a. I don’t think she has seen it, but she may (have).
b. This one needs to be repaired; the other already has (been).

Can you omit the second auxiliary? If so, is it more natural to omit it?

These examples are taken from The Cambridge Grammar (page 100):

Differences between stranding and the central do-support constructions
(a) Stranding not restricted to primary forms
One difference between stranding and the do-support constructions covered in §§2.1.1–3 is that while the auxiliaries in the latter are always primary forms, this is not so with stranding:

[29] i I don’t think she has seen it, but she may have ___. [plain form]
ii This one needs to be repaired; the other already has been ___. [past participle]
iii %He said I was being unfair, but I don’t think I was being ___. [gerund-participle]

This construction is subject to some regional variation: AmE doesn’t allow [iii], but it does have [i–ii], even if they are less usual than versions in which the second auxiliary is omitted.

The book is saying that it's more usual to omit the second auxiliary in [i–ii]. But I for one wouldn't omit the second auxiliary in [i] or [ii], so I have asked the question. Apparently, @EdwinAshworth seems to agree with me, but others seem to disagree. So I think it's worth investigating in ELU.

  • I'd not omit either; I'd class the second example as wrong on the grounds of unidiomaticity (and probably ungrammaticality). (a) is if anything a shortening of 'I don’t think she has seen it, but she may [yet] do'. Again, I'd reject as unidiomatic. Aug 17, 2023 at 9:49
  • In less flowery language, "on the grounds of unidiomaticity" = "because it's ungrammatical"
    – BillJ
    Aug 18, 2023 at 6:44
  • Do you speak American English? I think that their claims in that last paragraph are only intended to apply to AmE. CGEL's coverage of AmE is often...questionable.
    – alphabet
    Aug 18, 2023 at 16:16
  • @alphabet I think CGEL's saying that [29i-ii] is not limited to AmE but is allowed in BrE as well. So I'm not sure if their claims are only intended for AmE.
    – listeneva
    Aug 19, 2023 at 1:03
  • @listeneva They seem to be saying that in both dialects [i] and [ii] are valid with or without the second auxiliary, but that in AmE the second auxiliary is usually omitted. I disagree with them about AmE (see my answer below). Dunno about BrE.
    – alphabet
    Aug 19, 2023 at 1:21

4 Answers 4


The answer is different for cases a and b.

In case b, leaving out “been” is acceptable and doesn’t change the meaning. It may seem slightly more fluent to leave it out because it’s a “redundant” word, but it depends on context.

In case a, leaving out “have” potentially changes the meaning. With “have” the sentence is anchored to the present/past and reads as an epistemic statement: “To the best of my knowledge she has not seen it, but I could be wrong”.

Without “have” it reads more like a temporally uncertain statement: “I don’t think she has seen it at this time, but in the future she may see it”.

But the reason for the difference is in which aspects the auxiliary verbs contribute. “May” (or “might”) doesn’t have any past tense aspect(or a past tense sense you could substitute in), so you need to use “have” to contribute that.

  • 3
    I find the omission of 'been' unacceptable. Aug 17, 2023 at 9:44
  • 1
    I find the omission of "been" to be fine.
    – Stuart F
    Aug 17, 2023 at 10:58
  • I think the reason that (b) is more acceptable than (a) is because of the form of the ellipted auxiliary is the same as the form from the previous sentence. This isn't the case in (a) and it's a problem. Aug 17, 2023 at 13:41
  • 2
    @EdwinAshworth I find it fine. North/South divide maybe. Aug 17, 2023 at 14:38
  • Omitting words /word forms that have not appeared previously doesn't sound standard to me. Lack of parallelism, Stuart, Araucaria. Aug 17, 2023 at 15:44

(a) is an ordinary example of Conjunction Reduction, and therefore grammatical:

a. I don’t think she has seen it, but she may have seen it
== Conjunction Reduction ==>
a' I don’t think she has seen it, but she may

Conjunction Reduction is a syntactic rule, and is therefore mindless, automatic, and insensitive to meaning, either lexical or contextual.

(b), on the other hand, is ungrammatical:

b. *This one needs to be repaired; the other already has

(b) looks like conjunction reduction (you can put in but or use a semicolon, as you like),
but it doesn't follow the rules required for the rule. In conjunction reduction, the second clause must have the same structure as the first, except for the word that gets deleted because of identity. That doesn't work with this example.

If you try to undo Conjunction reduction with (b), you get

  • This one needs to be repaired; the other already has to be repaired.

And this is grammatical, but it doesn't mean the the other has already been repaired. Whoever put the sentence together used been, not be, before repaired, and thus did not replicate the structure required. Never mind what you want it to mean, it's not grammatical with that meaning.

So what to do with the been? Keep it; take off the parentheses. It can't be deleted under identity because it isn't identical.

  • It seems that intuitions differ on this. To me, [i] and [ii] sound equally bad without the second auxiliary, though I'm not sure I'd call either ungrammatical. This also doesn't explain why [iii] is valid in BrE but not AmE.
    – alphabet
    Aug 18, 2023 at 15:56
  • Multiple deletion rules involving meaningless auxiliaries are frequent sources of individual variation. The fact that every deletion rule produces multiple ambiguity is the reason. Everybody makes up their own grammar, after all, and many don't ever get around to those little details except sporadically and irregularly. Aug 18, 2023 at 16:08
  • 1
    But this fails at least for Br.E. (b) is preferable to (a) by far. I also question whether a listener would interpret (a) that way. They're far more likely to interpret it as "I don't think she's seen it, but she may (see it in the future)", methinks. Lastly, we don't need conjunction reduction to delete a VP after an auxiliary, do we? We can do that without permission from a conjunction environment. Just for the hell of it. Aug 18, 2023 at 16:46
  • 1
    CGEL makes it clear that there's nothing ungrammatical about your (b). It's fine in BrE too.
    – BillJ
    Aug 18, 2023 at 18:52
  • 1
    In my dialect "I don’t think she has seen it, but she may" unequivocally means I don’t think she has seen it, but she may decide to go see it in the future. This probably has to do with may/might usage - I can't use may in the sense of uncertain results, only in the sense of opting to do something. And may in that sense focuses on the act of choosing, so it isn't available in the past tense. There is no "may have seen it", only "might have seen it."
    – Phil Sweet
    Aug 18, 2023 at 21:29

This construction is subject to some regional variation: AmE doesn’t allow [iii], but it does have [i–ii], even if they are less usual than versions in which the second auxiliary is omitted.

The claim that the second auxiliary is usually omitted is only intended to apply to American English. I'm fairly confident that CGEL is just wrong on this point. I would prefer to include the second auxiliary in both sentences; the versions without them sound awkward, at least with the intended meaning, though I wouldn't call them ungrammatical.

In general, I've noticed that CGEL's treatment of the differences between American and British English is often incomplete or questionable. Both Huddleston and Pullum are Brits; though they've spent a great deal of time in Australia and the US, respectively, neither is a "native" speaker of AmE.

  • Pullum was Prof of English at the University of California, Santa Cruz, from 1981 to 2007, so he knows what he's talking about. Do you actually possess a copy of CGEL?
    – BillJ
    Aug 18, 2023 at 18:49
  • 1
    @BillJ Yes, and I've read about 2/3 of it straight through thus far. I've found a number of places where CGEL omits (or gets wrong) differences between AmE and BrE.
    – alphabet
    Aug 18, 2023 at 20:22
  • I agree with your assessment that the acceptability of (a) and (b) is substantially lowered if the second auxiliary is omitted. But I disagree that omitting the second auxiliary in [i-ii] is only intended for AmE. If the omission were only intended for AmE, they would have used % before [i-ii] as well, a symbol used in CGEL to indicate that the examples are grammatical in some dialect(s) only.
    – listeneva
    Aug 19, 2023 at 1:18
  • @listeneva They think that omitting the second auxiliary is possible in both dialects (hence no "%" signs in [i]-[ii]), but that AmE prefers to omit it. Which is false.
    – alphabet
    Aug 19, 2023 at 1:27
  • @alphabet Where do they say that AmE prefers to omit the second aux?
    – listeneva
    Aug 19, 2023 at 2:13

Too long for a comment, so an answer. I confess I find the question and the examples at odds. It looks like the stranding examples are talking about eliding the predicates - ie. not repeating them - and the highlighted words are identifying the stranded forms. I don't see how they could be optional.

In [29]i - I can't use may. Some might choose to not use may - preferring might, but I may not use may and must use might. So "I don’t think she has seen it, but she might have." The have is required.

In [29]ii - has been is required.

In [29]iii - I have never heard the being included locally. It sounds odd, but doesn't seem to hurt anything or shade the meaning one way or another.

American English with a southern bent.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.