Inversion gaps

[nschneid]

In a couple of places in the data there are multiple consecutive gaps with inversion: simplifying to remove a relative clause,

here are some emails -- --

The idea is that the subject ("some emails") is left alone and both the verb and predicative complement are extracted. The two gaps form a constituent where the first is the Head and the second is the PredComp.

But another strategy is demonstrated in the guidelines:

it works -- said -- Jim

where the verb stays in place and the subject and object swap to opposite sides—the subject into prenuclear position and the object into postnuclear position. Is this inconsistent, or is there a reason to treat these cases differently?

(Personally, I like the idea of keeping the verb in place as it is the head of the VP/clause, and there is no need for a constituent consisting of two gaps.)

[nschneid]

Note that we still need cases with gap as Head for subject-aux inversion (we say the auxiliary, not the subject, moves: to prenuclear position from head position). But perhaps we can prohibit constituents combining multiple gaps with no lexical material.

[nschneid]

CGEL p. 1385 discusses subject-dependent inversion:

In the great majority of cases the preposed element is a complement, usually of the verb be; [vi] shows, however, that with other verbs it can be an adjunct. It is for this reason that we speak of subject–dependent inversion – inversion of the subject and another dependent of the verb.

The verb itself is not described as pre- or postposed, which I think means that "here are some emails" should receive a similar structure to the reporting verb case.

(On pp. 1389-1390 are discussed special constructions with here/there + be, only some of which are subject-dependent inversions.)

[nschneid]

If we accept that subject-dependent inversion creates gaps around the verb, rather than at the end of the sentence with a constituent combining two gaps, that leaves one remaining gap-only constituent in the data:

[not only]/y did/x they [[--x --y] answer the phone at 4:50 on a Thursday] they hit the ground moving

This has "did not only" forming a VP, I suppose by analogy to "did not". But this feels different. The "only" is mandatory:

• *Not did they answer the phone at 4:50....

and coordination of "did not only" with another auxiliary cluster is not as typical as "did or didn't":

• *I thought you did and did not only wash the dishes.

though I found this from Fowler's Dictionary of Modern English Usage (1926):

• ...the result will or does or did not only follow, but follow without delay

In the more typical case, I wonder if "not only" should be an adjunct to the complement (answer the phone...) or even to the whole VP with the gap for did (did answer the phone...).

I guess the normal noninverted counterpart (without do-support) would be

• They [VP [Mod:AdvP not only] answered the phone...], they hit the ground moving.

suggesting

[not only]/y did/x they [--x [Clause [VP --y answer the phone at 4:50 on a Thursday]]] they hit the ground moving

[nschneid]

@BrettRey thoughts?

[BrettRey]

Sorry, for taking so long to get to this!

• I'm not sure whether to put any weight on this or not, but I think this behaviour is a relic of V2, and in V2, the verb moves to the left and then the topicalized item moves left.
• I'm not sure if this applies to the It works, said Jim construction, but CGEL is explicit about a post-posed subject in those cases.
• In the SAI, though it's explicit about the Vaux's being a prenucleus.

That said, I agree that the section on 1386 strongly suggests the same analysis as the It works, said Jim construction.

I'll add a section for subject-dependent inversion in the documentation.