An annoyingly technical question about right nonce-constituent coordination in Huddleston & Pullum (2002)

Question

I'm currently reading Huddleston & Pullum (2002), aka CGEL, specifically the chapter on "right nonce-constituent coordination"; the relevant pages are pp. 1342-1343. My question is about how they would analyze a particular sentence; my apologies for writing a post that is so technical and littered with their terminology.

My question is about this sentence (of my own creation, not from CGEL), which I think is grammatical:

1. You can rely on no friend too much or enemy too little.

"Friend too much or enemy too little" would appear to be a case of right nonce-constituent coordination, since in "you can rely on no enemy too little," "enemy too little" is not a constituent. The problem with this analysis is that, while "enemy" is (part of) the complement of "on," "too little" is an adjunct in clause structure. So there's no sensible place to put "friend too much or enemy too little" in a syntax tree, along the lines of the diagram at the bottom of p. 1342, since the coordination as a whole can be neither part of a prepositional phrase headed by "on" nor a constituent at the clause level.

Huddleston & Pullum address this issue in a footnote, note 65 on page 1343, using the similar example "We'll be in Paris for a week and Bonn for three days." In that case, they say that the answer is to treat this as a case of ellipsis, with "in" omitted before "Bonn"; if you undo the ellipsis, you get "We'll be in Paris for a week and in Bonn for three days." This can then be treated as a normal case of right nonce-constituent coordination, since both "in Bonn" and "for three days" occur at the clause level.

In (1) we seem to have the same situation; again we must posit that there is some ellipsis happening before "enemy." Here, however, we cannot just posit that the preposition "on" is ellipted before "enemy"; you would get the ungrammatical sentence *"You can rely on no friend too much or on enemy too little." This is, of course, because "enemy" is not a complete NP acting as the complement of "on" but a nominal within the scope of the determiner "no."

Instead, I think you have to posit that "on no" is being ellipted. So, undoing this ellipsis, we get:

2. You can rely on no friend too much or on no enemy too little.

The problem is that, strictly speaking, (2) is not quite equivalent to (1), since it changes the scope of the "no." In (1), "no" has scope over "or" (as in "you can rely on no friend or enemy") whereas in (2) "or" has scope over "no" (as in "you can rely either on no friend or on no enemy"). Instead, (1) is really equivalent to this:

3. You can rely on no friend too much and on no enemy too little.

This seems to lead to a problem. On the one hand, you can't treat (1) as an ordinary case of right nonce-constituent coordination, because (as stated earlier) the coordination "friend too much or enemy too little" can't be sensibly placed anywhere in a syntax tree. On the other hand, you can't describe this as a case of ellipsis, as their footnote suggests for the Paris/Bonn example, since you run into issues of quantifier scope. Indeed, this is exactly why they don't use ellipsis to account for other cases of right nonce-constituent coordination (see example [19iib] on p. 1342).

So, what is the correct way of analyzing (1), in H&P's terms? What, if anything, is ellipted in the coordination "friend too much or enemy too little"?

Answer 1

This is best described an example of a complex case of gapped coordination, which is outlined on p1338-1339.

You can rely on no friend too much or ___ enemy too little.

As noted on p1340, gapped coordination does not always have equivalence with the basic version of the coordination, hence we cannot reasonably posit an ellipted element or elements and must account for this as a simple gap instead - i.e. it's a special kind of coordination that plays by rules that ellipsis doesn't explain.

[15]

ii.

a. Kim wasn’t at work on Monday or Pat ___ on Tuesday.

b. Kim wasn’t at work on Monday or Pat wasn’t at work on Tuesday.

[a ≠ b]

...

Semantically we again have just one negative here, with scope over the whole coordination. We have noted that “not A-or-B” is equivalent to “not-A and not-B” (I didn’t like his father or mother = I didn’t like his father and I didn’t like his mother), and the same applies here, so that [iia] is equivalent not to [iib] but to Kim wasn’t at work on Monday and Pat wasn’t at work on Tuesday.

As far as the tree goes, an elegant solution would be to assume there are a sequence of elements making up the antecedent for the gap, then it'd fit nicely as a PP + AdvP nonce constituent that would have it looking much like the tree on p1342.

You can rely [on no friend | too much] [or ___ enemy | too little].

Of course if we insist that there must be enough in the gap to make up the full clause, then we'd have to have You can rely on no as the gapped element. But since this would produce an unacceptable result as far as the tree goes, we'd then have to separate the gaps into You can rely and on no, which would produce the same result as above, but with a more complex tree.

Answer 2

Instead, (1) is really equivalent to this:

You can rely on no friend too much and on no enemy too little.

Two main clauses:

You can rely on no friend too much and you can rely on no enemy too little.

Simple ellipsis made complex...