Numerical measurement phrase: incompatible with attach-low heuristic?

[nschneid]

Prompted by nert-nlp/legal-cgel#51: For "45 days after the request", there are a couple options:

• [Det:DP 45] [days [after the request]]
• [Nom [Nom [Mod:DP 45] days] [after the request]]

Unless we want to say that "after the request" is an external modifier, which seems wrong.

Measurement phrases can contain coordination but indicate a single measurement, e.g. "2 years and 45 days". (Such measurements can take singular agreement: 2 years and 45 days is too much.) Maybe:

• [Nom [Head:Coordination [Nom [Mod:DP 2] years] and [Nom [Mod:DP 45] days]] [Mod:PP after the request]]

Also, in such a compound measurement, "the" would not normally distribute: "the 2 years and 45 days after...", not "the 2 years and the 45 days after...".

I don't think this is a gap situation ([2 years __ and 45 days __] after the request). Distributing "after the request" gives the wrong semantics.

Considering regular count quantities with 2 DPs:

• The 3 cars and the 2 trucks in the parking lot should not be there.

In this case I guess it's a delayed PP:

• [NP-Coordination The [3 cars __] and the [2 trucks __]] [Mod:PP in the parking lot]

[bwaldon]

Also manifests in nert-nlp/legal-cgel#53

[BrettRey]

I think the issue isn't about measurement phrases with coordination specifically but about joint coordination in general (CGEL p. 1281–). It seems we need a new heuristic which is "attach high in cases of joint coordination."

[nschneid]

In this case I guess it's a delayed PP:

• [NP-Coordination The [3 cars __] and the [2 trucks __]] [Mod:PP in the parking lot]

In retrospect I'm not sure why I wrote that. Presumably the PP attaches outside the coordination.

Anyway, the original question was about "45 days after the request" and whether it should be "45 days" + "after the request" in spite of the attach-low heuristic. We were thinking through coordination as a constituency test.