The Analytic Fallacy

In Stewart Shapiro's excellent book (with an excellent cover photo), Thinking About Mathematics, Shapiro writes: "Quine's thesis is...there is no sense in saying that a sentence is true in virtue of language alone...critics of Quine's view point out that some sentences are in fact true in virtue of meaning.  Can we really contemplate experiences that could get us to deny that cats are feline, bachelors are unmarried, or 6=6?"

Shapiro's strategy is to suggest that Quine should not be taken too literally.  Shapiro writes: "In any case, I think that Quine can concede that some sentences are true in virtue of meaning and so are analytic...Quine's point is that analycicity cannot play the central role that the logical positivists had for it."  In other words, Quine didn't really mean that *no* sentences are true in virtue of language alone; what he really meant was that *almost no* sentences are true in virtue of language alone, or that almost no interesting sentences are true in virtue of meaning alone.  Being a real scholar, Shapiro goes on to pull some quotes from Quine (written much later than the first quote) that support this point.

I am a Dilletante so I cannot give you quotes.  All I can give you is my deep thoughts.  Here they are:

Of course I can contemplate experiences that would get me to deny that cats are feline.  There is a particular animal that I call a dog.  If I go outside and hear that everyone else calls this animal a cat, I will deny that cats are feline, unless people start using the word "feline" differently.

Why?

There are two kinds of linguistic propositions.  The first is empirical: it states an attribute of what people generally mean when they use a term.  So the proposition "bachelors are unmarried" can mean, "when people use the term bachelor to describe someone, that someone is usually unmarried."

The second kind of linguistic proposition is truly definitional, when I define a term that I will be using in my own communication.  So I could say, "I will use the term bachelor to mean an unmarried man", or, "I will use the term gnosis to mean unmediated knowledge, or valid intuition."  This kind of a proposition is not really a proposition at all.  It doesn't really say anything; rather, it lays the groundwork for me to be able to say things, perhaps more clearly (or perhaps less clearly).

I still don't quite understand the proposition 6=6, but I do think that a lot of the logical positivist thought about math is really formalism in disguise.  So let me get back to you on 6=6.

So I don't know if Quine is an anti-linguistic purist or not.  But I am.  No sentences are really true in virtue of language alone.

In fact, I think Kant's distinction between analysis and synthesis was a big mistake and has caused us a lot of headaches.