It’s clear that different thinkers want different things from a semantic theory of a language, or a theory of meaning. I want to schematize the different possible desiderata for a theory of meaning. Here is the beginning of a schema, and a comment on one of its problems.
A good semantic theory (where a theory is construed as a class of sentences) of a language L is/expresses:
(a) What we must understand
(b) What we must know
(c) What we must believe
(d) What we must know-true
(e) What we must believe-true
(f) What we must act as if true
in order to
(1) Understand the sentences of L
(2) Know the meanings of the sentences of L.
A conception of the goals of semantic theory comes from picking "is" or expresses in the first clause, one item on the lettered list, and one item on the numbered list.
I include (f) for those who think that semantics is a branch of sociology, not psychology – who think that the meanings of sentences are not mental representations or attitudes towards mental representations, or are not best studied through mental representations of meaning or attitudes towards those representations, or are not determined, in any interesting sense, by mental representations or individuals’ attitudes towards them. But I feel like, to accommodate just these sorts of folks, there should be something corresponding to (f) in the numbered list. “Count as a(n expert) speaker of L” or “Count as a member of the linguistic community centered around L” is too strong, because some phonological or pragmatic know-how or behavior goes into these things, but is not properly semantic. I’m at a loss.
There are also more ways of clarifying what we want from a semantic theory by placing different sorts of restrictions on the values of L – idiolects, dialects, the verbal behavior of maximal sets of mutually intelligible speakers, or whatever.