Morton White on Analyticity
by Steve Bayne
What follows is an account of what Morton White says about analyticity in the chapter "Analyticity and the Synthetic" in his book _Towards Reunion in Philosophy_. Atheneum. It is somewhat compressed as is intended only for reader reference.
There are many things in other sources relevant to the discussion which preceded it. Quine's _Word and Object_. MIT. 1960. Among the most relevant are discussions in Arthur Pap (_Semantics and Necessary Truth_. Yale. 1959) and Sellars ("Synthetic Apriori"). I am deferring discussion of these sources and simply presenting White's illuminating and historically important views.
One may ask whether all logical truths are analytic and miss much of the point of introducing a term like 'analytic' in the first place. One may, for example, miss its possible applicability to a priori knowledge generally even if we assume that the truths of logic are a priori. This may very well be because by 'a priori' we do not *mean* 'analytic'; and if we did, the analyticity of 'All a priori statements are analytic' would be just that of 'All a priori statements are a priori statements'.
One doesn't attempt to show a purportedly analytical statement false by counterexample, any more than one tries to show that 'Not all bachelors are unmarried' by discovering a married bachelor. These are the things that I believe compel White to make the observation that many who attempt to refute the claim that all a priori knowledge is analytic do so by counterexample: these philosophers do *not* believe in the synonymity of 'a priori and analytic'. White says (p. 135)In a curious way the very formulation of the issue can lead philosophers to use words whose meaning is at issue, for once we ask whether 'a priori' is synonymous with 'analytic' we we might also be led to ask whether the statement 'All and only a priori true statements are analytic' is itself a priori and analytic. (ibid.)The common man surely understands words like 'touch' and 'taste' and along with these 'experience'. That same person might get introduced to 'a priori' as something we "don't need to get to know" by experience. So one can come to know as a priori what is *not* known a posteriori (that is, by experience). I take it therefore that there is a misprint at p. 134 and that in...so the predicate 'a posteriori' suggests 'non- a posterioriwhat White wants is 'a priori' in place of 'a posteriori'. The important point is that we can introduce a predicate without even knowing whether it is satisfied. But now let's take a look at 'analytic' and ask how the common man may learn this from already understood terms in the way he learned the meaning of 'a priori' with the help only of terms such as 'experience' and 'not' (plus a couple other words like 'know').
Such a person may have it pointed out that knowing some sentences to be true requires knowing only a couple of synonyms. For example, knowing 'All vixens are foxes' requires merely knowing that 'vixen' means ' female fox'. Such sentences might be called 'analytic'. White, next, makes his smoothest move yet. If I come to know the words 'a priori' and 'analytic' in this fashion, I do not ipso facto know a truth is analytic if and only if it is a priori. What I *may* know is that if it is analytic it is a priori, but I do *not* know that it is analytic *if* and only if it is a priori.
Given that the issue is whether all and only analytic sentences are a priori then we may argue against the positivist position that a truth is a priori only if it is analytic by arguing that there are non-analytic sentences, synthetic ones, that is, which are a priori. White distinguishes four positions, but he pusues only two in some detail.
This philosopher says "that our conviction that a statement is a priori is always based on identity of meanings." The platonist who is a positivist will hold that we begin with a logical truth (known a priori) and then substitute terms with *identical meanings* in arriving at analytical truths. So given that the meaning of 'vixen' is identical to the meaning of 'female fox' we can get to the sentence 'All vixens are female foxes' by substituting into the predicate adjective position of 'All vixens are vixens' the expression 'female foxes' for 'vixens'.
This sort of philosopher will hold that there are a priori truths that resist being arrived at by substituting terms with identical meanings in sentences which are a priori (logical truths). So take a sentence like 'A cube has twelve edges'. This is a priori because we know it is true on the basis of our understanding of the meanings of the terms. What we *cannot* do is arrive at it from some such logical truth as 'A cube is a cube' by substituting 'has twelve sides' for 'is a cube'. But now White makes a VERY important move.
From the fact that platonist persists, just as does the nonplatonist, in insisting that the issue be couched in terms of sameness of *meaning*, White goes on to say that we move FROM ontology to "ideology" ala Quine. What a move! Because of this move, in my opinion, there is no wonder that White can now dismiss advocates of synthetic apriori truths:It should be be noted therefore that the doctrine of the synthetic a priori cannot profit by the criticism l leveled against positivism *at this point* (my emphasis sb), for advocates of the synthetic a priori are just as dependent upon the phrases 'is identical with' as applied to meanings and 'is synonymous with' as applied to predicates as positivists are. (p.137)White is careful to remind us that when terms are not synonymous but merely have the same designation analyticity is not preserved under substitution. So we can't infer 'All men are featherless bipeds' is analytic if we can infer 'All men are featherless bipeds' from "All men are men" and the merely extensionally equivalent 'men' and 'featherless bipeds' by sustitution.
I think it is important to note that the advocates of synthetic a priori truths (or statements) *only* fall into the trap of being "just as dependent" on synonymy as positivist *if* they characterize synthetic a priori truths NEGATIVELY as sentences that cannot be derived by substituting a synonym in a logical truth! It is altogether possible to argue, I believe, that there are categorial truths that in no such way depend on logical truths. Furthermore, even if we couldn't derive 'Every vixen is a female fox' from 'Every vixen is a vixen' plus the synonymy of 'vixen' and 'female fox' all this would prove would be that on one construal of 'analytic' not all a priori truths are analytic. Nothing really excludes resisting Quinean ideology in favor of ontology.
There is, if White is correct, a problem with meanings. The problem is that they are not like other things. We can tell whether, to use White's example, Memorial Hall is taller than the Eiffel Tower by standard measuring methods, but how do we determine relations among meanings? It would appear that the platonist approach is in trouble. White identifies two fundamental problems (p. 140).1. Detecting the relation the platonist requires between meanings in order to ground synonymy is difficult. 2. Asserting the relation the platonist requires arguably presupposes the very synonymy that relation is intended to explain. (sound familiar? - sb)White assumes that the platonist does not infer the existence of meanings from mere facts of synonymy. White places the existence of meanings on the same level as the existence of molecules, and asks us to consider what phenomena they explain. Can we invoke meanings as explainers in the same way we do molecules to explain, say, the so called Brownian movement associated with tiny flecks of pollen suspended in water? A big problem with analogically introducing meanings is that they are neither mental nor physical so we have some difficulty figuring what sort of events or facts they might explain (p.143).
So White recommends an alternative to an ontological account of analyticity based on meanings which in some sense lie beneath the surface of language. White entertains two possibilities. The first is one where we might describe analytic sentences as sentences which when denied yield us a self contradiction. There is a problem, however: If we discover 'All men are rational animals' is analytic how is it that we are able to determine that it is self-contradictory to deny it? We have already ruled out the ontological approach in the form of an appeal to meanings or 'intensions'. But neither can we appeal to the synonymity of 'men' and 'rational animals' in deriving 'Some men are not men'. This approach, then, collapses. So what about the other option?
This is one where to say 'All X are Y' is analytic is to say that if we were given a non-Y we would not call that non-Y thing an X. The problem here is that there are sentences that appear to be clearly non-analytic which fit the criterion; so, given a nonfeatherless biped, we would not, then, call the thing a man; *but* one could hardly infer the analyticity of 'All men are featherless bipeds'.
White appear to anticipate Quine's discussion of 'radical translation' but only to the extent that he considers how we adjudicate competing translation manuals. The conclusion he draws is not quite as radical as Quine's, however. White only says (p. 146) that no amount of prompting the native speaker will allow us to distinguish terms which are coextensive and terms which are synonymous.