water is wet". If, by contrast we translate "water2" as "XYZ",

COGNITIVE SCIENCE 105
water is wet". If, by contrast we translate "water2" as "XYZ", then the proposal
reduces to " 'water 2 is wet' expresses the de dicto belief that XYZ is wet";
and so forth for other possible construals of "water2". These various alternatives
will be discussed severally further on.
These considerations may suggest that there is some problem about
construing the communicative intentions of English2 speakers vis avis " water 2
is wet" that does not arise when we try to explicate our communicative
intentions with respect to the homonophonic expressions of English. And
it is true, of course, that " 'water is wet' expresses the belief that water is
wet", though perhaps uninformative, is at least well formed. However, the
appearance of asymmetry is surely spurious. When we do the combinatorial
part of semantics, we have some justification for simply assuming that the
vocabulary of the object language is available in the metalanguage of choice;
we thus take the semantics of "good" and "actress" for granted and show
how the semantics of "good actress" arises therefrom (see [2]). But, of course,
we cannot do that when we are embarked on Putnam's project, which is
precisely that of lexicographic analysis. If there is an issue about what "water2"
means, and if it is question-begging to answer that it means water 2i then
surely there must be the same issue about what "water" means and it must
be equally question-begging to answer that it means water. Maybe what this
shows is just that lexicography is a mug's game; indeed, I strongly suspect
that that is true. However, that line is unavailable to Putnam, whose discussion
is explicitly "almost entirely about the meaning of words rather than about
the meaning of sentences" ([9], p. 216). It must, in short, be obvious that
if Putnam's examples make the relation between meaning and de dicto
propositional attitudes problematic for English2, they must also do so for
English. The home language cannot be viewed as privileged in this sort of
study.
Second gambit: "Water2 is wet" is used to express the de dicto belief
that water is wet. Reply: I take it that we can set this proposal to one side;
not because it is obviously false (on the contrary, I shall eventually argue
that, for all Putnam has shown, it may well be true) but rather because if
it is true, then there is no Twin Earth problem for us to solve. As we saw
above, if Putnam's example is a problem for cognitive science, that is because
it seems to show that molecularly identical people can have de dicto propositional
attitudes that differ in content. What invites this conclusion is the
"Gricean" assumption that linguistic forms which differ in meaning must
ipso facto differ in the propositional attitudes they are used to express.
By contrast, the present proposal is that whatever may be the case with the
meanings of "water is wet" and "water2 is wet", they are used to express the
same propositional attitude: viz., that water is wet. On this account, the only
moral to be drawn from Putnam's examples would be the irrelevance of the
semantics of natural language expressions to the individuation of the propositional
attitudes of speaker/hearers. (This is a moral which I shall eventually
endorse, though on a very narrow construal. I propose to take the sting out
of it by suggesting (a) that the content of a linguistic expression should be
distinguished from such of its semantic properties as its truth conditions;