[hist-analytic] Roger's Questions about Anayticity
danny.frederick at tiscali.co.uk
Tue Jan 27 15:10:18 EST 2009
I almost agree with you.
Your claim was that the attribution of analyticity to a sentence is
synthetic. My objection was that the attribution of analyticity to a
sentence is either necessarily true or necessarily false - provided that the
sentence in which the attribution is made picks out by means of a rigid
designator the sentence to which analyticity is attributed.
The first question here is how the objection, which is about necessary truth
(and falsehood) relates to the claim, which is about analyticity. This is
not such an easy question. If we say that analyticity is truth in virtue of
meaning, we meet the issue of whether meanings are 'in the head.' When Kant
spoke of analytic truths, he assumed that (roughly) 'analytic' means true in
virtue of meaning and (since meanings are 'in the head') 'analytic' means
true a priori. Since Kripke, this has been untenable. 'Hesperus =
Phosphorus' is necessarily true, and it is true in virtue of the meanings of
the terms. But it is not a priori true, because the meanings of the two
singular terms are not fully transparent to us; that is, we only grasp so
much of the meaning of each (at least enough to use the term to talk about
its reference), but not enough to know that they refer to the same thing.
Thus we could say that 'Hesperus = Phosphorus' is analytic because, given
what its terms mean, it is impossible that it should be false (and thus it
is true in virtue of the meaning of its terms). But it is not knowable a
priori. So Kant might call it 'synthetic.'
Given this distinction between objective meaning ('outside the head') and
subjective meaning ('inside the head'), it seems to me that the
analytic-synthetic distinction is best dispensed with. Let us talk only
about necessary truths (and their negations) and non-necessary truths (and
their negations). As for 'a priori' and 'empirical,' we can consign those
designations to the dustbin of philosophical history for Duhemian reasons.
So I am kinda with Quine, except that I would insist on the distinction
between necessary and contingent truths (though we can never know for sure
which is which).
Now to the more minor points.
My 'rebuttal' did not hinge upon the claim that the only way of referring to
the semantics of a language is by means of a rigid designator. We always
have the option of a definite description. What my rebuttal depended on is
that an unambiguous singular term (rigid or non-rigid) referring to a bit of
language refers EITHER to uninterpreted symbols OR to interpreted symbols. I
was complaining that you tended to switch between the two. Uninterpreted
symbols can be given different interpretations (in which case they cease to
be uninterpreted symbols); but uninterpreted symbols cannot be necessarily
true. Interpreted symbols can be necessarily true; but if they are so, they
are necessarily so.
Similarly, if we mean by 'the French language' just the noises or marks that
people make when using it, then what we mean is something that has no
semantics. But if we mean by 'the French language' the interpreted,
meaningful noises or marks that people utter or inscribe when using it, then
we mean something that has a specific semantics. We can use 'the French
language' either way. But we should use it consistently whichever way we
choose. Then we will avoid the apparent paradox that
'.' is necessarily true
is itself not necessarily true.
I feel sure that this issue will have been fully discussed in modal logic
texts. Perhaps Steve Bayne knows where. Doesn't Kripke deal with it in his
semantics for modal logic? I've not read such stuff since 1987, so I am
More information about the hist-analytic