SLASH list not empty but the structure unifies with the root?

(probably related to this thread: In-situ wh-words in languages which also do extraction)

Why is the leftmost tree not ruled out by the nonlocal constraints on the root?

(Note: this is a pseudolanguage, not English):

Screen Shot 2020-01-28 at 4.49.12 PM

In the leftmost tree, cat sleeps is licensed by the extracted adjunct rule. So that middle S will have a non-empty SLASH list. Then the top S is licensed by the head-adjunct rule. But the head-adjunct rule is appending its daughters’ slashes. So the topmost S must have a non-empty SLASH-list. The root of course expects non-local-none on its SYNSEM. Why do I get that tree then?

Perhaps the SLASH value on the top S is fishy, I am not sure:

Perhaps that can somehow be unified with an empty SLASH list?

That was because the adposition was not inheriting from zero-arg, so the append was broken.

zero-arg isn’t right for the adposition though – it has arguments. It should be basic-one-arg, I think (assuming we’re not letting it take a subject).

1 Like

Hmmm indeed. Then I don’t know what’s going on. With basic-one-arg, I get the exact situation that I describe in the original post.

Is the sole ARG-ST element of the adposition in fact linked to the thing on its COMPS list?

1 Like

It wasn’t!! Thank you!