1 Carnap, presents a version of the linguistic theory in Meaning and Necessity (Chicago: University of Chicago Press, 1960), pp. 52ffGoogle Scholar, especially p. 62. See also Quine's, discussion in Word and Object (Cambridge, Mass.: M.I.T. Press, 1964), pp. 213ff.Google Scholar Quine does not adopt the linguistic theory. I should mention that the linguistic theory is easily modified to handle de re belief: ‘John believes of Mary that she is beautiful’ is read as asserting a relation between John, Mary, and the open sentence ‘she is beautiful’. The theory can also be modified to handle token reflexives: e.g., ‘John believed that it was raining’ is read as ‘John believed ‘It is raining’ I won't spell out the details. Henceforth, I shall speak as if all that-clauses were de dicto and eternal (i.e., free of token reflexives).

2 Here, as elsewhere, I have in mind only eternal English sentences. Notice that one need not speak English — or any language — to stand in the ‘believes’-relation to ‘the moon is round’.

3 Church's, arguments appear in ‘On Carnap's Analysis of Statements of Assertion and Belief’, Analysis 10 (1950), pp. 97–99.CrossRefGoogle Scholar Many different versions have appeared since then; I present here the best forms oft he argument that I've been able to construct—not always Church's version.

4 Ore presupposition of this paper is that such notions as synonymy and (unique) translation are well-defined, in some if not all contexts. Readers who believe—as I do not—that Quine has shown the contrary may wish to read the argument of this paper as a gambit against the lntensionalists: I try to show that, even if we grant Church the intensionalist machinery he needs for his argument, we can yet resist taking propositions as the objects of belief.

5 In Frege: Philosophy of Language (Harper & Row, 1973), p. 372.

6 The view of logical form here assumed is essentially Davidson's; by a semantical unit I mean any name, relation symbol, function symbol, connective, or quantifier. I should add that it is not in my view entirely out of the question for an analysis of logical form to discover a semantical unit not explicitly present in the original, but such cases are usually difficult to Justify. For example, if one takes the view that logical form is grammatical deep structure, it is somewhat plausible that (2) is the logical form of (1), but much less that (7) is: what sort of transformation could have removed the word ‘English’ in deriving (1) from (7)? (Only a transformation which would ignore the word ‘German’, even though ‘English’ and ‘German’ seem in all other respects to be in the same syntactic category.)

7 See Carnap's, Physikalische Begriffsbildung (Karlsruhe, 1926).Google Scholar A not implausible alternative to Carnap's analysis would introduce an abstract entity, the degree Fahrenheit (call it d); (9) would be read as asserting a certain 3-place relation to hold between the air, the number 50, and the entity d. Carnap's analysis is more useful for the analogy I want to develop.

8 In speaking of mental states I intend no irrevocable commitment to such intensional entities as states or properties. It is commonly the case that when we are quite in the dark about a subject matter, we quantify over the intensional in ways that later turn out to be eliminable. The intensional ‘There's something that argon and xenon have in common’ gets replaced by the extensional‘argon and xenon have full outer shells’. So perhaps here. In any case, such intensionality as appears here appears informally, in the metalanguage; our reading of such object-language sentences as (1) is extensional.

9 The most illuminating attempts I have seen to characterize the analogy occur in the writings of Sellars, W.; also in Harman's, Gilbert book, Thought (Princeton, 1973).Google Scholar

10 I want to add a brief remark about Davidson's ‘On Saying That’ (Synthese (1968), pp. 130–45) which seems to present a quite elegant solution to the problems I have been discussing. I say ‘seems’ because transposing Davidson's analysis of says that to an analysis of believes that is not as easy as one might think. In Davidson's analysis of ‘John says that the moon is round’ the referent of the word ‘that’ is taken to be, not a sentence, but a (concrete) utterance-event: for Davidson, the objects of indirect discourse are certain episodes of people speaking or wirting. One cannot however maintain that the objects of belief are certain episodes of people speaking or writing: we all have many beliefs which have never been and never will be expressed in any language. (Consider your belief that Carter is President or 107 a number. This one has now been expressed—no doubt for the only time in history—but there are many more like it.) A natural remedy would be to take the objects of belief to be, not actual utterances, but kinds of utterance — generic rather than concrete events. Generic events are of course Chisholm's states of affairs: they are a species of intensional entity not easily distinguished from propositions; this version of Davidson's theory would not be acceptable to Davidson, in view of his preference for extensionalist accounts. It is, however, worth taking seriously — for example, it is an intensionalist analysis which is immune to the ‘logical equivalence’ argument I gave above.