1 Lewis, David, ‘General Semantics’, in Semantics of Natural Language, Davidson, D. and Harman, G., eds., Dordrecht, 1972, p. 169.CrossRefGoogle Scholar I choose Lewis as my stalking horse here because of his clarity and forthrightness. Many others might have been chosen.

2 Lewis, , op. cit., p. 173.Google Scholar

3 Katzand, cf. J.J.Fodor, J., ‘The Structure of a Semantic Theory’, in The Structure of Language, , Katz and , Fodor, eds., Englewood Cliffs, 1964.Google Scholar

4 This seems to be the position of Donald Davidson, at least in ‘Truth and Meaning’.

5 Lewis, , op. cit., p. 174.Google Scholar

6 Actually, I think what we can all pretty well tell is much less general than this-much more specialized than what is needed to determine what Lewis is calling an extension. But that point can be allowed to pass for the present.

7 Lewis, , op. cit., p. 174.Google Scholar

8 I am assuming that, e.g., the extension of a sentence on an occasion corresponds to whether, on that occasion, it, or what is said in producing it is true. Otherwise, one can always stipulate all sorts of extensions for sentences, but the connection between such stipulations and what the sentences mean would be quite obscure. This tactic will be discussed in more detail in the next section, in connection with the interpretations of predicates.

9 I have recently discovered that David Kaplan draws a distinction very much like this. Those elements of meaning related to factors of the first kind he calls ‘character’; those related to factors of the second kind, i.e., those yielded on an occasion by factors of the first, he calls ‘content’. Kaplan is concerned with the distinction primarily in the cases of demonstratives and indexicals. I am concerned with it in language in general. My claim is that in the general case, expressions don't have definite characters. A fortiori, they don't have contents even on occasion, though what is said via them might. The meaning of an expression is nothing like a content. That is another way of saying why extensionalist semantics is mistaken. [I do not specifically claim here that Kaplan is wrong about indexicals, and in any case I do approve of his distinction. I think, however, he is overly concerned with associating conditions with forms of words.]

10 Lewis, , op. cit., p. 175.Google Scholar

11 Note, incidentally, that the above claim about ‘the boy’ can't be quite correct as it stands. ‘The boy" certainly isn't restricted to referring to what is a boy, but in the right context (e.g., an historical narrative) might at least refer to what was or will be a boy. I think this forebodes deeper difficulties with this kind of project, but for present purposes I will leave that line to one side.

12 I have attempted just that, however, in Travis [1976].

13 If you like, you can suppose that Max isn't puce, although I don't think that's necessary. Harry, who obviously hasn't noticed Max's colour might later say (correctly) ‘Oh, good heavens, look! He's puce too.’

14 For one use of this term, cf. Kripke, S., ‘Speaker's Reference and Semantic Reference’, Midwest Studies in Philosophy, v. II, Morris: 1977.Google Scholar

15 Words mean things, and so, in another sense, do people. But it is what is said that may be either true or false. Concentration on meaning as opposed to saying, has, I believe, led to a misunderstanding of both, largely responsible for the above dichotomy - on the one hand, the erroneous view that what an utterance means (i.e., how it is to be taken) is largely a function of what's inside the speaker's head, on the other side, misunderstandings of the nature of truth (or what makes things true).

16 One certainly might try. The most likely way is by trying to distinguish what is strictly or literally said from what passes as said, e.g., through some sort of ‘implicature’. One problem is that it is hard to identify any plausible ‘strict’ or ‘literal’ use of the words which will do the job in simple cases like the above. Then if one stipulated such a sense, this line would ultimately prove unfruitful, largely for reasons that will become clear through the discussion of other maneuvres. At any rate, I leave this idea to one side for the present, and take the data about what's true or false in the above cases at face value.

17 The requisite condition will, at any rate, have to be equivalent to this. I will not repeat why it would be semantically trivial to say something like, ‘x is in the extension of ‘is green’ on occasion O iff on O ‘is green’ attributes P and x is in the extension off iff C One might get by with a version of 2 which doesn't explicitly mention propertiesjust in case O determines what property is attributed. I mentioned properties in 2 just to emphasize that that is what O must determine.

18 So note, that, inter alia, on this advice, truth becomes a relative matter-relative to who is doing to assessing. I don't think that this is a bad consequence, merely that the advice misplaces the relativity. Note that on this interpretation, extensionalist semantics shifts its attention away from truth, extensions, etc., to predicting what various people would regard as truths, extensions, etc. Whatever the merits or demerits of that, at least the result is not extensionalist semantics in the classical {e.g., Lewisian) sense.

19 I am not raising a question about the utility of some such notion in defining valuations, e.g., for a first order predicate calculus. The question I am raising is about natural languages (and natural thoughts). That such a notion is useful in formal logic should supply some clue as to what logic is about -e.g., that calculi are not skeletal languages.

20 One might also explain the meaning of the words ‘is green’ to someone by showing him cases of clearly green things. But the success of such activities depends on sociological facts in a way which doesn't lead to coherent and correct truth conditions. To understand what he is being taught, the learner must imagine the right sorts of occasions - ones where the words would be used to predicate to those things what is true of them. The sociological fact (for successful teaching) is that the right sort of occasions are easily imagined - so much so that it is easy for us to forget, when doing semantics, that that is what we are doing.

21 However, certainly not in all cases. Suppose Max and Sam are sorting blue paint chips into ones with greenish and ones with reddish overtones. Then Max might hold up a blue paint chip with the right greenish cast, and say truly of it, ‘This one is green.’ I think it is rash to conclude (though not out of the question) that the blue paint chip might correctly be described as being green.