determiners providing discourse referents for future pronouns

While talking with Bill Ladusaw (ooh... look at me name drop. Albeit in a rather restricted context) about my linguistics thesis yesterday, he brought up the issue of whether nominal phrases with "a few" and "few" (the objects of my investigation) can serve as referents for future pronouns in the discourse. I told him I hadn't thought about discourse yet. His suggestion was that "a few" does, but "few" doesn't. A good observation, I thought, and definitely one I could incorporate into my thesis. But it sounded vaguely familiar somehow.

So I went back and looked in my journal from the summer. Sure enough, I had jotted down some examples of this very phenomenon:

1) A few Republicans oppose the tax cuts. They think they're welfare for the rich.
2) *Few Republicans oppose the tax cuts. They think they're welfare for the rich.

The asterisk, for those unfamiliar with the conventions of scholarly linguistics, indicates that the mini-discourse in 2) is somehow ungrammatical or, at the very least, bizarre.

Rather exciting to realize that I came up with the same non-trivial observation as one of the world's top semanticists. We'll just ignore the fact that I managed to forget it for 4 months...

UPDATE: To clarify a bit, 2) above is ungrammatical if "they" is meant to be co-referenced with "Few Republicans." A perfectly grammatical reading (and, indeed, the natural one) is "they" being co-referenced with those Republicans who don't oppose tax cuts. But since Republicans who support tax cuts are rather unlikely to think of them as welfare for the reach, this reading's rather bizarre, too.