Funky a

« previous post | next post »

A couple of days ago, I took Roy Peter Clark to task for claiming that phrases like "a million dollars" show that the indefinite article a can be used with a plural head ("Slippery glamour", 7/4/2008). I observed that the structure is clearly [[a million] dollars], not [a [million dollars]]; that expressions like "a million" are just numbers, fitting into the normal syntactic slot where numbers go; and that million in this case is morphosyntactically singular.

In the comments, Russell Lee-Goldman pointed out that

There are, however, a few cases where it really looks like "a" is acting funky:

– He was there for a good seven years.
– An additional three people are required.
– A mere four nations recognize that standard.
– She collected an amazing and heretofore unprecedented forty million dollars.

Things like this are indeed quite common — here are a few from the web:

That was the year he twice smashed the world record, the second time dropping it to a stunning 19.32 seconds.
Her distance held up throughout that day although the sole American woman jumper came within a surprising seventeen meters of her.
The Pats scored a record setting 589 points this season (they beat out the Vikings who had 556 back in 1998).
The court record reflects an extraordinary seven days of hearings and argument related to Martin’s efforts to avoid being forced to testify under oath.
Kate Beckinsale suffered from a severe case of anorexia when she was younger, dropping down to a shocking 70 pounds (32 kilograms).
BlogRodent offers up 701 posts with a mind-boggling 281187 words for you to happily consume.
I had the lyrics written down – before I really knew what they meant – and I had the riff kicking around for a good ten years
Those that survive will head out for a final four hours in the desert under conditions simulating a real mission in the "Power Wear Off."

I'm no syntactician, and I don't have access to CGEL or other wide-coverage grammatical descriptions of English this morning. But these examples seem to me to represent a generalization of the phenomenon on display in phrases like "a million dollars": English number-expressions have inherited from their partitive history a limited ability to act like singular noun phrases.

However, I'll admit that the constituent structure doesn't feel like
[ [a <modifier> <number>] <noun>]
but rather feels like
[ [a <modifier>] [<number> <noun>] ]
— for what little those feelings are worth.

And with that, I'll turn it over to readers for (I hope informed) comments.

[Update — note work by Tania Ionin & Ora Matushansky, "A singular plural", and Torodd Kinn, "Demotion of numeral nouns from heads to dependents" (abstracts from a 2004 Workshop on Numerals in the World's Languages).]


  1. Claire said,

    July 7, 2008 @ 6:26 am

    I can't say I would necessarily consider myself "informed," except in the sense that I am a native speaker of (American) English. But it seems to me that these phrases have an implicit " of" of some sort. For instance:
    a stunning [record/time of] 19.32 seconds. a surprising [distance of] seventeen meters a shocking [weight of] 70 pounds&c.

    My impression here is probably incorrect, possibly an overzealous attempt to find a Reason. But I thought I'd share it.

  2. Brett said,

    July 7, 2008 @ 6:27 am

    The CGEL briefly mentions this on p. 353, but when I asked Rodney Huddlesont about it, he wrote back and indicated that he wasn't completely comfortable with the analysis there and that he would spend some time thinking and consulting about it.

  3. Mark P said,

    July 7, 2008 @ 6:30 am

    I would say — "a million dollars is a lot of money" and never "a million dollars are a lot of money". Which means to me at least "a million dollars" is singular.

    Likewise, when I can make them the subject, the other phrases all seem singular to me — " a good ten years is a fair time to spend" but not "a good ten years are how long it took".

  4. Adam B said,

    July 7, 2008 @ 7:27 am

    @ Mark P:
    I'd say also things like "A good fifteen people were in the phone box"

  5. dd said,

    July 7, 2008 @ 7:35 am

    This construction does seem to "singularize" the phrase, lending support to the [ [a ] [ ] ] construction.

  6. dd said,

    July 7, 2008 @ 7:36 am

    hm, serves me right for copying and pasting? The construction I posted was the 2nd one in the article, to clarify.

  7. Adam Roberts said,

    July 7, 2008 @ 8:25 am

    A similar funky a, common on the school playgrounds in the UK, and deriving (I think) originally from the a BBC comedy programme, The Catherine Tate Show. Tate's schoolgirl character is cheeking one of her teachers: she says something outrageous, he replies 'I beg your pardon?' and she ups the outage stake by retorting: "Sir, are you deaf? Are you a deaf?' The shift from using a word as an adjective to a noun (the second functions as a compacted version of 'are you a deaf person) is signalled by the funky a; but what's interesting about it, I think, is that it is language use that gets a laugh, because it is novel and unexpected, whilst also being comprehensible. In other words, the funky a in this case is a comic marker. (I've also heard 'are you a gay?', where the intention is to be ironic and amusing rather than just offensive)

  8. Adam Roberts said,

    July 7, 2008 @ 8:27 am

    "…deriving (I think) originally from the a BBC comedy programme…"

    That funky a is inadvertent, I'm afraid: "…deriving (I think) originally from the BBC comedy programme…"

  9. James said,

    July 7, 2008 @ 8:58 am

    "A million dollars is a lot…" does make 'a million dollars' seem singular, but consider

    "Twelve people is too many (to take in your minivan)."

    Since it's 'too many', the subject *must* be plural (right?); but the 'is' is singular.
    Geoff?? Rodney?? Help!

  10. MM said,

    July 7, 2008 @ 9:02 am

    I'm a bit out of my depth here (agree with the second lot of square brackets), but used to introducing foreign learners to this feature of English (and possibly skewing the observations in doing so). I agree with Mark P.
    A thousand dollars is a lot of money / Fifteen people were in the phone box … The former is thought of as a unit, the latter as individuals.
    It would be rare for a sum of money to be regarded otherwise. Possibly more variation with drinks: six pints isn't enough to get me drunk – three whiskies were lined up on the counter.

  11. Radek said,

    July 7, 2008 @ 9:17 am

    Two arguments favor Claire's suggestion, in my opinion. Take the following sentence as an example (from the original post):

    BlogRodent offers up 701 posts with a mind-boggling 281187 words for you to happily consume.

    The first argument is semantic. The adjective "mind-boggling" does not modify "words". In that case one would have to say "281187 mind-boggling words". The adjective really modifies the dummy word "amount".
    The second argument comes from word-order. If my non-native intuition is correct, one cannot say "I bought tasty two apples" but it is fine to say "I bought two tasty apples". I.e. the modifying adjective always follows the numeral, unlike in the construction discussed.
    So the structure seems to be [funky a [Adj [ amount [ of [numeral [NP]]]]]].

  12. Steve said,

    July 7, 2008 @ 9:52 am

    @ Mark P – I think that 'a million dollars' or 'a good ten years' take a singular verb in his examples because 'money' and 'time' are both uncountable, and therefore we treat the subject of the verb as uncountable too. Compare 'A million dollars are spent in this shop every week' and 'A good ten years have passed since I saw him' – where a plural verb seems to me preferable, although I admit that a singular verb is possible too.

    The use of the singular article seems to me to be connected with the presence of an adjective, rather than a singular/plural or countable/uncountable noun. If we move the adjective next to the noun instead of the number, there is no longer any need for an article – 'a good ten years' but 'ten good years' – though of course the meaning is different too. So I would certainly agree with Mark Liberman's intuition about the underlying structure.

    I think Claire's impression about an understood 'noun + of' is on the right lines, though possibly not the full answer. The 'of' is perhaps not particularly significant, but the implied (singular) noun is surely what justifies the singular article, even when – as in my example 'A good ten years have passed' – the verb is plural.

  13. Jeff Petersen said,

    July 7, 2008 @ 11:26 am

    It seems from the given examples, you can look at what "a" is attached to by seeing what you can eliminate and still have a reasonable sentence.

    "the second time dropping it to 19.32 seconds" makes sense by eliminating "a stunning" but "the second time dropping it to a 19.32 seconds" does not.

  14. Henry said,

    July 7, 2008 @ 12:12 pm

    I agree with Steve about the difference between countable and uncountable. "A million dollars is a lot of money." "A million dollar bills are a lot of money".

  15. Russell said,

    July 7, 2008 @ 12:12 pm

    I decided that a more full discussion (on my part, anyway) was in order, so I put up a more lengthy reply on my blog (direct link here). But a few notes here might be useful.

    The singular/plural variability is, for those fortunate enough to have it, discussed in the CGEL around page 501. For those not fortunate enough, the bit of particular relevance concerns "measure phrases" like "a number of X" and "heaps of X." Depending on the particular "measure word" and the surrounding context, you can get either singular or plural agreement (such variability also happens with plain plural noun phrases, but it's more apparent with measure phrases). That probably says that Claire and Radek are on the right track — though one need not invoke unpronounced or implicit *words*, but at least implicit semantic structure specially created by this combination of adjective and quantified noun (you could call it a "construction"). It also meshes with dd's idea that the adjective "singularizes" the nominal — it places a semantic measure-y thing around it, so to speak. If that's right, It's not immediately clear to me what, if anything, that says about the syntactic constituency.

    I have the vague feeling that "to play a solid third base" and "to sing a mean alto" are in the same family of slightly-odd modification and determination. Maybe also "the always-controversial San Francisco".

  16. BFB said,

    July 7, 2008 @ 2:46 pm

    "Twelve people is too many (to take in your minivan)."

    In this and some of the other cases, the only strangeness is that a collection of discrete units is being treated as an amount.

    "It can take a three-ton load."
    "Can it take four tons?"
    "That's too many…or too much."

    And I think that's okay. English allows both of those.

  17. Josh Millard said,

    July 7, 2008 @ 2:56 pm

    Minor detail: my instinct is to to treat the "a" in "a million dollars" as different from the "a" in "a heretofore unprecedented forty million dollars". While the latter case is kind of the interesting meat in here, I'm not sure if anyone has underscored that in the first case, "a" functions (or at least can be read to function) as simply an analog to "one".

    Which may be blindingly obvious, which itself may be why no one has bothered to make the point. But the path from "a million dollars" to "a stunning one million dollars" involves changing "a" to "one" and folding in a whole new "a" for the construction, not sticking some modifiers between the original "a" and the "million dollars".

    (Which is to me a neat little thing I'd never considered before — that "a-as-one" can collapse under the weight of a construction like this. You can't very well say "a stunning a million dollars". But that may be wandering too far from the premise.)

  18. John Baker said,

    July 7, 2008 @ 3:06 pm

    The treatment of quantities as singular, as in "A thousand dollars is a lot of money, and two thousand dollars is even more," is well-established. It's on page 56 of the Merriam-Webster Dictionary of English Usage, for example. I'm not sure what's going on with the other examples, which seem generally to be in the form of a + modifier + number + plural noun. They aren't just quantities treated as singular, or they wouldn't take plural verbs, as in the given example, "A mere four nations recognize that standard." My intuition tells me that whatever is going on does not undercut the status of "a" as singular, though I'm not sure exactly why.

    What does "a" modify in these examples? Surely it modifies the number or the intervening modifier, rather than the plural noun.

  19. Paul Wilkins said,

    July 7, 2008 @ 4:03 pm

    Is the shoehorn a grammatical devive?

    Most of the examples are perfectly sensible with the a-clause removed. Such as "scored a record setting 589 points this season" v. "scored 589 points this season". Record-setting is acceptably jammed in there with the a-antecedent, when any number of other tricks coulda been used, such as a subordinate clause (do we still use them in English?), parens, a prepositional phrase or even another sentence.

  20. ohioans everywhere said,

    July 7, 2008 @ 4:25 pm

    What does "a" modify in these examples? Surely it modifies the number or the intervening modifier, rather than the plural noun.

    As I understand it, "a" modifies the quantity, which is expressed as compound of the number plus the plural noun.

  21. sidereal said,

    July 7, 2008 @ 5:25 pm

    I would say — "a million dollars is a lot of money" and never "a million dollars are a lot of money". Which means to me at least "a million dollars" is singular.

    I think you've rigged the game a bit by including the 'a', which already presupposes singularity.

    Consider 'Those million dollars' vs 'That million dollars'. Both sound okayish to my ear, though the latter is certainly preferred.

  22. Michael said,

    July 7, 2008 @ 5:30 pm

    Re: MM's example of "six pints isn't enough" vs. "three whiskies were lined up," can't the former be read as basically an elliptical construction for "six pints of beer isn't enough," where the natural tendency is to make the verb agree with the nearest noun, rather than the actual head ("pints") in front of the prepositional phrase? There can be no such ellipsis in "three whiskies."

  23. Eric said,

    July 7, 2008 @ 6:10 pm

    I agree with Russell's discussion mentioning "measure words" or "measure phrases."

    In English, they're not discussed on their own very often, but I think anyone who has studied an East Asian language would probably come to the same conclusion that they are relevant in this sort of discussion of the foibles of plurals in English.

  24. dr pepper said,

    July 7, 2008 @ 9:40 pm

    So how does this work in french, which has only one word for "a/one".

  25. Mark Young said,

    July 7, 2008 @ 10:21 pm

    Isn't what you're talking about here part of a wider linguistic phenomenon? Googling "a pre-Taxi Judd Hirsch" gets four hits (plus duplicates), in examples such as:

    In this earlier incarnation, a pre-Taxi Judd Hirsch is Sgt. Dominick Delvecchio, a no-nonsense Los Angeles cop who balances his detective work with studying to become a lawyer.

    "A Chastened Imus Returns to Radio" (here) looks to my untutored eye to be the same kind of construction as "A mere four nations recognize that standard" (from the article above) — a case where adding the adjective requires adding an article to introduce it.

    Number/quantity doesn't seem to be a necessary part of the funky a….

  26. Rubrick said,

    July 8, 2008 @ 12:02 am

    My own intuition is that the presence of the adjective is somehow singularizing the noun. "A good forty yards" feels parallel to "A dozen roses". This makes some sense, I think: "good" isn't modifying "forty", nor does it apply to each yard in the set (which, as others have pointed out, would be "forty good yards"). "Good" is turning "forty yards" into a singular unit.

    On the other hand, now that I think about it, "A dozen roses" is rather odd. Is "dozen" the only quantifier in English that doesn't require "of" (a gross of popsicles, a crapload of spam…)?

  27. Russell said,

    July 8, 2008 @ 1:06 am

    Mark Young: it seems so to me (see the last part of my comment above). However, it's interesting to consider the sorts of adjectives that are licit for quantified nouns as opposed to, say, proper names. For quantified nouns, there are three major categories: evaluative words like "staggering" and "good", historical words like "unprecedented" and or amount-related words like "estimated" and "mere". Other possibilities are there, to be sure, like "steady" (often in rate expressions) and "normal." Perhaps due to the syntax, or maybe the meaning, of proper names, there are less restrictions on the adjectives there.

    Rubric: I have the same intuitions about the "singularizing" effect of adjectives (also dd, above), and I think you could read Mark's constituency structure as reflecting the semantic intuition that the adjective modifies the number+noun combination, not just one or the other.

    As for quantifiers without "of," there is "all," "many," and "some" have both nominal and determinative uses: all/many/some (of) children. Only "all" acts like a noun that takes on of-less complement: all the children. If you consider "times" a quantifier in "hers was three times the size" then that might count. And finally, there is "hella," as in "there are hella ways to talk about amounts." But maybe the "of" is still there.

  28. Tarlach said,

    July 8, 2008 @ 5:10 am

    There is something I'm starting to notice about Language Log. People who post on here write about things about language that are obvious to people who use the language. But to the people posting ABOUT the language, they act as if there is some strange interesting phenomenom going on, and usually in a condescending way. As if the people who use and understand language are the weird ones, not the ones who look back upon language through the scope of their arbritrary rules.

  29. Brett said,

    July 8, 2008 @ 8:29 am

    The structure that Mark Young noticed above (i.e., a pre-Taxi Judd Hirsch is, I believe, one that Geoff Pullum noted here
    And my sense is that the two situations are related.

    Rodney Huddleston has gotten back to me with an explanation, which I've tried to convey here:

  30. Amy Vaughan said,

    July 8, 2008 @ 8:53 am

    @Adam Roberts: Similarly, my friends and I will occasionally accuse each other of being 'a dumb' or 'a fat'. We never heard of Catherine Tate until she got on Doctor Who, but. What I wonder is if you can really call it a 'comic marker' so much as a 'comic lack of a noun' – the common thread seems to be purposely creating an illusion of linguistic incompetency by leaving off what everyone knows should be there for comedic effect – so, rather than (*a) deaf, a deaf *(person).
    I would think of some tests, but it's early in the morning for me!

    @Tarlach: Don't you think gravity is pretty obvious? Then why have we spent so much time documenting and studying it? Why do we still care about gravity? Check yourself for hindsight bias.

  31. Edward Vitasek said,

    July 8, 2008 @ 2:02 pm

    Russel: ***I have the vague feeling that "to play a solid third base" and "to sing a mean alto" are in the same family of slightly-odd modification and determination. Maybe also "the always-controversial San Francisco".***

    I'm getting an idea that supports the [ [a ] [ ] ] reading. Compare the progression:

    1.a) Kurt Russel made [[a surprisingly good] [Elvis Presley]].

    No problem, here. "Elvis Presley" is clearly a category, and the adjective "good" describes the implied evaluatee "performance as" with the evaluator "likeness". It's both semantically and syntactically unremarkable.

    1.a) In this picture you can see [[a younger] [me]].

    Again, the speaker transforms himself into a category for the sake of comparing age.

    2.a) [[An additional] [three people]] are required.

    Now this is, I think, the same principle, constructionwise. You transform "three people" into a category and give them the attribute "additional".

    In your blog you say:

    ***It seems sort of limited: a grueling 100 miles, but ?an asphalt-paved 100 miles.***

    Here, I'd argue, it's not so much a question word-classification but of "evaluator-appropriateness". I'd have no problems with:

    "I'd rather drive across an asphalt-paved 100 miles than across 25 miles of gravel."

    To summarise: "Five dollars" may be "a measly five dollars" if you're two people and intend to go watch a movie together, but they may be "a whopping five dollars", if you just want to buy a popsicle. To me that makes "five dollars" a semantic unit.

    So does the indefinite article have a sort of categoriser function, here categorising "[number] [noun]"?

  32. old maltese said,

    July 8, 2008 @ 4:36 pm

    Claire and Steve may be on to something, particularly regarding the assumed 'of'.

    We'd say 'a fistful of dollars' or 'a stack of dollars'.

    Maybe we started with the parallel 'a million of dollars', and then dropped the 'of' as assumed.

  33. blahedo said,

    July 9, 2008 @ 1:12 am

    Part of my master's thesis demonstrated the necessity of a distinct phrase type NumP for numeric modifiers, with a unary rule in the system that could promote NumP into determiner for the standard analysis. The evidence for this had to do with the fact that the word 'each' (and similar words like 'altogether' and 'apiece') used binominally requires its distributed phrase to be "quantified" by a number or number-like phrase, but no actual quantifier (though "a" can act as "one" here and is moderately acceptable):

    Alex and Sasha read three stories each.
    *Alex and Sasha read the stories each.

    There's a bit more to it, but this margin lacks the space to detail it. :) (In the thesis, Chapter 2 has the long list of examples, and Chapter 6 the main analysis.) The reason I bring it up here is that the "funky a" constructions here *can* occur in this NumP slot:

    Alex and Sasha read a surprising three stories each.

    I take this to be very strong evidence that the 'a' is attached to the number before the noun: (NP (NumP a surprising three) stories).

  34. Brett said,

    July 9, 2008 @ 6:05 am

    blahedo, if a is attached to the number, how do you deal with:
    [a surprising three] days and five weeks

  35. Brett said,

    July 9, 2008 @ 7:28 am

    Rodney Huddleston has now clarified my poor attempt to explain his position. His explanation can be found here:

  36. John Payne, Rodney Huddleston said,

    July 9, 2008 @ 11:18 am

    The suggested analysis of (1) is precisely the one proposed in The Cambridge Grammar (CGEL), p.353:

    (1) a measly [million bucks]. And not (2):

    (2) *[a measly million] bucks.

    The evidence in favour of (2), although there was not enough space to
    present it in CGEL, is constructions like (3):

    (3) a measly [three weeks and two days],

    where "measly" evidently applies to the whole time period denoted by the nominal (Nom) in square brackets.

    Concretely then, in (1) "a" is the indefinite article functioning as
    determiner, "measly" is an adjective modifying the nominal "million
    dollars". In "million dollars", "million' is a numeral modifier.

    The oddity/funkiness of the construction is the ability of the indefinite
    article, in conjunction with a set of adjectives like "mere", "good",
    "additional", to select a plural nominal containing a numeral modifier.

    On the other hand, CGEL takes the "a" in "a million", just like the "one" in
    "one million", to be a component of the numeral. The analysis of "a
    million bucks" is then (4):

    (4) [a million] bucks

    The expression "a million" here follows rules for the internal structure
    of numerals, see CGEL pp 1715-18, and functions, as a single unit, as
    determiner. The "a" which occurs in "a million" however only occurs when
    this and analogous large numerals function as determiners. Thus it is
    impossible to have "a", unlike "one", in (5), where "million" is a

    (5a) the [million] bucks which you lent me
    (5b) *the [a million] bucks which you lent me

    Contrast (6):

    (6a) the [one million] bucks which you lent me.

    For the rule governing the appearance of "a" in numerals see CGEL p372.

    We note finally that it would have been better if we had said in
    heading (c) on p. 353 of CGEL "Dependents or sequences of dependents". In "a good three hefty steaks" we have "a" as determiner and "good" as modifier: "a good" is not a single dependent.

  37. Russell said,

    July 9, 2008 @ 11:23 am

    Edward: I'm not sure what the motivation for [ [a surprisingly good] [Elvis] ] is, as opposed to [a [ [surprisingly good] [Elvis] ] ]. In fact, if the speaker "transforms himself into a category for the sake of comparing age," this suggests to me [a [younger me]], rather than [[a younger] me]. But I must be missing something. Personally, I am wary of using my intuitions about a phrase's meaning to diagnose its syntactic structure.

    About the asphalt-paved 500 miles, I agree. I think now that time/distance uses are different. See the link in Brett's post for a good contrast between "surprising" and "exuberant."

    blahedo: Cool thesis! I'll have to take a look at it. Does it handle "at a time" or "in a row"? (a couple of nasty beast that I thought about for a bit).

  38. AEW said,

    July 9, 2008 @ 1:42 pm

    As a non-linguist, this doesn't seem too mysterious, but maybe I'm missing something.

    The original examples all seem to focus on the number rather than the collection of things. The first three examples can be transformed along the lines of

    "Out of one hundred and thirty-two nations, a mere four recognize that standard."

    So, I guess to my ear it's [a mere four] nations, not a mere four [nations].

    John Payne and Rodney Huddleston seem to refute this with

    (3) a measly [three weeks and two days],

    But still the emphasis is on number rather than collection. (3) seems like an extension of the use of, say, "a lovely wife and child" for "a lovely wife and a lovely child," (Even though "a measly three weeks and a measly two days" doesn't quite make sense.)

    Similarly, you could say both

    "Six minutes is a good pace," and

    "Six minutes and thirty seconds is a good good pace."

  39. Edward Vitasek said,

    July 10, 2008 @ 3:10 am

    @Russel: Actually, I'm not sure what I thought myself. I began having doubts on my way to work. (I was in a hurry; I'm not posting hastily again.)

  40. blahedo said,

    July 10, 2008 @ 3:10 pm

    @Russell "Does it handle "at a time" or "in a row"?":

    Those didn't come up at the time, but it's plausible. "At a time", at least, seems to have similar requirements for numeric quantification, and semantically it feels like it's doing something with the distributivity; on the other hand, the distributivity is more to do with time than with distributing over a Range of actors (the "subject" in the simple examples). Maybe since my analysis was extensional only, 'at a time' will do something with a temporal parameter? "In a row" seems a bit more like a straight-up adverbial phrase, other than requiring plural constituents somewhere.

    @Brett "[a surprising three] days and five weeks":

    Hmm, I have no idea. That definitely requires further thought.

RSS feed for comments on this post