Verbless clauses

[BrettRey]

The kids in tow may be a clause semantically, but a syntactic clause in CGEL is a projection of the VP. In a footnote on p. 1286, CGEL says, that "the ultimate head of hat in hand is in..., with hand an internal complement (in hand constituting the predicate) and hat an external complement (more specifically, the subject).” This is then a kind of 3rd layer on the PP analogous to the NP over the Nom or the Clause over the VP. So, instead of calling this a nonce, we could call it a PP_Clause: [PP_Clause [Subj hat][Head [Head in] [Obj hand]]]. Without any clothes one would be [PP [Head without][Comp:PP_Clause [Subj any clothes][Head on]]]].

We could then extend this to other verbless clauses:

1. with the children so sick: [PP [Head with][AdjP_Clause [Subj the children][Head so sick]]]
2. Although no longer a minister: [PP [Head although][Comp:NP_Clause [Mod _no longer][Head a minister]]]
3. once away from home: [PP [Head once][Comp:PP_Clause [Head away from home]]]
4. when drunk: [PP [Head when][Comp:AdjP_Clause [Head drunk]]]
6. if necessary: [PP [Head if][Comp:AdjP_Clause [Head necessary]]]
7. his face pale with anger: [AdjP_Clause [Subj his face][Head pale with anger]]
8. some of them primary school children : [NP_Clause [Subj some of them][Head primary school children]]
9. etc.

[nschneid]

What about using fusion and calling this Head-PredComp? The last phrase is of a kind that can be a PredComp (NP, AdjP, PP, but not AdvP), there's just no verb to serve as the usual VP head.

By XP_Clause you mean a clause headed by an XP, right? I don't see why that has to be given in the category if it can be inferred from the structure:

[Clause [Subj:NP the children] [Head:AdjP so sick]]

or [Clause [Subj:NP the children] [Head-PredComp:AdjP so sick]]

"with" + Clause vs. AdjP_Clause: It's not a problem for a PP to take a Clause as a complement. It is less common for a PP to have an AdjP complement. Perhaps "at large" and "at first" would be analyzed that way, but I don't know if it occurs apart from multiword expressions.

The "Head-PredComp" notation makes it easy to recognize why the clause's head is not a VP. Unlike other cases of fusion, there would not need to be an intermediate node on one branch—this is just saying that AdjP fills both the Head and PredComp functions of the Clause. (Or we could have a VP layer, in which case it would be both the Head and PredComp of the VP.)

[BrettRey]

I tried out the fusion idea on Geoff, and he was sort of luke warm towards it, but when I thought about it more, I realized that a fused-head NP can do anything any other NP can do, but verbless clauses don't have the same distribution as any other kind of clause, so I don't think this works, as attractive as it seems.

[BrettRey]

[Clause [Subj:NP the children] [Head:AdjP so sick]] doesn't work any more than [NP [Head:AdjP so sick]] or [PP [Head:AdvP quickly]]. A PP_clause would be a PP, not a clause.

[nschneid]

Well we have a notion of clause type, which partly accounts for the distribution of different types of clauses. It's just not in the tree (yet).

I think it will help to separate the ones with subjects from the others (it's not obvious to me that they're the same kind of construction). Considering those:

"Some of them primary school children" doesn't have the distribution of a normal NP—it can't be a subject or object, for instance.

"His face pale with anger" resembles a clausal predicative complement:

• What I saw was him leaving
• What I saw was his face pale with anger

With-absolutes are tricky because the meaning of "with" can overlap with its canonical-preposition meaning. (Grammaticalization!) But I would argue for the following groups:

with+Clause

• With the election approaching, we get a lot of campaign mail.
• With the election nearby, we get a lot of campaign mail.
• With the election over, we no longer get much mail.
• With the election around the corner, we get a lot of campaign mail.*

with+NP

• With the approaching election, we get a lot of campaign mail.
• With the nearby election, we get a lot of campaign mail.
• *With the over election, ...
• With the proximity of the election, we get a lot of campaign mail.

* Actually this one is ambiguous. It could mean the election that is around the corner (as opposed to others), which is the NP reading. But the clausal reading ('given that the election is around the corner') strikes me as more likely.

[BrettRey]

The notion of different clause types won't save the fusion-of-functions analysis because what that does is change the inside, while saying to the outside syntax "I'm a normal XP." Fusion of functions is a clever analysis because the result is NOT a different type of NP.

Because fused-head Clauses have radically different distributions than any other kind of clause, they just don't make sense.

In contrast, NPs with N_pron has a slightly different distribution from others and same with VPs with V_aux. A Clause_rel has a different distribution than a content clause. So, I think that extends easily to say that an NP_clause would be a special kind of NP with a special kind of distribution. If "clause" is causing a problem, then maybe NP_{prop(osition)}.

As you point out, though, the subjectless cases have a different distribution, so the NP_clause/prop analysis can't capture both those with subjects and those without.

[nschneid]

How is the distribution of "the election over" related to the distribution of a PP?

[BrettRey]

It is a small subset.
The election over, traders' attention returned to an increasingly sickly European economy.
With the election over, traders' attention returned to an increasingly sickly European economy.
After the election, traders' attention returned to an increasingly sickly European economy.

[nschneid]

Decision: Leave as future work, make a note in the guidelines

[nschneid]

Discussion from our paper draft: