Davidson on Truth, Meaning, and Conceptual Schemes

Rorty defends his claim that we need no theory of truth, and his consequent rejection of the entire discipline of epistemology, by appealing to Donald Davidson’s philosophy of language. Davidson is a philosophical descendent of Quine, and thus, he bears some of the imprint of pragmatism. Davidson himself abjures the epithet "pragmatist," but Rorty argues that he deserves it nonetheless.

We must begin with Tarski’s famous Convention (T):

(T) "S" is true iff S,

where "S" is a variable ranging over sentences. Tarski offered Convention (T) as a constraint on theories of truth: any theory of truth that doesn’t entail (T) is surely wrong. Now, think of a theory of truth not as an analysis of the concept of truth or of the meaning of the predicate "… is true," but rather, as a full blown theory of some particular language, L. Such a theory will take it upon itself to determine which sentences of L are true. That is, it will entail a "T-Sentence" that instantiates (T) for each sentence of L. So, a theory of truth for the English language should entail, inter alia:

"The cat is on the mat" is true iff the cat is on the mat.

Now, suppose that Jones, who speaks English, is developing a theory of truth for German. If her theory is correct, it must entail this T-Sentence for German:

"Der Apfel ist rot" is true iff the apple is red.

A complete theory of this sort would be pretty impressive.

In fact -- and here is a basic Davidsonic insight, or claim, depending on how sympathetic you are to his views -- such a complete truth-theory for German would seem very much like a translation manual for that language. It would tell Jones just when every German sentence is true; it would correlate German sentences with states of affairs that Jones can recognize by way of her own linguistic capacities. Davidson suggests that if we are "field linguists," such a truth-theory as just described is all we can have. A "field linguist" is someone whom we suppose has gone into a community of speakers of a language under study. This linguist has no antecedent knowledge of this language. Davidson labels this enterprise "radical interpretation." The field linguist cannot describe the "inner experiences" of speakers of L, nor tell us antecedently what the significance of their behavior is. All the field linguist can do is correlate the linguistic behavior of speakers of L with situations experienceable by them in the world around them. And these situations would be, of course, characterized in the linguist’s own language.

Davidson develops his philosophy of language in two ways that Rorty seizes upon. First, such a truth-theory is exhausted by the T-Sentences it entails and does not offer us anything that might be called an "analysis of the concept of truth." This is Rorty’s "less is more" theme. And there’s more: a truth-theory does involve spelling out how nouns and other referring expressions of L correlate with objects, as well as how predicates correlate with properties or classes of objects. (It must do that, because language is compositional: its sentences are composed of, inter alia, referring expressions and predicates.) But such correlations are (a) merely what Rorty calls "fallout" from the truth-theory, i.e., adjustable instruments for making the truth-theory work, and (b) no more analyzed by the theory than is truth.

Second, Davidson puts this line of thought to dramatic use in his argument in "On the Very Idea of a Conceptual Scheme." He is there concerned with the thought, suggested by Kant and Hegel, and popular now amongst many philosophers and especially social scientists (anthropologists come in for especially rough treatment by Davidsonians), that different communities "see the world through different lenses," that they experience the world mediated by different conceptual schemes. A conceptual scheme is a network of concepts that interprets experience, or the world, for us. Davidson wants to argue that the very idea of a conceptual scheme is confused: there cannot be divergent conceptual schemes, and thus, the concept has no utility to us, or the field linguist.

The argument:

  1. Suppose that Jones confronts some behavior that she thinks is linguistic behavior. What’s her job then? It is, consonant with the Davidsonic philosophy of language, to generate a truth-theory for the language.
  2. Now, suppose (for reductio) that Jones develops a truth-theory for L according to which a great many of the utterances of the speakers of L come out false. E.g., when confronted with an apple, the speakers of L typically say something that Jones translates with her sentence, "Hey, there’s a tiger!" We would think that Jones has done a poor job in building a truth-theory for L. This suggestion is called the "Principle of Charity:" it is the linguist’s duty, in building a truth-theory for L, to make most of utterances in L come out true.
  3. What if we don’t seem able to construct such a "charitable" theory for L? Davidson suggests that we would then come to have doubts that the behavior we thought is linguistic really is. We are confidant that German and Chinese are languages, not just because they are spoken and written, but also because we can construct enough of the truth-theory for them that most of what speakers of those languages say is true.
  4. Davidson infers that any behavior that we can recognize as linguistic involves sentences most of which are true.
  5. If that’s right, then we can translate what speakers of L are saying by way of true sentences of ours. They don’t see the world (significantly) differently than we; they see it largely just as we do.
  6. So, all languages largely use our concepts.  
  7. Ergo: Since all languages largely use the same set of concepts, there is no sense in talking about alternative conceptual schemes. There cannot be alternative conceptual schemes. There is only one set of concepts, ours, i.e., the ones that accurately describe the world around us.

Furthermore, it follows from the conclusion that to the extent that we and the speakers of L differ, one of us is wrong. (This is the Davidsonic rejection of relativism.)

Suggestions for criticism:

  1. Davidson confuses whether we have reason to believe a community is speaking a language with whether they are in fact doing so. This is a covert, verificationist premise, not likely to convince Kant, at least.
  2. In construing a truth-theory for L defined by Convention (T) as a translation manual from L into the theorist’s language, Davidson is construing meaning as "extensional," "truth-functional:" he is construing any two sentences that mutually imply each other truth-functionally, such as, "grass is green" and "snow is white," as synonymous. This is obviously wrong. (Davidson does entertain this objection in "Truth and Meaning.")


Davidson's "Truth and Meaning" (1967) and "On the Very Idea of a Conceptual Scheme" (1974), are both collected in his Inquiries into Truth and Interpretation (Oxford, 1984).