Third gambit: "Water2 is wet" is used to express,

106 J. A. FODOR
and (b) that content is—though truth conditions are not—a construct out of
the communicative intentions of speaker/hearers.)
Third gambit: "Water2 is wet" is used to express the de dicto belief
that what is called "water2" around here is wet ("around here" being used
to index Earth2). Reply: I put this one in because some things that Putnam
says about the indexicality of kind terms may suggest it. It is, however, highly
implausible and I very much doubt that Putnam actually has this solution in
mind. For one thing, it is too metalinguistic sounding; whatever belief "water2
is wet" is used to express is surely one that animals, prelinguistic children and
(nb) people who have never heard of the word "water2" can share; and none
of this would be so if one accepted the present proposal about what "water2
is wet" is used to say. To put this point quite generally: the belief that
"water2 is wet" expresses must turn out, on any acceptable analysis, to be
identical with the belief that water2 is wet (whatever belief that turns out to
be; see above). But, surely, it must be possible to have the belief that water2
is wet without having any metalinguistic beliefs at all. The belief that water2
is wet is a belief about water2, not a belief about language.
Such considerations suggest that the indexicality story does not provide
necessary conditions for being the belief that "water2 is wet" expresses. What
is more important, however, it does not provide sufficient conditions either.
Consider the parallel proposal that the (English) formula "tigers have stripes"
expresses the belief that what are called "tigers" around here have stripes.
Well, if this is true it must follow that having the belief that what are called
"tigers" around here have stripes is sufficient for having the belief that tigers
have stripes; for surely, that tigers have stripes is the belief that "tigers have
stripes" is used to express. But that's no good for the following reason: you
could have the belief that what are called "tigers" around here have stripes
and not have the belief that tigers have stripes if, for example, you happen
to think that what are called "tigers" around here are pyjamas. In that case,
you would have one false belief (viz., that pyjamas are called "tigers") and
one true one (viz., that pyjamas have stripes), the conjunction of which is,
patently, not equivalent to the belief that "tigers have stripes" is used to
express.
Similarly, with bells on, for such proposals as that "water2 is wet"
expresses the de dicto belief that this stuff is wet ("this stuff being used
to index some water2), You could believe that this stuff is wet while believing
that this stuff is, say, tomato juice. In that case, believing that this stuff is
wet would not be believing that water2 is wet, even though this stuff is, as
a matter of fact, both wet and water2.10
Fourth gambit: "Water2 is wet" is used to express the de dicto belief
that XYZ is wet. Reply: I think that this is what a lot of philosophers would
say who share Putnam's intuitions about how lexicography should be pursued.
For example, Burge [ 1 ] has recently accepted the corresponding solution for
a class of examples which, as he remarks, are in important respects quite similar
to Putnam's but do not involve terms for natural kinds. Nevertheless, it seems
clear to me that quite familiar considerations preclude taking this line. Since
the reasons for denying that "water2 is wet" expresses the de dicto belief that
XYZ is wet are equally reasons for not accepting Burge's account of the