A Verification Principle
Consider a kind of sentence X
X: Any assertion that a thing has a certain property or is of a certain kind (or lacks that property or is not of that kind.)
In other words, X is the application of an adjective to describe a thing or the use of a noun to categorize a thing (or to deny that the adjective correctly describes it or to deny that the noun correctly categorizes it.) "The action was right" and "Joe is a zombie" are examples.
I propose the following verification principle, which has the form "if p, then q":
V: If S knows what S means by an X, then S has an idea of what S would take as evidence that the X correctly described a given object and what S would take as evidence that the X incorrectly described another given object.
In other words, if you assert something meaningfully by an X, then you have an idea of what would count as evidence that what you've said was false, and what would count as evidence that what you've said is true. I also accept that if S has an idea..., then S knows what S means by X. So, the verification principle could be made bidirectional (p iff q) if you like. However, I am only interested in the first direction: "if p, then q." You'll see why as we go along. What's important is whether terms and assertions have meaning, not (in the other direction) whether we can determine that someone has no evidence for an utterance on the grounds that he doesn't know what he means by it.
Also, I have nothing to say about whether the evidence is analytic or empirical in nature. It doesn't matter. It's a 21st Century verificationism, if you like, without the logical positivism that bedeviled its grandparent almost a century ago. Let's just not talk about logical positivists and their verification principle. Objections to it are relevant only if they are objections to V.
Here is an argument for V. Suppose S asserts that X. Suppose also that we ask S what would count as evidence that X was true and what would count as evidence that X and what would count as evidence that not-X. If S falls silent, shrugs, says "I have no idea," then we begin to wonder whether S has no idea what he's said when he said X. In fact, if a thorough probe showed that S really had no idea of what would count as evidence, then we would be justified in concluding that S had no idea what he meant by X.
Is V analytic? Yes. It's analytic philosophy, in which we analyze concepts to produce interesting tautologies. Oddly, the analytic-synthetic distinction is fuzzy, however, and V is ever slightly empirical in its derivation. In analytical philosophy, we notice which terms (such as "evidence") we would apply in imagined situations, in order to adduce evidence for statements about what we mean by other terms, such as "meaning." This imagining flushes out meaning and therefore helps construct analyses.
I intend to use V to argue for interesting results in philosophy of mind, epistemology, and metaethics. So, it is important to know whether V is true. Are there any problems with V?