What is common to Burge's example and Putnam's,

108 J. A. FODOR
a principle of charity which operates to prohibit accusing one's fellows of
inconsistency in this flagrant and inflammatory way.
What is common to Burge's example and Putnam's is that in each case
something that is taken to be part of the meaning of an expression that
speakers use is nevertheless assumed to be something that the speakers need
not grasp. In Burge's example, it is a necessary truth constitutive of the
meaning; in Putnam's example, it is the extension. My point is that that is all
right so far, but you get into trouble with the principle of charity if you
also make the assumption that, in effect, what a verbal form means is interchangeable
with the concept that it expresses: in the present case, that
"contract" expresses the concept contract or that, mutatis mutandis, "water2"
expresses the concept XYZ. For then all sorts of innocently false statements
("verbal contracts do not bind"; "water2 is not XYZ") are going to be taken
to render self-contradictory beliefs, and this the principle of charity forbids.
I have the impression that there are a lot of philosophers who think
it is all right to say that "water2" means XYZ or, to continue with Burge's
example, that Jones uses "contract" to express the concept contract. It may,
therefore, be worth making explicit the ingredients of the present bind and
considering in some detail the various options Burge has for getting out of it.
Burge's difficulties arise from the interaction of the following five
assumptions:
1. Verbal contracts bind is constitutive of the meaning of "contract".
2. The meaning of a word is a construct out of the concept it expresses;
in particular, words that express the same concept are synonymous.
3. When Jones says "contract" he expresses the same concept that we do
when we say "contract".
4. The principle of charity.
5. The intersubstitutability of synonymous expressions in de dicto belief
(or "says that") contexts.
Of these assumptions, Burge is more or less explicitly committed to 1 and 3,
and something like 2 would seem to be required if the relation expressing
the same concept as is to be distinguished from weaker forms of semantic
equivalence that words may enter into.11 Assumptions 4 and 5 strike me
as pretty plausible, but Burge might want to try ditching one or both. Neither
possibility seems very attractive, however, as we are about to see.
Suppose you give up the principle that you can substitute synonyms
for synonyms in de dicto specifications of beliefs. That would suffice to
block such inferences as from "Jones believes that (says that) verbal contracts
do not bind" to "Jones believes that (says that) what binds when not verbal
does not bind when not verbal"; similarly, mutatis mutandis, it would serve
to block the inference from "Jones2 doubts that water2 is XYZ" to "Jones2
doubts that XYZ is XYZ". The trouble is, however, that if you do block these
inferences, it is hard to see what is left of assumption 1 (or, mutatis mutandis,
of the claim that "water2" means XYZ.) Claims about lexical meaning seem to
turn very largely on the issue of what substitutes for what in de dicto contexts.
Moreover, it will not do for Burge to just say that "contract" and "what
is binding though verbal", although they express the same concept,12 are