Ellipsis woes

In collecting handbook comments on government of verb form by the nearest, I came across advice about a very different sort of “failed parallelism” bundled together with advice about the verb form cases. What the two phenomena share is data involving ellipsis in coordinate sentences and an appeal to parallelism.

So in the same passage Pink’s Dictionary of Correct English (1948) cautions against a change of “tense” in

(1a) always has [done] and will continue to do

and a change of number in

(2a) so marked is his preference and so cosmopolitan [are] his tastes

(where the ellipted material is in square brackets). And Bernstein’s Watch Your Language (1958) generalizes about ellipsis, saying that the understood word must be in the same form — number, tense, person — as the explicit one, and giving as parallel violations

(1b) all men who have [served] or will serve

(2b) Ammunition was seized and 150 suspects [were] freed.

(similarly in his Dos, Don’ts, and Maybes of English Usage (1977)).

Here we see a common tactic in the advice literature: putting different proscriptions together as instances of a more abstract piece of advice. As I said in a fairly extensive Language Log posting of 2005 on (claimed) failures of parallelism, such umbrella advice mixes together phenomena of very different statuses and typically overgeneralizes in ways the advice writers don’t appreciate.

In the cases at hand, the (1) examples illustrate factorable coordination, and the (2) examples do not. Even if you view factorable coordination as involving a kind of ellipsis — not all syntacticians do — factorable coordination is a very different sort of construction from the ellipsis construction in the (2) examples, which syntacticians call Gapping.

Simplifying things considerably, in Gapping an omitted main verb in the second of two conjoined clauses picks up its meaning from the main verb of the first conjunct (in traditional terminology, it is anaphoric to this earlier verb, which is its antecedent). So in

(3) Kim prefers sushi, and Sandy ___ sashimi.

the omitted material is understood as supplied by prefers. (The position of the anaphor, the omitted verb, is indicated by the underine, and its antecedent by boldfacing. [The previous sentence has a somewhat more complex sort of Gapping in it; the omitted material has the antecedent is indicated, which is a two-word expression. Clearly there’s a great deal more to be said about Gapping.])

What I just said about Gapping treats the anaphor-antecedent relationship as resolutely semantic. Given some further assumptions about semantics and morphosyntax, such an account predicts that examples like

(4) Kim prefers sushi, and her friends ___ sashimi.

should be impeccable. For me, they are, as are the (2) examples above. (Remember that I’m someone who doesn’t use things like the (1) examples, so this is a big difference between factorable coordination and Gapping.)

The further assumptions have to do with the morphosyntactic features (or “inflectional categories”) of expressions. Crudely (again), these come in two varieties: semantically potent ones, like number on (must) nouns and tense on (most) verbs; and morphosyntactically determined ones, like person and number of (most) verbal expressions. (Yes, I know, there are manifold complications, many of them well studied. In particular, determined forms can often be used to convey features of an absent determinant: “Am going to London soon” in a letter, for instance. Here I’m trying to sketch the larger outlines of these phenomena, not to paint a complete picture.)

What this means is that in (4) the number difference between anaphor and antecedent is, for a great many speakers of English, ignorable, though (4) could not be understood as conveying ‘Kim prefers sushi, and her friends preferred sashimi’.

So much for one variety of English. The question is then: what was going on for Pink and Bernstein?

Two possibilities: one, these writers — no doubt there are others — were simply reporting their judgments about examples like (2) and (4), and have packed a proscription against them in with a proscription against examples like (1); or two, they conceived an abstract principle about ellipsis that led them to become wary of things like (2).

I’m not trying to gauge what was in people’s minds, though I am floating the idea that they might have been making an entirely “theoretical” objection based on some sort of reasoning from “grammatical logic”. Who knows?

What is clear from what these writers (and some others) said about Gapping is that they viewed the anaphor-antecedent relationship as a formal (rather than primarily semantic) one, as a relationship between two expressions: an explicit one (the antecedent) and an understood one (corresponding to the elliptical anaphor); the two are required to be in the same form. So Pink said that (2a) involves a change of number (antecedent is, undersood anaphor are), and Bernstein talked about a similar change in (2b) (antecedent was, understood anaphor were).

Both Pink and Bernstein treat the “same form” requirement as a condition on ellipsis in general, which means that they predict that it applies to ellipsis constructions other than Gapping. In particular, they predict that it applies to the much-studied Verb Phrase Ellipsis (VPE) construction. And that prediction is just wrong.

Now there are edgy examples of VPE — Language Log’s discussion of these goes back at least to 2006 — but there are plenty of unexceptionable examples, easily found, of VPE where the antecedent and the anaphor are of different forms. Here are a few from the many examples in my collections:

“We cannot allow energy to divide Europe as Communism once did,” José Manuel Barroso, the European Commission president, told The Financial Times.  But it is ___. (Thomas L. Friedman, “The Really Cold War”, op-ed piece in NYT, 10/25/06, p. A19) [base-form antecedent, present-participial anaphor]

I lost weight with Jenny Craig, and you can ___ too. (television ad for Jenny Craig weight-loss program, seen July 2008) [finite antecedent, base-form anaphor]

But if you ask adultescents why they haven’t grown  up, they may give you a simple answer: Because they don’t have to ___.  (John Tierney, “Adultescent”, NYT Week in Review 12/26/04, p. 4) [past-participial antecedent, base-form anaphor]

In the ellipsis section of his Modern English Usage (pp. 134-6), H.W. Fowler takes up one example of VPE, and goes one step further than Pink and Bernstein in insisting on identity of antecedent and anaphor; for

Would not have really revised the Bill, as no doubt it could be ___.

he says that revised must be repeated, because although both antecedent and anaphor have the same form (past participle), the antecedent is active in voice (the past participle is in a perfect construction) but the anaphor is passive, so there’s a change of voice.

There are many more ellipsis constructions in English(not all involving verb expressions), and I have no idea what the range of opinions is about feature parallelism in them.

[I’ve skirted some subtle questions here about when word forms are the same and when they are different. To get something of the flavor of the issue, compare:

I prefer sushi, and Kim ___ sashimi.

I preferred sushi, and Kim ___ sashimi.

Some would bar the Gapping in the first example (insisting on using prefers instead) but allow it in the second. But in both examples, the first clause is 1st person and the second 3rd person.]

2 Responses to “Ellipsis woes”

  1. Nominal ellipsis « Arnold Zwicky's Blog Says:

    […] By arnoldzwicky In a recent posting I looked at some English constructions (Gapping, Verb Phrase Ellipsis) with verbal ellipsis in […]

  2. VPE mismatches « Arnold Zwicky's Blog Says:

    […] intended and caught my eye that way). One example of each type (there are a few more examples here): 1. BSE ant, PRP ell (the “McWhorter configuration”, so labeled for its deployment by […]

Leave a Reply


%d bloggers like this: