Unsurprisingly, precisely the same sort of trouble arises for beliefs about water2

COGNITIVE SCIENCE 109
nevertheless not substitutable in de die to contexts; he will have to give some
account of why they are not. The trouble is, however, that the only conceivable
reason why one should not be able to make such substitutions is
that beliefs about contracts and beliefs about what is binding though verbal
are not identical beliefs.13 But one wants to ask how beliefs about contracts
could be distinct from beliefs about what is binding though verbal if, as Burge
assures us, "contract" and "what is binding though verbal" express the same
concept.
Unsurprisingly, precisely the same sort of trouble arises for beliefs about
water2 and beliefs about XYZ. If you cannot substitute "water2" for "XYZ"
salve veritate in de dicto belief contexts even though, by assumption, "water2"
means XYZ, that must be because beliefs about water2 and beliefs about XYZ
are ipso facto different beliefs. But if we now add the (presumably untendentious)
premise that "water2 is wet" expresses a belief about water2, we get
a contradiction of the proposal we have been investigating: viz., that "water2
is wet" expresses a de dicto belief about the wetness of XYZ. In particular,
we have (1) beliefs about water2 Φ beliefs about XYZ (in order to account
for the unsubstitutability of "water2" for "XYZ" in de dicto belief contexts);
(2) "water2 is wet" expresses a belief about water2 (by assumption); hence
(3) "water2 is wet" does not express a belief about XYZ.
Where we are is: there is a prima facie clash between the principle of
charity and Burge's assumption 3, but the way out is not to jettison the
substitution salve veritate of synonyms for synonyms in de dicto contexts.
An obvious alternative, however, would be to give up the principle of charity
in these cases. So let us look at that.
This is not a decision to be taken lightly; charity should not be confused
with mere politeness. The point is—to switch the discussion back to Putnam's
example—it would be unreasonable for us to take English 2 speakers to be
expressing self-contradictions when they utter things like "water2 is not XYZ".
There might, for example, perfectly well be a point in the development of
their chemistry when water2 is not XYZ is the rational thing to believe given
the evidence. It can, no doubt, be rational to entertain a belief that is
necessarily false; but it is hard to see how one could rationally entertain a
belief with the de dicto content P and ~P. Could the weight of the evidence
favor a contradiction?
It may be felt, however, that this sort of argument is too good. For,
if the principle of charity precludes taking "water2 is wet" to express the
belief that XYZ is wet (on pain of attributing too many inconsistent beliefs
to English2 speakers) does it not also prohibit taking "water2" to mean XYZ
(on pain of attributing too many inconsistent sayings to English2 speakers)?
If "water2" means something like XYZ, then it looks as though the form of
words "water2 is not XYZ" is going to be something like analytically false.
I am perfectly prepared to accept this argument since I am not wedded
to Putnam's intuitions about the meaning of "water2". If, however, you do
not like it, there is a way of avoiding it. You can argue that, given Putnam's
assumptions, it is not obvious that the principle of charity should be applied
to what we say since what we say need not be, in any very direct way, an
expression of what we de dicto believe.