Davidson wants to abstain from any attempt to define or further analyse truth (in as far as analysis is supposed to lead to a definition, or, if not a definition, an understanding in terms of other concepts). Moreover, he does not only reject a definition of truth, but also any attempt to capture something of its essence by means of general and vague formulae. Davidson argues that it does not make sense to try to define the most fundamental concepts we have, like truth. He considers any such attempt futile because of the circularity that is often involved in such definitions: when a philosopher tries to analyse a certain concept on its own, she has no choice but to take an understanding of the other fundamental concepts (equally potential matter of philosophical analysis) temporarily for granted in order to try to define the concept at stake in terms of the others. Davidson thinks there is a hopeless circularity in such an approach.46
To judge from what Davidson does say on the concept of truth, his alternative to a definition is to illuminate the concept not by defining it, but by describingwhy it is important, what role it plays in our understanding of the world, and (hence) how it is related to the other fundamental concepts that philosophy is concerned with (like those of meaning and of belief).
A particular reason not to try to define truth for Davidson may be not to fall victim to circularity himself. The reason is that Davidson more or less analyses meaning, a fundamental concept that he deems much more obscure than truth, in terms of truth conditions (although he does notdefine
meaning either). If a definition of truth then would (implicitly) rely on an understanding of the concept of meaning, we would end up with the kind of circularity Davidson thinks is problematic.
Davidson has another argument for the undefinability of truth: he takes it to follow from Tarski’s undefinability theorem.47 Tarski’s theorem says that true-in-L cannot be defined in L. Davidson seems to infer from this that a definition of the concept of truth is impossible because such a defi- nition would have to be stated in a languageL, and would then implicitly also define true-in-L, which contradicts Tarski’s theorem. It is questionable
44 Cf. EPT, p. 190. 45CTTK, p. 139 and TP, p. 55 46 Cf. FTDT, pp. 19-21. 47Cf. FTDT, p. 23.
whether that as such is a good reason for abstaining from a philosophical
analysis of truth. Natural language (the language of philosophical discourse) has inconsistent sentences anyway, which does not prevent us from philos- ophizing.48 Even if we know beforehand that we will not succeed in giving a satisfactorydefinition of truth, our failures may bring to light important features of the concept.49
However, maybe it is not a surprise that the concept of truth resists analysis if it is, as Davidson assumes, the core concept of rationality. For Davidson grasping the concept of truth and grasping propositional content are interdependent, as we saw. Having the concept of truth depends on lin- guistic communication, but understanding language again depends on un- derstanding truth. It seems to naturally follow from this that the only thing we can do is describing how truth carries over from language to language, by relying on a prior understanding of truth, as Tarski showed. Davidson writes that it gradually dawned on him that whereas Tarski took for granted an understanding of meaning to explore the notion of truth, he himself had in mind the reverse: taking for granted the concept of truth, he tried to shed light on meaning.50 Davidson thinks further analysis of truth cannot bring to light more than we already know:51
Truth, as applied to utterances of sentences, shows the dis- quotational feature enshrined in Tarski’s Convention T, and that is enough to fix its domain of application. Relative to a language or a speaker, of course, so there is more to truth than Convention T; there is whatever carries over from language to language or speaker to speaker. What Convention T [...] reveal[s] is that the truth of an utterance depends on just two things: what the words as spoken mean, and how the world is arranged. There is no fur- ther relativism to a conceptual scheme, a way of viewing things, or a perspective. Two interpreters, as unlike in culture, language, and point of view as you please, can disagree over whether an utterance is true, but only if they differ on how things are in the world they share, or what the utterance means.
If Davidson is right about this, it explains the mistakes of deflationists and correspondence theorists. Deflationists argue, based on the apparent
48The undefinability theorem did not prevent Tarski himself from giving a vague defi-
nition of truth in terms of correspondence, which he thought was given content to by his theory.
49
Davidson ridicules Plato’s attempts to define fundamental concepts: one of the few Platonic definitional achievements was to define ‘mud’ as ‘water + earth’. Cf. FTDT, pp. 19-21. Nevertheless, Gettier’s proof that Plato failed to define knowledge was considered highly significant and inspired sophisticated scrutiny of the concept.
50
Cf. Davidson (2001b), p. xvi introduction.
51
triviality of T-sentences and their resistance to analysis, that truth is not only transparent, but in fact trivial. Correspondence theorists, on the con- trary, have tried (in vain, if Davidson is right) to analyse the relation between meaning and ‘how the world is arranged’ in terms of entities in the world corresponding to sentences. A sentence, however, is not true because of some entity in the world, but simply because of the way the world is, and the only way to express the latter is by using the sentence itself (it is therefore not possible to analyse what the truth of a sentence consists of, since that brings us back to the sentence itself).52