Frank Kogan

but he does for example discuss analytic and synthetic judgments in terms of subjects and predicates (which kept together would always form sentences)

Yes, of course, I should have thought of this: the analytic-synthetic thing is put in terms of subjects and predicates; a statement is analytic if the predicate adds nothing that isn't already thought in the statement's subject; the statement is synthetic if the predicate does add something that can't be derived from the subject.

Unless I'm missing something (well, I certainly am, not having read the Prolegomena in 34 years, and never having read the Critique), an immediate problem with the idea of "analysis" jumps out at me; a question would be whether this problem also jumped out at Kant and he took care of it, or whether it jumps out at me because lots of people have spent time since Kant attacking Kant's notion of analysis. Whereas people in Kant's time wouldn't have thought to attack it in this way.

The problem is that what's "in" the subject of a sentence - what the subject is - is there by way of the meaning of the words that present the subject to us. But "meaning" is a matter of experience and convention (and our learning the conventions is a matter of experience). And anything based on convention and on consensus experience can be challenged or disagreed with, so what's already contained within a subject can be challenged and disagreed with. Once we've decided that all bodies must have extension and decided what we mean by the word "contradiction," it's contradictory to say that a body doesn't have extension. But what if someone doesn't choose to play by these rules? "All analytical judgments are a priori even when the concepts are empirical, as for example, Gold is a yellow metal; for to know this I require no experience beyond my concept of gold as a yellow metal." (Prolegomena §2.) But what if someone comes at us with a different concept of gold, say based on atomic structure, or something not yet thought of? It might turn out that nonetheless all gold is yellow, but this is no longer part of the concept. Of course, "anything can be challenged and disagreed with" doesn't mean that everything can be challenged and disagreed with at once, since the challenges and disagreements depend on assumptions too; just that what will or won't get challenged and disagreed with isn't determined in advance; even if we believe that the vast majority of our concepts won't get challenged, we won't be able to predict with certainty which will and which won't. Kant's question in the Critique is "How are synthetic a priori judgments possible?" But I don't see how analytic a priori judgments are possible, since an "analytic" statement here seems to simply consist of "at the moment no one I know of has come up with a good reason to rework or find an alternative to the concept."

E.g., for an Aristotelian, a man's returning from sickness to health was an example of motion, since change in quality or state was contained in Aristotle's concept of motion. But it wasn't contained in Newton's concept, hence would have been "analytic" for an Aristotelian but not even synthetic for a Newtonian, just wrong.

But I'm getting ahead of myself. We don't get to Quine's attack on the analytic-synthetic divide until chapter four. But what I wrote in the previous paragraph seems to border on being a platitude, that different people can use words differently. Was it really only in the middle of the 20th century that this became a big deal for philosophers? Or were Dewey and James and Nietzsche already there? Marx? Montaigne? Or does Kant himself deal with this at some point? I've only now read about three pages.

Index to the Carus edit of the Prolegomena is lousy. It says that "predicate" only appears on pp 43 and 99, and here I am on pp 14 and 15 and it's all over the place.