I was a little puzzled by the original semantics question, to tell the
truth. Was it in response to my suggestion that we not use semantically
bizarre sentences for grammatical analysis? If such sentences are
naturally-occurring, they have to be accounted for. I was arguing
against making up weird sentences and then using them to understand
non-weird ones.

I am an extremely strong believer in using semantics as part of any
accounting of language, including an accounting of syntax. In fact, my
posting implied this when I said that we often have to take into account
the semantics of a construction in order to understand its grammatical
behavior. In fact, I was trying (not too clearly?) to say that semantics
is such a strong controlling factor in the grammar of sentences, that we
can't afford to set it aside by believing we can learn something about
grammar from semantically unacceptable sentences.

Maybe that makes what I was trying to say more clear. I hope so.

Johanna Rubba