pro-predicate do and verb phrase ellipsis

Have you read past the scary title of this post? Glad you're still with us! The phenomenon in question is how AmE and BrE speakers differ in their preferences for avoiding repetition of complex verb phrases in main clauses. (Still here?) So which of the following would you say?
(1) I ate all the chocolate, even though I shouldn't have done.
(2) I ate all the chocolate, even though I shouldn't have.
If you answered "(1)", then I'd be willing to bet that you're not American. Kevin of Berkeley, California wrote to me about this back in April, saying:
I particularly wonder if the American formulation is as jarring to British ears as theirs is to mine.
(I'll leave it to people with British ears to answer the 'jarring' point.) Since this type of construction was one of those things that I had in mind when starting this blog, I'm fairly surprised that I haven't given it proper coverage yet. I guess I've put it off because I feel the need to go over some basic grammatical concepts first. And then I got slowed down by an obsession with using sentence trees to do so. But while walking home from yoga class tonight (with my mind all open to startling truths), I reali{s/z}ed that one rarely makes new friends by presenting sentence trees to them. So, let's see how well I do without.

First, a little sentence anatomy. Both sentences (1) and (2) above are made up of two sentences (clauses) joined by a conjunction (but). The two clauses are: I ate all the chocolate and I shouldn't have (done). The second clause in both cases means 'I shouldn't have eaten all the chocolate', and in both cases the speaker is avoiding the awkward repetition of a form of the verb eat plus its complement (which in this case is a direct object) all the chocolate. So, eat all the chocolate is old information that doesn't bear repeating, but we have new information to impart, the feeling that the chocolate-eating was in some way a bad thing to do. So, we want to say the clause while leaving out the old information shown in brackets here:
(3) I shouldn't have [eaten all the chocolate].
The usual AmE solution to this problem is just not to say the bit in the brackets. (Bit is such a BrE noun to use, but not so exclusively BrE that I feel comfortable marking it as BrE.) This leaves a sentence without a full verb phrase (or predicate in traditional grammar terms). We have the modal verb (should), the negative marker (n't) and an auxilliary verb (have), which gives tense and aspect (the when and how-it-relates-to-time) information, but no main verb (the heart of any complete sentence) or complements (elements that the verb requires in order to make a complete verb phrase). The continuation of the verb phrase is just understood from context. This leaving-understandable-but-grammatically-important-things-out business is called ellipsis, and we are left with an elliptical construction.

In BrE, however, there is a preference for having a complete clause in these situations, with a main verb included. So, how do you do that without repeating a lot of already-heard, understandable-from-context words? You use a pro-verb (not the same as a proverb! Sometimes hyphens are important!) or a pro-predicate.

You might not have heard of a pro-verb or pro-predicate before, but you've probably heard of their cousin, the pronoun. All of these are pro-forms, that is to say, words that stand for a word/phrase whose meaning is recoverable from context. (English also has pro-adverbs.) If we wanted to use a pronoun to solve our problems with the 'eating all the chocolate' sentence, we could say (4)...
(4) I ate all the chocolate, but I shouldn't have eaten it.
...with it standing for the phrase all the chocolate. But that's still pretty repetitive.

What BrE speakers typically do here is to use do as a pro-predicate that stands for the main verb and its complements (at least). So done in (1) above stands for eat(en) all the chocolate.

Why does this grate on the ears of some AmE speakers, like Kevin? Because we just don't like using a pro-predicate with auxiliary or modal verbs in main clauses (see below for when we do use it). We (and BrE speakers too) are able to use do as a pro-verb, as in (5) where it stands for the main verb eat and nothing else, or as a pro-predicate that stands for an entire verb phrase (without modal or auxiliary verbs--we refer to these collectively as support verbs) as in (6).
(5) I ate all the chocolate, but I shouldn't have done it. [do= 'eat']

(6) I ate some chocolate, and Better Half did too. [do = 'eat some chocolate']
But most AmE speakers cannot use pro-predicate do in a clause with support-verbs in it, as in (1) above. (Note that do has other non-"pro" uses too, and so may be used with modals and auxiliaries in those cases.) There are some AmE dialects that are more tolerant of mixing support-verbs. See this article from American Speech by Mariana di Paolo for an example.

BrE uses support verbs with pro-predicate do very freely. So any of the following could be your answer to the question Have you sent Lynne any chocolate yet?
I have done.
I haven't done.
I will do.
I might have done.
I could do.
I could have done.
I should do.
I should have done.

(Note that the correct answer to that question should be the first one. Otherwise, go for the third one.)

On a(n) historical note, the aforementioned di Paolo article says:
Butters (1983 ["Syntactic change in British English propredicates" Journal of English Linguistics 16:1-6]) adds that pro-do was possible as long ago as Middle English although it was not common in England until about the 1920s in the written sources which have been examined. Butters also presents historical evidence suggesting that pro-do spread from subordinate clauses to main clauses in the early part of this century. Most dialects of present-day English, including American English, probably preserve the conservative forms in dependent clauses, as in the following example:
[...] I usually kinda take a back seat, which I know I shouldn't DO but...
So, we AmE speakers, like BrE speakers, can use pro-predicate do with support verbs in some clauses that are, like the above example, not complete sentences on their own (in this case the dependent clause is: which I know I shouldn't do). I'd have no problem (grammatically speaking) in saying the 'back seat' sentence, with pro-predicate do. But it's not quite as straightforward as 'propredicate do is good in AmE dependent clauses' because examples (1) and (2) above involve the subordinating conjunction even though, putting the shouldn't have (done) in a dependent clause. And I can't (in my native dialect) say that one, or include the do in this one:
(7) I usually take a back seat, even though I know I shouldn't do.
There might be a cline of 'subordinateness' operating here, with even though clauses 'feeling' more independent than which clauses, and therefore less likely to allow a pro-predicate do in AmE. (Or else the 'dependent clause' explanation of the exception is just too general/simplistic.)

Pro-predicate do is one of those Briticisms that I find myself using every once in a while, but I retain a certain self-consciousness about it. As well I should (do).


  1. (Am/E) I would either say, I shouldn't have done that. Or I shouldn't've, in a mumbled slide, as though awkwardly embarrassed.

    Of the later examples, in no small part because of military exposure, I would say
    Will do.
    I might've.
    I could.
    I could've.
    I should.
    I should've.

  2. Or, perhaps, I might could have sent you chocolates.


  3. In answer to Kevin's question, both 'I shouldn't have' and 'I shouldn't have done' sound OK to this speaker from the South East of England. I think I'm more likely to use the 'done' with contracted forms such as 'shouldn't've' or 'could've' rather than full forms such as 'shouldn't have' or 'could have'.

  4. PS

    >>The usual AmE solution to this problem is just not to say the bit in the brackets. (Bit is such a BrE noun to use, but not so exclusively BrE that I feel comfortable marking it as BrE.)<<

    So what would be the normal US version of 'the bit in the brackets' then?

  5. I remember noticing that my first son started using pro-verbs when he was tiny and only knew maybe ten verbs total. I observed this with awe and remember thinking it was evidence for neural hardwiring of grammatical structures. Was it?

  6. This was actually the topic of a paper I did back in College/Uni(versity). Though 'do' is probably the most common for this type of ellipsis, BrE also does it with 'have,' in non-auxiliary uses. My favorite example is from the Beatles' movie Help!:

    Superintendent : Oh come on now lads, don't be windy, where's that famous pluck?
    John : I haven't got any, have you George?
    George : Did have.
    Paul : I have had.
    Ringo : I will have! Lead on!

    (now I'm looking at my sentence trees and deciding you did the right thing)

  7. Zhoen, your examples are all fine in AmE or BrE. Notice that when you use shouldn't have done, you're putting the pronoun that in there, so using do as a pro-verb, but not a pro-predicate.

    Bingley, Jonathan's right that part would be more AmE than bit, but note that I didn't use parentheses there, I used, in the AmE sense (which works in BrE too in this instance) [square] brackets. If they'd been rounded like (this), then they'd be parentheses. I discussed that back here. (Leave new comments on brackets/parentheses back there, please.)

    Robert, using rather 'empty' verbs like do is fairly typical in English-acquiring children's early stages. They allow the child to start to string together sentences without having a big verb vocabulary. It's clever way to not let one's vocabulary restrictions stop one from acquiring grammatical constructions.

    Thanks for the example, Joel!

  8. Sorry Lynneguist, I just don't get this. Maybe I'm being obtuse. Though I am a BrE speaker your AmE examples sound better to me than the BrE ones. (2) and (6) are the only sentences that sound OK to me. (1) sounds clumsy and (7) horrible - is it meant to?

  9. 'I shouldn't have' - sounds right to me (Scot)

    'I shouldn't have done': altogether too gor-blimey for me.

  10. BrE speakers, does 'shouldn't have done so" sound any better than "shouldn't have done"?

  11. I (ScE) am with Dearieme on this one. And Elena, to me "I shouldn't have done so" sounds rather formal and a bit overcorrect for use in a discrete sentence.

  12. The Irish are with the Scots on this: "I could do" et al usually seem English English rather than British English. I think I do add a "do" on occasion, but more often I don't do :p I'm not sure if it's prosody or clarity that triggers it for me.

  13. Cameron, it surely is too formal for a casual phrase about chocolates. But ignoring the style and the context (correct me if I am wrong), 'done IT' or 'done SO' somehow sound more natural than simply 'done'.

    Seems like most people are ok with ellipsis and the [pro-verb+pronoun/pro-adverb] deixis while only some BrE speakers are comfortable with the pro-predicate thing.

    [I am a native Russian who went to school in Michigan.]

  14. To be clear here: all of the non-pro-predicate examples are fine in any English. All can use ellipsis or pro-verbs. The 'should have done' kind of thing is more typical of British Englishes than American Englishes, but (a) its popularity seems rather new, and more common in speech than in writing, and (b) that doesn't mean that every BrE speaker says it. (People are notoriously bad judges of what they actually say though. Tape yourself for a week and have someone else transcribe may be surprised!

    If you are a BrE speaker, you should at least be used to hearing (if not saying) should have done and similar phrases...unless you only speak to Scottish and Irish people and don't listen to the television or radio!

  15. Great topic. I'm glad that the example from "Help!" was mentioned, because whenever I try to emulate this pattern, I can't quite get comfortable with using 'do' with as a pro-form for verbs that don't involve a suitable "do-ishness," if you will. Is there a distinction between phrases which take 'have' and those that take 'do' as the preferred pro-form?

    I note that to my American ear, the addition of 'it' makes all the difference with these pro-verbs. This is especially true when the 'have' portion is mostly elided ('I shouldn'a done it' [would anybody use two 'dos?' such as in:
    Q: Did you do it?
    A: I did do]).

  16. It's pretty rare that I hear the (1)-style 'do' in Australia. Being American, I also find it pretty jarring, so I think I'd notice it if I did hear it. But that Di Paulo paper says it's used in Australia. I wonder if any native Aussies can weigh in.

    As for the sentence about taking a back seat, it seems to me that as you're reading or listening to the sentence, you expect that the 'which' will modify 'seat', but then the 'do' pops you up to a higher level and makes the 'which' modify 'take'. In the second version, the 'even though' can only modify the whole first part of the sentence, so the 'do' is not necessary.

    That's supposed to be an explanation of why the first sounds OK to Americans, but not the second. I can't use the proper vocabulary, but I hope it still makes sense.

  17. @jhm
    Yes, 'did do' exists, but it is more equivalent to 'used to do so'. For example:

    A. I thought you worked in a bookshop.
    B. Well, I did do, but the pay just wasn't enough so I got a job with a publisher.

  18. It seems like 'have' is only used where the original verb was 'have,' in the transitive 'possession' sense. 'Do,' on the other hand, may be substituted for lots of (or any?) other verbs. I think it's the (apparent) use of 'do' as an intransitive verb that sounds weird to Americans.

    I'm not sure about age; the earliest example I know of is from The Two Towers:

    “Sam frowned. If he could have bored holes in Gollum with his eyes, he would have done."

    Also, my dad (native of Washington State, US, child of the 50s) has always said that nothing about this type of phrase sounds strange to him, and he felt like he'd heard it a lot from older people, but I think he might be an anomaly.

  19. Would you put the common AmE '...will do' into this analysis? I can imagine it in an example such as "I hadn't thought of that, but--will do" in which even the pronoun is left out.

    It's a frozen phrase surely--but those can be a lovely bit of amber that fossilizes a form that doesn't occur otherwise--no?

  20. I hear "will do" as an abbreviated form of "I will do it." The opposite of "will do" is "no can do." I'm American.

  21. Hmmm, dare I leave a comment?
    I've accidentally stumbled across your site, and was drawn in. I'm not a linguist, but I am a Utah native. I didn't know until today what "propredicate do" was, much less that it was uncommon. Learning new things can be hard, but I have done!

  22. Is it possible to use a double 'do'? For example:

    Q: Do you regularly eat marmite?
    A: I do do every morning.

  23. You'd better not use the phrase "do do" around an American fourth grader. The phrase "doo doo" (note spelling) is slang for sh*t.

  24. Native Aussie weighing in as requested... I would more naturally produce (2), and would expect to hear it most commonly, but it also wouldn't be unheard of for me to encounter or produce (1). I certainly have no sense that the (1) form is in any way jarring or awkward, although it probably does sound noticeably English English-y to me too.

  25. Bingley said...

    So what would be the normal US version of 'the bit in the brackets' then?

    In my native dialect (Educated urban midwest American -- Chicago)? It would be "the stuff in the brackets." Though my grandmother would chide me; she doesn't think 'stuff' is a 'ladylike' word. She'd probably say "part".

  26. Wow, Lynne! Thanks for such a delightfully detailed answer, and for the descriptive terminology, especially. You're welcome to throw sentence diagrams around anytime, as far as I'm concerned. And you've even garnered quite the assortment of helpful comments, too (thanks to all of you, too!).

    I honestly can't recall having heard this pro-predicate (nice terminology!) do from before about ten years ago, I suppose, but that's just my exposure. As noted, I've always found it jarring--not wrong, per se, as I recognized what the do was doing. But it was just uncommon enough in my dialect that it draws attention. It's one of those oddities of American vs. British English that would be perfect for spotting spies, I would think!

    Thanks very much for addressing it. I'd be delighted to hear more!

  27. I hear "will do" as an abbreviated form of "I will do it." after a question like "Will you call me with the results?" the response "will do" substitutes "do" for "call you..."

    I do do every morning.

  28. Lynne,

    I do believe this is my favoUrite of all of your blog items! What an interesting subject -- and so well-described!


  29. Glad so many people liked the topic...I really should have done it much earlier!

    Catching up on comments...

    Wishydig (and Ken): I'm more accustomed to hearing AmE will do as a phrase on its own, without a subject, e.g.:
    - Could you send me some chocolate?
    - Will do!

    Rather than as I will do (and probably not any other person as subject either). It comes with its own response-type intonation. So in those ways, it seems more like a set phrase and less like the grammatically productive thing that I've described for BrE--as you say. But, do can certainly be classified as a pro-predicate there. It's standing for another verb phrase.

    James, I think your marmite example works. I found this one in a play by someone in Nottingham:

    Bill (Indicating whiskey): I’d have thought you'd have given that stuff up by now

    Alan: I do do - every morning - then at night I need the crutch to help me through

  30. For what it's worth - my advisor in grad school was from British Columbia and she said "could do" all the time. She's the only person I've ever heard use that construction in real life, so I always assumed it was a Canadian thing.

  31. Is that just could do without a subject? I think that's like will do...more of a set phrase than part of a grammatical pattern.

  32. But, dear lady, you have yet to face the challenge that is "I could do a pie".

    Do you ever do a pie before five?
    Oh, I do do do sometimes.

  33. Dearieme, what's given you the impression that I'm a lady?

    Do a pie doesn't as far as I can tell involve pro-predicate do or ellipsis, so I didn't think it belonged here. And besides, I thought we were talking about chocolate!

    But perhaps you'd like to shed light on the subject?

  34. Errr! Sorry dearieme. I ate all the pies.

  35. Oh dear. To "never comment before eating" I must now advise myself "never comment after Riesling".

  36. I just... love reading this blog.

  37. When I was a (n American) kid my friends and I loved this phrase from the English children's books we read avidly: "Do let's!" "Shall we walk with Nanny in Kensington Gardens?" "Oh, do let's!"

    So: any relation?

  38. nbm (even though you've got my mother's initials, I'm going to assume you're someone else!), do could be considered to be a pro-predicate here, but there's more going on...more than I think I want to go into in these comments. There are probably a half dozen more posts to write about do!

  39. Turner Classic Movies in the States shows movies from the 30s. In the Thin Man, I believe, you will hear some of the Americans using the Br/E pro-predicate do.

    Now, this could be the accent of the upper class in the US, but I don't hear it anymore in the States.

    Lynne, have you ever looked a modal differences between Am/E and Br/E? An interesting one is the I would/ I should for giving advice.

    Q: What is the best way to get to X?

    Am/E: I would take the train.
    BR/E: I should take the train.

    Surely there must be other differences.

    Bob Yates, University of Central Missouri

  40. Haven't done much on modals yet, and the situation is VERY complicated. As usual, I'd like to request that it NOT be discussed here in the comments, as it decreases the searchability of topics on the blog. (This is why I ask for requests for new topics to be e-mailed rather than left in comments.) If you want to see what else has been done on verbs on the blog, hit the 'verbs' tag at the bottom of the entry, and it will take you to other so-tagged entries. Ta!

  41. Re: the word "stuff" not being ladylike...I figure any word that appears in the KJV of the Bible is good sturdy English. That means "piss" is okay too. LOL

  42. New data point:

    Hank Williams in his song "Why Don't You Love Me" sings "Why don't you love me like you used to do/How come you treat me like a worn out shoe?", and it's hard to get more American than him.

  43. shouldbewrittingapaper;thisismorefun30 September, 2009 17:30

    I'm native to Utah so this is particularly interesting for me. Marianna Di Paolo is my professor for Varieties in American English. I'm finishing my bachelor's degree in Theoretical Linguistics and this semester is the first time in my life that I found out that this was an abnormal construction for AmE speakers. What I find odd is that I'm totally fine with sentences like:

    "My husband didn’t help a lot on Thanksgiving, but he did do more than his

    "I ate all the chocolate, which I shouldn't have done."

    But this sounds marked to me:

    "I ate all the chocolate, even though I shouldn't have done."

    Well, perhaps I should say that I wouldn't produce it, but I'm not sure I would even notice if someone else produced it. So not marked, just not natural to my mouth.

  44. You describe a typical American pattern, Shouldn't. Nothing too unusual there!

  45. Thank you so much for explaining this construction. I had noticed it before in the Harry Potter books and on some BBC series and wondered what was going on. It sounds so foreign to my AmE ear that I couldn't really come up with any usage examples to aid in searching. (Googling can be tricky when two of your main search terms are "word" and "do.") I should have thought to come here first. Oh well, in the future I will do. Thanks again!

  46. Thanks for the posts. Great fun. My wife and I are Americans living in England and the 'do' on the end just grates. Sounds redundant.
    'Can you take me to town?'
    'I can do'
    'Do you work in town?'
    'I do do.
    Second one doesn't work. But I hear the first often. Though it is true that it seems to be a broken rule. No one uses it all the time. It just pops up here and there.

  47. Massachusetts-

    With the increasing use of text speak it is now possible to reduce duplication even more.

    I ate all the chocolate. :(

    Though that could be construed at disappointment in there being no more chocolate, rather than shame at being such a glutton.

    More sophisticated emogees might be clearer.

    "I usually kinda take a back seat, which I know I shouldn't DO but..."

    While I would tolerate do here, I still prefer the clause without.

    In that particular sentence I would also let the comma which begins the clause do the work of which and strike which as well.

  48. James

    Hank Williams in his song "Why Don't You Love Me" sings "Why don't you love me like you used to do/How come you treat me like a worn out shoe?", and it's hard to get more American than him.

    Two points

    1. Do is a very useful rhyme. and used to is handy for getting it to the end of a line.

    2. Hank listened to a lot of blues.

    My blues concordance has

    Said hear me mama : who in the world been telling you
    You don't even treat me : nothing like you used to do

    Says my woman she quit me : keep me worried and blue
    Take me in your arms and love me : like you used to do

    Says you three time seven : and you knows what you want to do
    Sometimes you going to think : about the good things I used to do

    Come back baby : papa ain't mad with you
    Says I do just like : mama babe that I used to do

    And you stood and cried : what you want me to say to you
    I want you to think about the things baby : that me and you used to do

    So they can eagle rock me they can talk me : about the things that I used to do
    I got the Nehi blues mama : don't know what in the world to do

    I : ain't going to do like I used to do
    I'm going to stand right here : do the same old thing to you

    Now I want you to tell me baby : baby just what's getting wrong with you
    Now and you don't treat me : nothing baby like you used to do

    But the prize goes to a blues song much recorded in 1924:

    How come you do me like you do do do do?
    How come you do me like you do?

  49. Hooray, I'm not insane. I've been saying these extra occurrences of "do" seem odd, and maybe like a Britishism, and people have seemed to think I was speaking Martian.

  50. While those blues artists were singing used to do, Cole Porter was writing

    It's not 'cause I shouldn't, it's not 'cause I couldn't
    And Lord knows it's not 'cause I wouldn't
    It's simply because I'm the laziest girl in town

  51. In some of those song lyric examples (although not the Williams one), I don't think do is a pro-predicate at all. "The things we used to do" is like the places we used to go, the food we used to eat, the books we used to read, the songs we used to sing, etc. - the "do" is actually the real verb there.

  52. The reason that:

    - I usually kinda take a back seat, which I know I shouldn't DO

    sounds ok, Lynn, is that it uses lexical DO (or'main verb' DO if you prefer). This verb has a direct object, which is represented by a gap, and which is co-indexed with the word 'which'. In other words you can think of the word *which* as representing the direct object of DO there.

    The other exampes are off because, they genuinely do involve pro-VP DO.

    Incidentally, the American British split is not symetric. British 'unAmerican' nonfinite auxiliary DO is ungrammatical for most AmE speakers. But all the examples here are ok for BrE speakers, like me for example.

  53. I just linked to this post on Languagehat and was rereading the comments. Utah's variety of American English is known to have BrE influences in its phonology due to 19C immigration of English Mormons directly to Salt Lake City. The greater acceptance of pro-VP do in that part of the country may be related to this. (Unfortunately I don't have access to the di Paolo paper, which probably says just this.)


The book!

View by topic



AmE = American English
BrE = British English
OED = Oxford English Dictionary (online)