Kant can be easily misread (or: I myself easily misread him, for a long time) as claiming that no "existence claims" are analytically knowable. Technically, though, his system has it that (positive) categorical existence claims are not analytically knowable, but some assertions about existence are. This is so because his system allows that the following existence claims are true (and knowable by logic alone, so to say):
- A exists or A doesn't exist.
- No A exists such that A = ~A.
He can even allow for existential conditionals like, "If a vacuum collapse bubble exists, the universe as we know it will probably be 'destroyed' after thousands of eons have passed." (It would take that long on account of the bubble only expanding at the speed of light.)
Now, as far as Kant's presentation goes, the analytic/synthetic distinction is exhaustive. Either "the concept of the subject logically contains the concept of the predicate," or it does not contain that. The concept of existence is not really descriptive or qualitative; the phrase "a unicorn" and "an existent unicorn" call to mind virtually the same imagery, and whatever the latter "adds" to the image is relational, outwardly, e.g. a sense (or wish) that we might encounter a particular unicorn someday, somewhere. (For a long time, I thought George Lakoff's invocation of the actionable paradox of "don't think of an elephant" was an ironclad ostensive demonstration of an extremely crucial fact about language and imagination. However, I later realized that we could distinguish "don't think of a unicorn in general" from "don't think of a particular unicorn," and the latter "works" contra the way the former doesn't. So maybe Lakoff's insight is not so penetrating.)
But now consider:
- Either X exists or the concept/question of X's existence/X existing is meaningless.
I don't know that Kant ever considered such a proposition. Perhaps in his meandering doctrines of noumena and things-in-themselves, there is room for claims like (3). They would be "hyperanalytical" existence claims, so to say. I prefer framing the issue in erotetic terms and I think that Kant would have done so, or in some way (subconsciously?) did so, when he said that, for example, "What is the constitution of a transcendental object?" is in some sense a meaningless question. In fact, transcendental logic really is an erotetic logic; Kant inherited talk of categories from Aristotle, and so he also inherited Aristotle's (purported) appeal to "forms of possible questions" in this connection, I suspect. And a transcendental argument shows that the meaningfulness of certain questions presupposes the existence of facts that ground this meaningfulness, either due to our erotetic capacity in itself, or more "locally" (per the local content of the transcendental argument in play).
Finally, consider a "degenerate"(?) option:
- Either X doesn't exist or the question of X existing is meaningless.
Is talk of hyperanalytical knowledge of certain "weird" existence claims consistent with Kant's scheme for the analytic/synthetic distinction, or if not his scheme, at least the thematism of the entire first Critique?