A topic that recurs in the interviews with Quine is his views on analyticity and how his views have changed since “Two Dogmas”.

One of Quine’s responses surprised me. In the interview with Føllesdal, Quine says: 

So the acquisition of elementary, basic logic is just part of the acquisition of language. In this respect I do regard elementary logic as analytic in the sense of being something that is true, as they say, by virtue of the meanings of the words.

He goes on to clarify that what he means is that that analytic truths are truths is learned in the learning of the use of the words involved and nothing more. This is, he says, the sense in which “No bachelor is married” is analytic. Somewhere in the book he indicates that he goes into this in Pursuit of Truth. (As an aside, how does this book not have an index entry for “analytic”?) This description of what is going on with analytic truths seems to be backing off of the hard line position in “Two Dogmas”. I seem to remember him saying, in “Two Dogmas,” that not even logical truths are analytic. [Edit: In the comments, Daniel corrects this slip of mine.] Alternately, one might interpret this as toning down the rhetoric, since among Quine’s early criticisms of analyticity was the charge that there was no general characterization of analytic truth. He goes on, in the interview, to say that the problem with Carnap’s view was that he had no general criterion demarcating analytic from synthetic truths. 

Quine puts the point a little more sharply. 

As for a general criterion of analyticity, a general definition that I’ve used is that you might say a sentence is analytic for a native speaker if he has learned the truth of that sentence, learned to affirm that sentence, in the course of learning one of the words in it. … And the reason this criterion doesn’t serve satisfactorily for the general epistemological sort of application that I’ve criticized is that in general we don’t know how we learned our words and we don’t care.

A quick check back over “Two Dogmas” indicates that Quine, there, harped on the lack of a general boundary between analytic and synthetic statements. However, in the opening of the article, he does cite the belief in a fundamental cleavage of truths into two classes, analytic and synthetic, as one of the dogmas of empiricism. I cannot find a supporting passage, but I had thought that in “Two Dogmas” Quine was arguing against the coherence of the notion of analyticity. In the interviews he sounds like he endorses something more like Putnam’s position, namely that there are analytic truths but the notion of analyticity cannot do any substantive work in epistemology. I had thought that Quine’s view of analyticity had amounted to the claim that we cannot make sense of the notion of analyticity, but his revised view seems to be that we can make sense of the notion in particular, isolated cases but it cannot do anything of significance in general.

One question that pops up after reading this is: why does Quine specify a notion of analyticity in terms of learning the use of words when it doesn’t do any work for his epistemology? Isn’t the introduction of idle symbols something that he complained about elsewhere, e.g. against Principia Mathematica?  Granted, I haven’t looked at Pursuit of Truth and he might give some motivation for it there.

Before I continue, I want to make one remark about Quine’s views about learning logic. He says that we learn elementary logic in the course of learning language. If so, one would expect everyone in college to be reasonably good at elementary logical manipulations, since undergrads have learned their languages to some degree. This isn’t the case though. Quine was a logic teacher, so he was aware of the difficulty some people have in doing basic logic even though the same people have no trouble saying grammatical sentences, demonstrating they’ve internalized the grammar of their language. Perhaps Quine would chalk this up to pedagogical deficiencies. This seems to be something that his view about the learning of logic should be able to explain.

Returning to the interviews, Quine does seem to have backed off of his criticisms of analyticity some. He says in another interview that he has come to see the greater evil as the second dogma, and so his arguments for holism are the more important parts of that article. Oddly, this doesn’t seem to surface in his “Two Dogmas in Retrospect,” at least not from a quick skim of that article. The force of the holism arguments against Carnap is something I’ve never been quite clear on. In Logical Syntax, Carnap talks about Duhem, endorsing his points about holism. Quine must be accusing Carnap of bad faith, endorsing holism while repudiating it in the commitments of his further writings. Quine goes on to say that appreciating holism undermines the role of analyticity in epistemology. From reading these interviews, I get the feeling that I need to reread “Two Dogmas” focusing on the part on the second dogma, which in almost all my previous readings has definitely not been at the forefront.