Copyright 1996 and 1999 by Russell Eliot Dale All rights reserved
1. Lewis (1975), p. 169.w

2. Lewis (1975), p. 165.w

3. Lewis (1975), p. 167.w

4. Ibid.g

5. Lewis (1975), p. 187.

6. Ibid.g

7. Lewis (1969), p. 177.}

8. See Hawthorne (1989), Lewis (1991), Hawthorne (1992), and Schiffer (1993). In Loar (1976), p. 159 and in Loar (1981), p. 256, Loar addresses a related problem for a proposal of his of the actual-language relation. The solution he suggests in those places is essentially the same as that suggested later by Lewis in Lewis (1991). See section 5.5 below for a brief discussion of Loar's theory of the actual-language relation and his solution to the present problem. Martin Davis also raises a version of the problem and offers Loar's solution to this problem in Davis (1981), pp. 15-16. Davis cites Platts (1979) and Loar (1976) as sources of the discussion of this problem and explicitly credits Loar (1976) with the solution he suggests.

9. I will take issue with this claim in the section 2.2.4 below.

10. Since right now I am considering only whether [O] states a counter-example to [L'], I will talk only of a convention of truthfulness in a language and not of a convention of truthfulness and trust in a language. I will discuss [O] as an objection to [L] later.

11. I am not really sure that I think [B] is valid on any reading except, perhaps, where the phrases "a bachelor" and "a male" are read de re - and even there, the notion of validity is some sort of informal notion, obviously, and therefore somewhat questionable to my mind. But, suppose Edna believes that Gilbert is a bachelor because she heard David say that Gilbert was a bachelor. But for some odd reason she believes that Gilbert is female. Or suppose that Phil believes that he has arthritis because his doctor told him that he does, but he does not believe that he has a disease of the joints because he has some odd belief about the nature of arthritis. These would be cases where the inference read de dicto would seem to fail. Anyway, I consider these cases for the sake of thoroughness, but I am prone to be sceptical of a notion of validity based on 'sense-containment' relations. See Katz (1986) for a discussion of such cases.

12. Quine's notation in Quine (1955) is also helpful for seeing the point that I make here as well as some points that I will be making below.

13. See Loar (1976), p. 158, Loar (1981), p. 256, and Schiffer (1987), p. 259, where they paraphrase different but related attitudes with de re quantifier expressions as I do here.8

14. Recall that a language is understood as a function from expression types to propositions. A single pair, therefore, of a sentence type and a proposition will be a language. Thus, I can speak of truthfulness in a single such pair.i

15. As well as the other inferences required for it to follow from [S] that there is a convention of truthfulness in L+.

16. Lewis (1969), pp. 182-183. Lewis is using a notion of language here that is slightly more complicated than the one he uses in Lewis (1975) and that I am using, but that doesn't affect anything in the present discussion. The same objection is considered in Lewis (1975), pp. 180, and Lewis answers it in pretty much the same way there as he does in Lewis (1969).

17. See Heytesbury (c. 1339a), Heytesbury (c. 1339b), Bocheski (1961), Knuuttila (1982) and Tweedale (1982). I thank Tereza Saxlova for the last two of these references. Also helpful are Chisholm (1962), Sosa (1970), and Geach (1964).n

18. Lewis (1969), p. 65. Lewis cites Quine (1955), Montague and Kalish (1959), Lewis (1968), and Kaplan (1969) at the end of the paragraph from which this quotation is taken. But none of these papers, it seems to me, sheds any great light on what Lewis is really bothered about.-

19. Nothing that I am saying here is meant to suggest that there is nothing problematic with the view that I am ascribing to Lewis. I think that there is quite a bit that is problematic with it. But that doesn't matter to the point that I am making right now. See Loar (1976), p. 158 and Loar (1981), p. 256, for an obvious but serious problem with the view that I am ascribing to Lewis.,

20. In Loar (1976) Loar indicates that he understands Lewis as I have: "David Lewis has suggested that the mutual knowledge of the relevant regularity involves knowledge of all the sentences of L in sensu diviso...." (p. 158). See also Loar (1981), p. 256. And, again, see Schiffer (1987), p. 259, where Schiffer runs through an argument like mine that concludes, more-or-less, that Lewis needs to replace references to languages via singular terms with references to languages via de re quantifier phrases.

21. Henceforth I will leave off the "interest in communication"-clause. I will say some things about this clause in section 5.3 below.

22. Lewis (1975), p. 167.w

23. When I speak of actual usage here, I am not using "usage" in the more comprehensive way in which it is sometimes used in which it includes all sorts of internal sentence-processing matters. I mean something wholly external here.m

24. From now on I will just speak of Lewis's theory, even though he had more than one: what I say from here on actually applies to all of them.

25. I have in mind here Schiffer's counter-example to this view whereby processing, and thereby meaning, proceeds via a non-semantic translation device of some sort.

26. See Burge (1975).s

27. Burge (1975), p. 252. There are other counter-examples to Lewis's notion of convention in this paper as well, and I think that they are all quite good.

28. Schiffer (1972), p. 154. See Schiffer (1972), pp. 30-31 for Schiffer's definition of his notion of mutual knowledge*. For present purposes Schiffer's mutual knowledge* can be read as the same as Lewis's common knowledge.

29. See Schiffer (1972), chapter 5. Schiffer's gloss of convention is found there on p. 154.

30. Grandy (1977), p. 133.x

31. I will speak in terms of "common knowledge", but I am to be understood as talking about both Lewis's common knowledge and Schiffer's mutual knowledge*.+

32. See Lewis (1969), p. 59 and Schiffer (1972), p. 150.

33. See, for example, Gilbert (1983). Perhaps Schiffer's account of convention escapes Gilbert's problems for Lewis's account. For more technical problems with Lewis's use of game theory in motivating the account of convention he ultimately settles on, see Gilbert (1981).

34. See Schiffer (1993), pp. 232-233.

35. See Loar (1976), pp. 151-161 and Loar (1981), pp. 256-260. Loar's theory is related in important respects to Schiffer's theory of expression-meaning in Schiffer (1972), chapters 5 and 6. It is also somewhat related to the so-called "IBS" accounts of expression-meaning that Schiffer discusses in Schiffer (1987a), pp. 249-255 and Schiffer (1993), pp. 239-242, though these accounts don't explicitly use the notion of convention.r

36. He refers readers to Lewis (1969) and Schiffer (1972), chapter 5 for the details of the notion of convention.

37. Loar (1976), p. 153. Loar, later in Loar (1976) and in Loar (1981), unpacks the notion of serious circumstances he uses here. I am not entirely sure that Loar's notions of being free for meaning a proposition and of associative and non-associative improbabilities which he uses to clarify the notion of serious circumstances are unproblematic. But since I won't take issue with Loar here about this I won't bother using the unpacked version of his theory in my discussion.

38. Loar seems to have changed his mind about this somewhat by Loar (1981).-

39. This is essentially the suggestion that Lewis makes much later in Lewis (1991).

40. See Schiffer's discussion in Schiffer (1993), pp. 232-239.

41. See Loar (1981), p. 257. Loar didn't discuss this sort of counterexample in Loar (1976).

42. In his earlier piece, Loar seems to think that talk of cognitive relations to CMTs is out of place in a theory of meaning, even if it could be that, in fact, we do use our languages by standing in such relations to CMTs: "There may be a Chomsky sense of knowledge - having an internal representation - in which a speaker knows the rules of his language, but that is a psychological hypothesis and, however reasonable it is, we do not want to build it into an explication of what it is for L to be the language of P" (Loar (1976), p. 158). But in Loar (1981), Loar suggests that a fuller theory of meaning cannot be provided until we know more about our own psycholinguistics and how we might be related to CMTs for languages: "I suggest the Chomskyan idea of the internalization of the generative procedures of a grammar has got to be invoked to make sense of the entrenchment of a grammar, and therefore to make sense of literal meaning. The exact force of this can't be spelled out antecedently to a detailed psycholinguistic theory" (Loar (1981), p. 259). Thus, Loar seems to be going back on the earlier view that he seemed to have had about people standing in cognitive relations to CMTs and how this sort of possibility should figure into a theory of meaning.z