How could experience ever justify us in revising a putatively analytic statement like 'all bachelors are unmarried men'? I imagine Quine is entertaining the possibility that we may stumble across some married or female bachelors. But how could this ever happen? No one can ever be a counter-example to our statement because to do this they would need to be married or female and would then fail to be a bachelor, that is, a married man. Despite the attention it has received, I find it hard to see the plausibility of Quine's position.

Start with a different case. Take the sentence "Whales are a kind of fish".

Once upon a long time ago, that would have been taken to be a truism. And someone who then denied "Whales are a kind of fish" would probably have been suspected of not understanding "whale" (or "fish") -- whales are "by definition" a particular big kind of fish, it would probably have been said, just as bachelors are a particular kind of unmarried man. Philosophers of the time might even have said it was "true by definition", the kind of thing that is true just in virtue of the meanings of the words involved ("analytic" as some later philosophers would put it).

Yet nowadays we do routinely deny "Whales are a kind of fish". So if it really was once true just in virtue of the meanings of the words, and now it isn't true, we'd have to conclude that the meaning of the words has changed.

But is that right? Have we definitely come to change what we mean by the words "whale" or "fish"? Or is it perhaps that we have changed our beliefs about the way the world is (the nature of species, etc.)? Or is it a bit of both?

Quine's fundamental claim is that such questions don't have determinate answers (for a start, notions of meaning just aren't sharp enough). Our "web of belief" has changed, the interlocking network of sentences we accept as true has shifted. But, he thinks, we can't factor the changes neatly into a change-of-meaning component and a change-of-belief-about-the-world component. It's all a lot messier than that.

And if it isn't a determinate matter, in a particular case, whether change of assent involves change of meaning or change of doctrine, it isn't a determinate matter either whether some sentences are initially true-in-virtue-of-the-meanings-of-the-worlds involved (analytic), and then later not. An all-or-nothing notion of analyticity, Quine argues, does no useful work here. Rather we need some more shaded notions that come in degrees, notions like degree-of-entrenchment into our web of belief.

Now, you might respond, even if things are more complicated in the "whale" case, surely at least in the case of "all bachelors are unmarried men", this could only be rejected if we changed the meaning of "bachelor" or "unmarried" (or "man"). But maybe even that is not entirely obvious. Suppose that it became the case that the only legal marriages tended to be weddings of convenience e.g. for immigration purposes or tax purposes, and that such marriages had no implications for relationships. Rather, a different kind of civil partnership, with different duties and entitlements became the social norm (with the law remaining out of step with rapidly chnaging social practice). Then the ideas of (i) a man who is not in a long term relationship and could become a recognized member of a stable partnership and (ii) a man who is able to contract a legal marriage could peel apart. Suppose the word "bachelor" went with the first idea. Would that, quite determinately, be a change of meaning? I'm genuinely not sure (given all the life-style and partner-availability connotations of "bachelor"): again, the notion of meaning is arguably too unclear. Yet we'd no longer say "all bachelors are unmarried men".

Ok that really was a bit fanciful, and I don't want to insist on the example. But it perhaps it is just enough to give some colour to the Quinean point that these issues aren't clear-cut in the way that the traditional doctrine of analyticity would have it. (So it isn't that "Quine is entertaining the possibility that we may stumble across some married or female bachelors", but rather he is saying that issues about meaning and change of meaning, for example, aren't nice and sharp and determinate.)

But suppose you are inclined, despite all that, to dig in your heels and insist that "all bachelors are unmarried men" is true in virtue of the meanings of the words involved, and hence change of assent to that sentence must involve a change of meaning. Quine himself might disagree, but the more relaxed latter-day Quinean would probably not be too unhappy, and allow you a few trivial cases of analyticity like this. The argument would remain that the all-or-nothing notion of analyticity applies, if at all, only to such trivia, and is in bad shape for dealing with any more substantial cases of supposed necessary truths.

There is a really nice discussion of these things, if you can get hold of it, at the beginning of Paul Churchland's book Scientific Realism and the Plasticity of Mind. And another nice paper is Hilary Putnam's "It ain't necessarily so" (which you can find in one of the volumes of his Collected Papers).

Read another response by Peter Smith
Read another response about Philosophers