeze so much into such a small space. Dealing with the issues I have mentioned would merely have increased the word count. You have done well.

All the best,

Geoffrey