HCONS constraint lost - but not always - using the same rule?

I was inspecting a pseudolanguage and noticed that, while both the PP and the VP are licensed by the same rule, in the PP the HCONS link between the quantifier and the noun is lost. The VP looks fine.

11%20PM

What I mean is, I looked at the local AVMs for both NPs, they looked fine. Then at the PP level, the HCONS were broken in the PP.

But of course this can’t be the Head-Comp rule as it is the same in both situations. So I am missing something.

My guess is that the problem is actually with the P, but doesn’t cause a completely broken diff-list append until you get up one level higher.

1 Like

Right. I think the issue was that the P was lacking the no-hcons-lex-item parent. I am still not sure I understand the mechanics but I suppose being able to guess this is already something :).

That parent should say HCONS <! !>, and without it the HCONS is just underspecified, breaking the diff-list appends.