1. Of course, [D'] can be turned into a true theory for another class of expressions, viz.,
the base-4 Arabic numerals, provided we change the second axiom of [D'] so that it reads
"For any base-4 Arabic numeral x, x conventionally expresses fI(x)". The resulting
theory's definition of the function fI will have five useless clauses - the ones for the
simple symbols "5", "6", "7", "8", and "9" - but this doesn't affect the theory's truth with
respect to the base-4 Arabic numerals. But none of this matters to what I am doing in
the main text. The point being that [D'] simply entails falsehoods about the Arabic-decimal numerals.
2. In the case of a CST that contains as an extra addition the statement of an analysis of
its semantic notions, you still need an independent understanding of the semantic notions
to judge the CST for truth or falsity since you would need such an understanding to know
whether the analyses offered were correct.
3. There are, of course, two notions of definition being used here. One notion applies to
linguistic expressions, the other to notions, that is, ideas or properties. I am just speaking
loosely, but things should be clear enough.k
4. I take a quoted sentence to count as a structural description of the sentence here. But
maybe that is not the best way to look at things.
5. A problem for this characterization of a CTT is the following sort of theory: "(s)('s'
is true in L just in case s)" where "" is a substitutional quantifier, "L" is a name or
description of some language L and the substitution-class for "s" is the set of L-sentences.
This is a finite theory that entails a truth of the appropriate form, but it is not what I want
to count as a CTT. But I will pretend I haven't noticed this.^
6. Well, perhaps you can invent a language with finitely many sentences and give a CMT
for it. For example, Consider J-English which consists of the one sentence "June likes
to dance" which means in J-English that June likes to dance. Here is a CMT for J-English: "'June likes to dance' means in J-English that June likes to dance". Okay. For
the simplest languages, maybe, then, you can give a CMT. Read me in the text, then, as
meaning "the simplest BIG language" where a BIG language is one that is too big to have
its members enumerated by us for mundane practical reasons. But a BIG language doesn't
have to be infinite.[
7. See Davidson (1967). But the seeds of the idea are already in Davidson (1965).
Probably his clearest statement of the project he envisioned, though, is found in Davidson
(1973). Note that even if Katz and Fodor antecede Davidson with respect to concern for
the compositionality of natural languages (see Katz and Fodor (1963)), still, their early
proposal for a theory that deals with the compositional features of natural languages is
probably better characterized as compositional-translation theory than as a compositional-semantic theory. I will talk about compositional-translation theories and attempts to use
such a theory to explain meaning when I turn in chapter 6 below to Schiffer's recent
theory.
8. See Davidson (1968).v
9. See Schiffer (1987b), chapter 5, "Sententialist Theories of Belief" for a somewhat
compelling critique of Davidson's theory of propositional attitudes.
10. See Peacocke (1976). But note that Peacocke redefines the notion of a language for
his talk of an actual-language relation. So, the truth is, he is really talking about a
different notion from the one Lewis spoke of. But I am being so general in the discussion
right now, this doesn't really matter much.
11. Field (1972).o
12. Fodor (1987), p. 98.k
13. See Soames (1984).t
14. See Fodor (1987) and Fodor (1990), chapters 2, 3, and 4.
15. Schiffer (1992).r
16. This might seem slightly uncharitable to Fodor since he has a sophisticated view of
the logical form of propositional-attitude ascription that blocks the Frege-style point in a
case like in the Schiffer story. See Fodor (1989b). But, Fodor's understanding of
propositional-attitude ascription is independent of his views about the naturalization of
meaning and there are a number of serious problems with Fodor's view of attitude
ascription. See Schiffer (1992).
17. Chomsky (1980), p. 84.x
18. Op. cit., p. 82.r
19. Op. cit., pp. 69-70.v
20. Ibid.g
21. See Fodor (1975), pp. 119-122, and Schiffer (1987b), chapter 7. Fodor, however, has
insisted that in spite of the possibility of the following sort of explanation of language
processing, still there must be CMTs for Mentalese and natural languages. His arguments
will be addressed in just a bit.
22. It is not necessary to take a LOT here to be the very weak sort of LOT that I
discussed at the end of chapter 3. A LOT here can be as strong and as interesting as you
like for the present discussion. That is, the form of the following argument is "Suppose
that there we think in a LOT on any version of the LOT hypothesis you like...."
23. See Fodor (1989) and Fodor (1991).
24. See section 4.3.2.t
25. In other words, the argument Fodor has in mind is something like the following.
scusse(1) Naturald of-language users have impressive skills that must be explained.
you
li(2) Certain CST stories explain these skills and certain syntactic arg-translation-device
stories do as well.
a LOT(3) There are no other stories besides these that would explain the skills in
question.
ber of(4) If Mentalese enjoys a CST, then so will any natural language if the syntactic
-translation-device story.
theory(5) Mentalese enjoys a CST.mant
c theo(6) So, whichever of our two stories that explains natural language processingsuch
turns out to be true, natural languages will enjoy CSTs.
w to SSo the really controversial premise here is (5), that Mentalese enjoys a CST.anin-
26. Schiffer (1991), p. 191.z
27. Fodor (1991).o
28. Op. cit., p. 306.s
29. And this is not yet to mention Fodor's attitude towards reduction in others of his
writings.
30. Op. cit., p. 318.s
31. Fodor does speak of identifying the properties assigned to expressions by a
supervenience theory with the semantic values of the expressions. (See, for example,
Fodor (1991), p. 306: "...the temptation might be overwhelming to simply identify the
semantic values of words....") Usually the expression "semantic value" is used to talk
about what a CMT or a CST assigns to expressions. But Fodor clearly doesn't have this
sort of thing in mind when he talks about identifying supervenience properties with
semantic values. He still, I imagine, thinks the semantic value of "cow" will be
something like cows, not the physical property assigned to "cow" by a compositional-supervenience theory which will surely not be anything like cows! I believe that Fodor
in talking about identifying the properties assigned to expressions by a compositional-supervenience theory with semantic values is merely trying to talk about identifying the
physical properties assigned to expressions by a compositional-supervenience theory with
meaning properties, that is, with properties like the property of meaning cow or the
property of meaning that snow is white.
se I tIn one place Schiffer responds to Fodor's talk of ind'semantic values in the context of
this discussion as if Fodor were talking about what is usually meant by "semantic value".
No harm in doing so, over all, but to take Fodor in that way, it seems to me, is more to
construct a straw-man than anything else. See Schiffer (1994b), p. 322, note 18.7