I am using 'deflationism' as an umbrella term subsuming several different deflationary theories of truth, among them Ramsey's redundancy theory, Quine's disquotationalism, Horwich's minimalist theory, and others. Deflationary theories contrast with what might be called 'robust' or substantive' theories of truth. It is not easy to focus the issue that divides these two types of theory. One way to get a feel for the issue is by considering the traditional-sounding question, What is the nature of truth? This 'Platonic' question -- compare What is the nature of knowledge? (Theaetetus); What is the nature of justice? (Republic) -- presupposes that truth has a nature, a nature that can be analyzed or otherwise explicated in terms of correspondence, or coherence, or 'what conduces to human flourishing,' or what would be accepted at the Peircean limit of inquiry, or something else.
The deflationist questions the presupposition. He suspects that truth has no nature. He suspects that there is no one property that all truths have, a property the having of which constitutes them as truths. His project is to try to account for our truth-talk in ways that do not commit us to truth's having a nature, or to truth's being a genuine property. Of course, we English speakers have and use the word 'true.' But the mere fact that we have and use the predicate 'true' does not suffice to show that there is a property corresponding to the predicate. (Exercise for the reader: find predicates to which no properties correspond.)
So if we can analyze our various uses of 'true' in ways that do not commit us to a property of truth, then we will have succeeded in deflating the topic of truth and showing it to be metaphysically insubstantial or 'lightweight.' The most radical approach would be one that tries to dispense with the predicate 'true' by showing that everything we say with its help can be said without its help (and without the help of any obvious synonym such as 'correct.') The idea here is not merely that truth is not a genuine property, but that 'true' is not even a genuine predicate.
Consider two assertions. I first assert that snow is white, and then I assert that it is true that snow is white. The two assertions have the same content. They convey the same meaning to the audience. This suggests that the sentential operator 'It is true that ___' adds nothing to the content of what is asserted. And the same goes for the predicate '___ is true.' Whether we think of 'true' as an operator or as a predicate, it seems redundant, or logically superfluous. In "Facts and Propositions" (1927), Frank Ramsey sketches a redundancy or logical superfluity theory of truth. This may be the first such theory in the Anglosphere. (Is there an historian in the house?)
For Ramsey, "there really is no separate problem of truth but merely a linguistic muddle." Ramsey tells us that ". . . 'It is true that Caesar was murdered' means no more than that Caesar was murdered, and 'It is false that Caesar was murdered' means that Caesar was not murdered." (F. P. Ramsey, Philosophical Papers, Cambridge UP, 1990, ed. D. H. Mellor, p. 38) But what about a case in which a proposition is not explicitly given, but is merely described, as in 'He is always right'? In this example, 'right' has the sense of 'true.' 'He is always' right means that whatever he asserts is true. As a means of getting rid of 'true' in this sort of case, Ramsey suggests:
1. For all p, if he asserts p, then p is true.
But since "the propositional function p is true is the same as p, as e.g., its value 'Caesar was murdered is true' is the same as 'Caesar was murdered,'" Ramsey thinks he can move from (1) to
2. For all p, if he asserts p, then p.
If the move to (2) is kosher, then 'true' will have been eliminated. Unfortunately, (2) is unintelligible. To see this, try to apply Universal Instantiation to (2). If the variable 'p' ranges over sentences, we get
3. If he asserts 'Snow is white,' then 'Snow is white.'
This is nonsense, because "'Snow is white'" in both occurrences is a name, whence it follows that the consequent of the conditional is not a proposition, as it must be if the conditional is to be well-formed. If, on the other hand, the variable 'p' is taken to range over propositions, then we get the same result:
4. If he asserts the proposition that snow is white, then the proposition that snow is white
which is also nonsense. Unless I am missing something, it looks as if Ramsey's redundancy theory cannot succeed in eliminating 'true.' It looks as if 'true' is an indispensable predicate, and thus a genuine predicate. This does not, however, show that truth is a genuine property. It merely shows that we cannot get rid of 'true.'