Accentuate the negative

« previous post | next post »

A curious case of a forced-choice sentence-completion question on a ninth-grade exam at a high school in Taiwan is briefly discussed on Lingua Franca today, for a very general non-linguist readership. It merits a slightly longer and more serious treatment, which I thought Language Log readers might appreciate. The exam question basically asks for a decision on the question of which one of these sentences is fully correct and which deserves to be called ungrammatical:

(a) Lydia knows few things, and so does Peter.
(b) Lydia knows few things, and neither does Peter.

Because continuation with neither does… is widely taken to be a test for negative polarity, this amounts to asking whether Lydia knows few things is a positive clause like Lydia knows everything or a negative one like Lydia doesn't knows anything. And a friend of mine in Taiwan reports having asked a number of English speakers, with a truly surprising result. He finds a split between the two great English dialect groups, the North American dialects (AmE) and the British and Australasian dialects (BrE). The AmE speakers that he asked all said (a) was correct, while the BrE speakers all said that (b) was correct.

There are other tests for negation than neither continuations (see The Cambridge Grammar of the English Language, pp. 786–7, drawing on "Negation in English" in The Structure of Language, ed. by J.A. Fodor and J.J. Katz, Prentice-Hall, 1964). One is particularly well known: reversed polarity tags. Positive sentences take negative confirmation tags (isn't it?) while negative sentences take positive confirmation (is it?). There is a minor difficulty with this: in addition to the standard negative confirmation tag on a positive clause, as in (c), which simply adds a chance for the interlocutor to confirm that the answer to the question in the tag is "Yes", there is the possibility of a snarly sort of sarcastically-tinged positive tag as well, as in (d):

(c) Lydia knows everything, doesn't she?
(d) Oh, so Lydia knows everything, does she? Well, let's see if she can name all eight of the Hawaiian islands correctly!

That means the two have to be distinguished, which makes everything a bit more subtle. So the most useful fact is that negative clauses simply do not allow negative tags in standard AmE or BrE:

(e) *Lydia doesn't know anything, doesn't she?.

That seems like a solid ungrammaticality, confirming that Lydia doesn't knows anything is negative.

Another test for a negation is continuation with not even, which negative clauses allow:

(f) Lydia doesn't know anything about football, not even the names of the teams.

Now, three further complexities turn out to be relevant. The first is nonverbal negation. Using certain words in certain positions, you can negate a clause without putting n't on an auxiliary verb or adding not before a verb phrase. Words like nobody, nothing, and never are examples, though this is somewhat more formal style. Thus (g) is a negative clause:

(g) Lydia knows nothing about football.

The tag-question and not even tests confirm it:

(h) *Lydia knows nothing about football, doesn't she?
(i) Lydia knows nothing about football, not even the names of the teams.

The second complexity has to do with approximate negation (CGEL §3.3, pp. 815–821). There are seven words (barely, few, hardly, little, rarely, scarcely, and seldom) that negate a clause in a sort of statistical way that allows for a relatively modest number of exceptions. Absolute negation is illustrated by Pandas never eat meat, which is falsified by any occurrence of a panda eating meat (it is false, in fact, since occurrences of a hungry panda departing from its vegan bamboo diet by catching and eating a rat have occasionally been observed in the wild). But Pandas rarely eat meat has the approximate negator rarely, and claims only that cases of carnivorousness are statistically rare in Ailuropoda melanoleuca, which is true. That it is a negative clause is once again confirmed by the tag-question and not even tests:

(j) *Pandas rarely eat meat, don't they?
(k) Pandas rarely eat meat, not even if offered it.

One of the seven words that effect approximate negation is the determinative few. Unlike a few, which is decisively positive, few negates any clause in which it appears on the subject NP,

(l) Few pandas eat meat.
(m) *Few pandas eat meat, don't they?
(n) Few pandas eat meat, not even if offered.

The third complexity is that nonverbal negation works best if the negating word is early in the sentence. It is most acceptable in the subject slot, and acceptability is lower when it is in the predicate. That is, Nobody had my permission to take this action is much more clearly perceived as a negative clause than I gave permission to do this action to nobody.

The Taiwanese high school exam question unfortunately has an approximate negator in the direct object combined with a rather weak diagnostic for clause negation. I have to admit that I find it very hard to be sure about my judgment that (a) is ungrammatical and (b) is grammatical:

(a) Lydia knows few things, and so does Peter.
(b) Lydia knows few things, and neither does Peter.

I definitely don't think detecting the putative grammaticality in such a case makes a good exam question for foreign learners.

But the worst thing is this. Suppose the survey conducted by my friend in Taiwan were substantiated by a wider investigation — that is, suppose BrE speakers agree on the grammaticality of (b) while AmE speakers disagree and regard (a) as grammatical.

The acquisition puzzle in that case is what worries me the most. How could the facts possibly be learned by the two communities? This is not something that can be attributed to universal grammatical constraints genetically built into the human mind: the facts are different for two geographically separate but not genetically isolated communities of humans.

Is the frequency of hearing "so does…" after clauses with approximate negators so much higher in America? Or is nonverbal negation with approximate negators] so much less frequent in AmE that Americans develop a grammar that fails to define them as negative?

I would actually prefer to think that my friend's informal survey was a victim of small numbers and unreliable judgments, and there is no dialect split. Because if it is a solid fact that the Atlantic Ocean divides the people who accept (a) from the people who accept (b), syntacticians have some explaining to do.


  1. Viseguy said,

    January 28, 2018 @ 6:17 pm

    AmE here (NYC born and bred). At first blush both (a) and (b) seemed incorrect or somehow off to me, but on further consideration (a) is (marginally) acceptable while (b) is not. For some reason, (a) becomes more clearly acceptable if I change it to "Lydia knows little, and so does Peter." Conversely, I'm totally comfortable with "Lydia doesn't know much, and neither does Peter." In the final analysis, I have a hard time thinking of "Lydia knows few things|little" as a negation. It may be derogatory, but it's still affirmative to my ear.

  2. TIC said,

    January 28, 2018 @ 6:38 pm

    I'm AmE born and bred and, for whatever it's worth, a) sounds right to me. Even more so with the equivalent phrasing "as does Peter."

  3. Rachael said,

    January 28, 2018 @ 6:40 pm

    I'm British and I'd pick (a). Although both sound stilted to me – I'd never say "Lydia knows few things". "Few" (as opposed to "a few") is rare in my idiolect; I'd more naturally say "Lydia doesn't know many things".

    Also, your examples (e) and (f) say "Lydia doesn't knows anything". I was confused by this because (e) is starred as ungrammatical and I thought at first that this was why.

  4. I know few things, too/*either said,

    January 28, 2018 @ 6:40 pm

    I'm an AmE speaker and have the same judgment about the Lydia/Peter sentences (a is better than b) — but, perhaps interestingly, "Pandas rarely eat meat, do they?" and "Pandas rarely eat meat, don't they?" both sound OK to me (on readings other than the "she does, does she?" type).

    A: Fluffy didn't eat any meat yesterday. And he didn't eat any meat today, either!
    B: Well, pandas rarely eat meat, do they?

    A: Here's my list of animals that rarely eat meat. Can you think of any more?
    B: Pandas rarely eat meat, don't they?

    I wonder if it's due to the difference between "pandas RARELY eat meat" and "PANDAS rarely eat meat".

    In fact, I think focus can do a lot to go against the negative polarity patterns:

    A: I need a naïve speaker to test the example sentences for my new experiment on! Is there anyone around here who knows absolutely nothing about linguistics?
    B: Well, LYDIA knows absolutely nothing about linguistics, doesn't she? (*does she)

    It's possible that the intonation of the tag question can also play a role. But, going back to Lydia/Peter, is it possible that North American and non-North American English have different readings available (or salient) for bare 'few' that would imply different focus?

  5. Rachael said,

    January 28, 2018 @ 6:44 pm

    To pick a comparable example that doesn't sound stilted to me, "Lydia says little, and so does Peter." Even though there's approximate negation, I wouldn't say "and neither does Peter". Also, with "pandas rarely eat meat", I think I prefer "and so do chickens" rather than "and neither do chickens".

  6. JohnG said,

    January 28, 2018 @ 7:04 pm

    AmE here, but (b) sounds better to my ears and seems more in line with the intended meaning (that Lydia generally lacks knowledge). We're excluding Lydia from the class of people who know many things (assuming knowing more is better than knowing less) and we're similarly excluding Peter.

    If I were writing it, I'd probably notice that it's a bit awkward and look for an alternative.

    Another variant I find interesting and flips the sense is:

    (z) Lydia knows a few things, and so does Peter.

    Knowing "a few things" seems to imply that, while spotty, the knowledge Lydia has is important and the sentence is now a positive statement about her and Peter.

  7. Tim Leonard said,

    January 28, 2018 @ 7:11 pm

    I'm struck by the assumption that a construction is either grammatical or ungrammatical. Surely there's a spectrum, and the grammaticality of some constructs is not clear. If you ask a copy editor, "When I start a sentence with the name e. e. cummings, do I capitalize the first letter or not?" the answer will almost certainly be, "Restructure the sentence to avoid the issue." Similarly, when I'm not sure of the grammaticality of a construct, I find a way of expressing myself that doesn't use it. I'm sure other people do too. So a better answer (which the test probably doesn't allow) is that both (a) and (b) should be rewritten.

    Do modern linguists take this position, or try to claim that grammaticality is binary?

  8. Viseguy said,

    January 28, 2018 @ 7:11 pm

    I don't immediately get the "spotty knowledge" implication from "Lydia knows a few things, …", although I concede it's possible. More emphatically positive would be, "Lydia knows a thing or two," or especially, "Lydia knows a thing or two about [blank]."

  9. David L said,

    January 28, 2018 @ 7:29 pm

    They both sound wrong to me. The first would work if you put 'a' before 'few':

    Lydia knows a few things, and so does Peter.

    But then of course the sense is different: the statement says that Lydia does know some things, not that she doesn't know a whole lot.

    [BrE turned AmE]

  10. John From Cincinnati said,

    January 28, 2018 @ 7:42 pm

    Mephistopheles: Why this is hell, nor am I out of it.
    –Christopher Marlowe, circa 1593

    AmE here. The above memorable line from my youthful studies completely colors my reaction to the question. Answer (b). I also like "and nor does Peter."

    I appreciated GKP's discusssion of few / a few. The only thing that would make me comfortable with the grammaticality of (a) is if Lydia knows *a few things; but as David L said, the sense is different.

  11. John Roth said,

    January 28, 2018 @ 7:44 pm

    AmE speaker. My first look says that A is, at best, awkward and probably wrong, while B is perfectly acceptable.

  12. Laura Morland said,

    January 28, 2018 @ 8:35 pm

    I echo John Roth completely (fellow AmE speaker).

  13. Jim Breen said,

    January 28, 2018 @ 8:52 pm

    Australian, i.e. BrE speaker. I'm only comfortable with (b).
    I checked with my wife (as one does). She's much further into
    grammatical issues than I am, and was quite adamant that (b) is
    the correct one. She remarked that she's had tussles in editorial
    situations with AmE speakers about use of "neither" without an
    accompanying "nor". For us it's OK and for them it was a no-no.

  14. Jon Galliazzo said,

    January 28, 2018 @ 9:33 pm

    I think that the ambiguity or weakness of the sentence construction makes choosing one or the other as grammatical or not isn't actually possible. I think these are both probably strictly grammatically correct, although an argument could be made that the sentence structure and lack of more definitive signals makes these de facto ungrammatical in the sense that the specific words can be used with each other but a sentence can't be structured the way both have been. Either way, I would argue that most native English speakers would consider both of these sentences poorly written and would be highly unlikely to ever say things this way.

    Maybe I'm a bit more on the fence while also considering that they may both be te'hnically grammatical because I am Canadian and so influenced by both British and American English usage.

  15. Chas Belov said,

    January 28, 2018 @ 9:50 pm

    Southwestern Pennsylvania English speaker here.

    (a) *Lydia knows few things, and so does Peter.
    (b) *Lydia knows few things, and neither does Peter


    (a) Lydia knows a few things, and so does Peter.
    (a') Lydia and Peter know few things.

    would be correct. Darned if I can say why.

  16. Chas Belov said,

    January 28, 2018 @ 9:56 pm


    Lydia knows few things, doesn't she?
    *Lydia knows few things, does she?

    BTW, is "One is particularly well known: positive sentences take negative confirmation tags (isn't it?) while negative sentences take positive confirmation (isn't it?)." a typo? I would have expected:

    One is particularly well known: positive sentences take negative confirmation tags (isn't it?) while negative sentences take positive confirmation (is it?).

    [Yes, there was a typo overnight; I've fixed it. Thanx. —GKP]

    but would take the tendency to add such tags to be overwhelmingly BrE.

  17. Ellen Kozisek said,

    January 28, 2018 @ 10:08 pm

    I'm American, and if forced to pick I'd pick B, but really both sound wrong. Not so much clearly ungrammatical as just something one wouldn't say. The first part, "Lydia knows few things" is not something I would say. Not how I would express the idea. And not something I'd expect others to say.

  18. David Morris said,

    January 28, 2018 @ 10:43 pm

    Australian, usually closer to BrEng, but in this case, it's a) (but only just – it's very awkward).

  19. NatShockley said,

    January 28, 2018 @ 11:41 pm

    Another Australian speaker, in full agreement with David Morris and even more so with Viseguy's first comment: both (a) and (b) feel wrong but (a) is perhaps marginally acceptable, whereas (b) feels completely wrong. Something in my brain screams with reflexive pain on encountering that "neither" there.

  20. NatShockley said,

    January 28, 2018 @ 11:49 pm

    The overall impression I get from the comments is that no native speaker would be likely to say either of these things. You'd try to express yourself in some other way. I found Tim Leonard's comment very interesting in this regard: is there a technical term for these areas of grammatical unclarity that people try to avoid?

  21. Rebecca said,

    January 28, 2018 @ 11:55 pm

    AmE speaker here. For me, (a) sounds off, but (b) sounds fine. I was very surprised to read there was a dialect difference (and I seem to fall on the wrong side of the line)

  22. Rebecca said,

    January 28, 2018 @ 11:59 pm

    After writing the above, I recalled that, as a "positive anymore" speaker, I often am surprised by other people's negative polarity judgements. I wonder if that will factor into this.

  23. Joyce Melton said,

    January 29, 2018 @ 12:04 am

    I'm actually a professional book editor. Both sentences fail the grammatical test I use: do I have to read them more than once to get a clear meaning?

  24. Martha said,

    January 29, 2018 @ 12:17 am

    I'm American and I thought (a) was fine and (b) was ungrammatical. But I agree with others why say that despite (a) being fine grammatically, it isn't something someone would say.

    My husband (also American) didn't feel comfortable with either and when I asked him which was better, he said they both were "wrong."

    Which, in my opinion, doesn't make for a useful grammar test. When my fellow (ESOL) teachers and I get together to analyze each others exams, this is the exact sort of thing we would tell each other to rewrite or get rid of.

  25. Rubrick said,

    January 29, 2018 @ 12:36 am

    To my (AmE) ear, both of these variants, marked as grammatical by Geoff, sound off:

    (k) Pandas rarely eat meat, not even if offered it.
    (n) Few pandas eat meat, not even if offered.

    In both cases, I strongly want to drop the "not": "Pandas rarely eat meat, even if offered it."

  26. tangent said,

    January 29, 2018 @ 1:59 am

    To my ear (AmE) they're both wrong. If pressed I'd say (a) is technically grammatical but only by abuse of the fact that the form is positive — I'd only say it to make a grammatical point.

  27. tangent said,

    January 29, 2018 @ 2:01 am

    What about these, though:

    (1) Alice said nothing, and so did Bob.
    (2) Alice said nothing, and neither did Bob.

    I actually think they're both valid, with different senses. So maybe my grammatical judgments are getting mixed up with my taste.

  28. cliff arroyo said,

    January 29, 2018 @ 2:38 am

    AME here and I'm surprised that hardly anyone is mentioning intonation and stress and context.

    a) is okay if 'few' has primary stress (Lydia and things have secondary roughly equivalent stress) in the first clause and Peter is stressed a bit more strongly than usual. this implies a context where the topic of discussion is how well informed people are

    Lydia knows _few_ things (ie is poorly informed) and so does Peter (who is similarly poorly informed).
    It also implies that they've been kept ignorant by the speaker or someone known to the speaker….

    b) no intonation I can think of makes this okay grammatically although the meaning is similar, it's maybe marginally okay with a longer than usual break between the two clauses (giving time for the meaning that Lydia doesn't know much to soak in)

  29. rosie said,

    January 29, 2018 @ 2:44 am

    For me (BrE), "neither" is ungrammatical with a mere approximate negator. If Peter knows few things, then
    (b) *Lydia knows few things, and neither does Peter.
    fails because "neither" implies complete negation (like "nothing"). If Peter knows nothing, then
    (b) *Lydia knows few things, and neither does Peter.
    fails because "neither" implies equality with an antecedent (like "so").

  30. Jamie said,

    January 29, 2018 @ 2:45 am

    To me (BrE) they both sound unnatural. Changing "things" to "people" makes it slightly less awkward (for some reason) and I then dislike (b) less than I dislike (a).

    Interesting that some commenters have changed "few" to "a few", deliberately or otherwise.

  31. stedak said,

    January 29, 2018 @ 3:02 am

    American here: both (a) and (b) are bad. That is, (b) is completely impossible, while (a) is something I would never say but might grudgingly accept if someone else said it.

    Also, I agree with Rubrick above about the "not even if" sentences: the "not" should be dropped. I would only use "not" if it's an absolute negative and strongly emphasized by breaking up the sentence: "Pandas never eat meat at all. Not even if offered it!"

    But mainly, I'm posting to ask you to please fix the typos
    Lydia doesn't knows anything (4 times!) and positive confirmation (isn't it?)

  32. Martin said,

    January 29, 2018 @ 3:06 am

    BrE here: definitely prefer (a) to (b), for reasons similar to rosie. I think the original idea that it's a plain BrE/AmE difference can be laid to rest.

    (The article would be improved if the repeated "doesn't knows"s were fixed.)

  33. John Walden said,

    January 29, 2018 @ 3:11 am

    Because question tags which are negative when the affirmation is positive and vice versa:

    John speaks English, doesn't he?

    as well as echo tags which follow the same "polarity":

    John speaks English, does he?

    both exist, I'd say a more reliable test could be how to deny the truth of the affirmation:

    'John hardly ever listens' 'No, he doesn't' is agreement

    'John hardly ever listens' 'Yes he does' is disagreement.

    But 'hardly ever' is a paid-up member of the 'treated-as-negative' club

    On to 'Lydia knows few things' which I suspect is different.

    'She does' vs 'She doesn't' . For me, on balance I think the former agrees and the latter disagrees. But I wouldn't say or expect it in the first place.

  34. Hans Adler said,

    January 29, 2018 @ 4:15 am

    The results are probably tainted by the fact that the distinction between *few* and *a few* is clearly on its way out. I know this from the Duolingo Dutch course for English speakers. A large number of native English speakers in that course are sure that only one of the two expressions is grammatical. (Actually the main dispute – now mostly deleted – was about *little beer* vs. *a little beer*, but it seems to be a general phenomenon.) Apparently, the indefinite article in "a few" has become so inaudible in a lot of ordinary speech that no longer all native speakers are getting enough input to learn the distinction. Most of those affected seem to believe that *few* and *a few* both mean what is still *a few* in standard English. They split into those who think *few* is colloquial for *a few*, those who think *a few* is (borderline) ungrammatical due to an extra indefinite article, and those who think it's fine either way.

    My impression was that AE speakers were affected by this more than BE speakers. If true, AE speakers are likely to actually 'see' the following options:

    (a) Lydia knows a few things, and so does Peter.
    (b) Lydia knows a few things, and neither does Peter.

    Given this choice, clearly only (a) is grammatical.

    (Also, to make earlier comments more explicit: There are 4 instances of the typo "doesn't knows" in the post. As these are in sentences whose grammaticality is under discussion, they really should be fixed.)

  35. Miles said,

    January 29, 2018 @ 4:57 am

    British speaker here.

    Having read all the comments, I've now lost a sense of whether either is acceptable, partly because (as others have suggested) both feel a bit unnatural.

    However, following on from tangent's comments, I would be comfortable with
    (1) Alice said nothing, and neither did Bob

    but not
    (2) *Alice said nothing, and so did Bob.
    (except perhaps in a deliberately humorous usage).

  36. Andreas Johansson said,

    January 29, 2018 @ 5:02 am

    L2 speaker, usually closer to BrE than AmE (I think). FWIW, my immediate reaction was that both were odd but only (b) is wrong.

    But we seem to have a fairly strong consensus that whether grammatically legitimate or not, both are unidiomatic?

  37. Bart said,

    January 29, 2018 @ 6:02 am

    Here I have a difficulty that arises with many similar threads. People discuss whether a given piece of language (eg 'Lydia knows few things, and neither does Peter.') is grammatical or is not grammatical.

    Plainly you can’t do that without having a clear idea of what you mean by saying of any particular piece of language (whatever it may be) that it is grammatical or is not grammatical.

    Everybody on this thread so far seems to be adopting a certain meaning of ‘grammatical’ that is never made explicit, as if we all understood exactly what it meant.

    But if you did define explicitly what you meant by ‘grammatical’, wouldn’t that be a very useful first step towards deciding whether any particular tricky item such as this one did or did not meet the criterion?

    As I say, the same question arises on numerous other threads about the grammaticality of specific pieces of language.

    Am I the only one troubled by this?

    Is everybody who enters the discussion in fact using the same definition of ‘grammatical’ or are there important discrepancies?

    If everybody is in fact using some fairly straightforward, generally accepted definition of ‘grammatical’, where can I find it written down?

  38. poftim said,

    January 29, 2018 @ 6:11 am

    I might as well chime in (BrE).
    (a) feels grammatical but decidedly weird to me, while (b) doesn't feel grammatical at all.

  39. David Marjanović said,

    January 29, 2018 @ 6:14 am

    John Walden wrote at what's shown to me at 3:11 am:

    But 'hardly ever' is a paid-up member of the 'treated-as-negative' club

    And there, I think, we have it. I think the issue in the OP isn't so much an issue of syntax as an issue of agreement with meaning vs. agreement with grammatical form. Generally speaking, BrE does more of the former than AmE, but most or all Englishes do it more often than, say, German: the police is seems to be flat-out ungrammatical in BrE, because the police are many people, but it's a common option (though apparently not the only one) in AmE, while in German "the police are" is wholly unthinkable.

    I agree, of course, that L2 learners shouldn't even be exposed to such unidiomatic example sentences as used in that exam question.

  40. David Marjanović said,

    January 29, 2018 @ 6:17 am

    …Specifically, I'm claiming that knows few things is grammatically affirmative, but semantically negative enough for some people to give it negative agreement (and neither)

  41. Terror Incognita said,

    January 29, 2018 @ 6:33 am

    Just to throw my hat into the ring as a BrE speaker: for me (b) is grammatical, (a) is not. 'Few' in this case is (to me) very clearly a negative marker that's used to emphasise how little Lydia knows. This is in stark contrast to 'a few' which would always be positive to me… which leads me to think that Hans Adler is onto something in a comment a few above, and there may be some amount of evolution in various idiolects in how 'few' and 'a few' (or 'little' and 'a little') are used. It appears that it's definitely not a BrE/AmE distinction though, as shown by the comments here.

  42. RP said,

    January 29, 2018 @ 7:06 am

    BrE. I prefer "Lydia knows few things, and so does Peter", although my opinion has changed back and forth a few times in the course of thinking it over.

    However, if the conjunction were "but", I'd prefer "Lydia knows few things, but neither does Peter".

    On the other hand, if "few" were the subject, then I would definitely prefer "Few people know Lydia, and neither does Peter" – not "few people know Lydia, and so does Peter".

  43. Lane said,

    January 29, 2018 @ 7:35 am

    AmE. The first is barely acceptable, the second not at all.

  44. /df said,

    January 29, 2018 @ 8:10 am

    BrE Neither sounds natural but (b) wins.

    But how about these:

    (a')*Lydia wears few clothes, and so does Peter.
    (b') Lydia wears few clothes, and neither does Peter.
    (c')*Lydia doesn't wear much, and so does Peter.
    (d') Lydia doesn't wear much, and neither does Peter.
    (e') Expecting another scorching day, Lydia wears few clothes, and so does Peter.
    (f') Expecting another scorching day, Lydia doesn't wear much, and neither does Peter.

    For me (a') is wrong, but not (e').
    Seemingly in (e'), the 'expecting' makes the 'few' positive, whereas its default interpretation is negative. But a lack of something negative doesn't work:

    (g')*Lydia's genius has few limits, and so does Peter's.
    (h') Lydia's genius has few limits, and neither does Peter.

  45. John Finkbiner said,

    January 29, 2018 @ 8:17 am

    I find the sentences seem less weird with “people” substituted for “things” — with that substitution (a) seems much better to me. (I have spent my entire life in NA.)

  46. Coby Lubliner said,

    January 29, 2018 @ 9:43 am

    I think it's unfortunate that this discussion is framed in terms of a kind of Chomskyan notion of "grammaticality". To me a far more useful notion is idiomaticity, and it seems clear that neither sentence is idiomatic in any variety of English.

    [It's quite true that the exam question insists you decide between two sentences neither of which sounds quite right. That is my main reason for not liking it at all as an exam question. It is, however, amazingly common for English exams in non-English-speaking countries to use this type of easily-graded but pedagogically inadvisable question. I keep getting emails from people in countries like Iran, Saudi Arabia, Israel, India, Pakistan, Bangladesh, China, Korea, and Japan asking for my opinion about the correct answer to questions like this. I wrote about the issue twice on Lingua Franca: here and here. I want to simply say, don't use this type of question. But of course some of the people writing to me are students, who are required to answer such questions! —GKP]

  47. Mark Meckes said,

    January 29, 2018 @ 10:49 am

    To my American ears neither (a) nor (b) sounds idiomatic, though (b) sounds slightly less strange. So I'm one weak data point against a transatlantic dialect split here.

    But if such a split is real, here's some poorly-informed speculation about a possible source. The following assertions are based only on my own intuition, so take with an appropriately large grain of salt: The set phrase "not a few" is unambiguously positive, and is more common in BrE than AmE. This tends to reinforce negative polarity for "few".

    (Wait, is it really true that both "a few" and "not a few" are always, or at least usually, positive? Seems right to me, anyway, but amusing.)

  48. Rose Eneri said,

    January 29, 2018 @ 11:16 am

    As an AmE speaker, I find both a and b unacceptable mainly because I find "X knows few things" itself to be ungrammatical. A native speaker would never say it (except under the most exact circumstances), so how can anything follow it? But on purely hypothetical grounds, I would say that "X knows few things" is positive and must be followed by option a "and so does Y." Honestly, I can't even parse b.

    For more grammatical sentences,"X knows a few things" would be followed by "and so does Y." But, I also think "X knows so few things" still would be followed by "and so does Y." But, a native speaker most likely would follow "X knows so few things" with something more like "and the same could be said for Y" which still is positive.

  49. Cary Coutant said,

    January 29, 2018 @ 11:42 am

    Wow, the number of people who prefer (a) over (b) completely caught me off guard. I'm an AmE speaker who finds (a) wrong and (b) natural. The distinction between "knows a few things" as a positive statement and "knows few things" as a negative statement is clear to me and ingrained in my use of English.

    In fact, I had this come up very recently while copy editing an article. The original sentence was "There are few customs or observances for [x]," and the suggested correction was to change it to "a few." But to my mind, that would have changed the intended meaning of the sentence from emphasizing that there were only a few customs or observances to one that simply points out that, yes, there are some (which is less interesting). And by changing it to "a few", the conjunction is wrong; it should be: "There are a few customs _and_ observances…."

    Has the language really lost this subtle distinction?

  50. stedak said,

    January 29, 2018 @ 12:20 pm

    I agree with Rose that "Lydia knows few things" by itself is archaic and stilted, and no one would say it although they might write it. But we can sidestep this, and the question of "few" vs. "a few", by substituting another approximate negative. They all have the same problem for me (American):

    *Lydia rarely eats meat, and so does Peter. (The scope ambiguity of "so" hits me in the face: Peter does eat meat, or Peter rarely eats meat?)
    *Lydia rarely eats meat, and neither does Peter. (I don't like "neither" with approximates. It should be absolute.)
    Lydia eats meat less than once a month, and so does Peter. (OK)

  51. RP said,

    January 29, 2018 @ 1:06 pm

    @Cary Coutant,
    If the negative polarity of "little" is no longer observed (or no longer universally observed) in certain contexts, that isn't the same thing as saying that the meaning has changed in any substantive way (after all, "Lydia knows an extremely small number of things" is an unambiguously positive statement from a syntactic viewpoint, and yet is synonymous with "Lydia knows few things"). Even if that change in polarity (at least in certain contexts) is sufficient to constitute a substantive change of meaning, that is still a long way from the distinction in meaning between "few" and "a few" necessarily being lost.

  52. Jon W said,

    January 29, 2018 @ 1:57 pm

    AmE. I suspect that, for me, the reason that (a) seems merely troubling, and (b) seems screamingly wrong, is that "knows few things" strikes me as a sufficiently awkward construct that my mind stops to process it as the positive "is ignorant" before proceeding to the rest of the sentence. FWIW, I respond even more strongly to "Lydia speaks few words" as a positive statement.

  53. Mr Punch said,

    January 29, 2018 @ 2:17 pm

    I find that I (AmE) am an absolutist on approximate negation: "Lydia knows nothing, and neither does Peter" is fine, but "Lydia knows next to nothing, and neither does Peter" seems incorrect.

  54. FM said,

    January 29, 2018 @ 2:27 pm

    AmE, and agreed with TIC. (b) is ungrammatical, (a) is questionable, and I prefer ", as does Peter."

  55. BZ said,

    January 29, 2018 @ 2:33 pm

    AmE but originally from Russia:
    "Lydia knows few things" sounds barely grammatical even on its own, and completely ungrammatical with either (a) or (b). In my idiolect (I think) a "neither does" requires some form of the word "do" in the antecedent, so "neither does Peter" requires "Lydia doesn't know anything" or some such, so to me, "neither does" is not a negative polarity test at all.

    Similarly "Pandas rarely eat meat, not even if offered it." sounds wrong to me because to me a "not even" construction requires an absolute negative ("pandas don't eat meat" or "never eat meat"), so clearly "Lydia knows few things" cannot take a "not even".

  56. Carol Saller said,

    January 29, 2018 @ 2:43 pm

    AmE, and I thought both sounded confusing!

  57. RP said,

    January 29, 2018 @ 3:33 pm

    Gunnel Tottie (Fuzzy Negation in English and Swedish, 1977) asked respondents to fill in the missing word in the sentences "Few people like him. ___ do I" and "I see few people these days. ___ does Mary".

    In the first case ("few" as subject), 96% of respondents specified either "neither" or "nor".

    However, in the second case ("few" as object), only 72% answered with "neither"/"nor" (the remainder answering "so"; non-compliant responses are excluded).

  58. Jerry Friedman said,

    January 29, 2018 @ 5:35 pm

    AmE. Like some others, I find (a) weird and (b) definitely wrong. What I'd actually say is something like "Lydia knows very little about this. Peter doesn't know much about it either." Or "Lydia knows very little about this—much like Peter."

    Cary Coutant: I'm familiar with the distinction between "few" and "a few". The former is more common with "very few". My difference from you is syntactic: "few" isn't a negative-polarity trigger for me.

  59. maxh said,

    January 29, 2018 @ 6:01 pm

    AmE: I'd consider B more proper; A sounds like an attempt at being clever. I don't really like either, though.

  60. RfP said,

    January 29, 2018 @ 6:13 pm


    1. They’re both wrong.
    2. B is slightly less wrong than A.

  61. Martha said,

    January 29, 2018 @ 7:08 pm

    I just wanted to disagree with the comment above that "police is" is a "common" option in American English. I'm not saying it doesn't exist; I'm saying that most people don't say it. "Police" is plural.

  62. Adrian said,

    January 29, 2018 @ 9:37 pm

    I know from my spell teaching English overseas that the disctinction between "few" and "a few" is one of the things that is taught, but in common with other things that are taught, the importance of these words, or of the distinction between them, in the language is overplayed. As others have pointed out, bare "few" and "little" have become relatively rare in general speech, to the extent that many native speakers think they are mistakes for "a few" and "a little".

    Anyway, despite the fact that I do know the difference, (a) still makes some kind of sense and (b) is alien to me, so I think it's fair to say that most BrEng speakers would get this question "wrong".

  63. Mark S said,

    January 29, 2018 @ 9:58 pm

    BrE speaker, but I live in US:

    (b) sounds stilted, but possible, (a) sounds impossible.

  64. Isaac Grosof said,

    January 30, 2018 @ 2:11 am

    AmE speaker:

    (a) sounds mildly ungrammatical, (b) sounds more clearly ungrammatical.

    For me, a significant part of the issue is the construction "knows few things", which feels borderline unacceptable on its own. A more standard usage of "few" in this way would be "This tree has few apples on it." Substituting this into the sentences, we get:

    (a') "This tree has few apples on it, and so does that one."
    (b') *"This tree has few apples on it, and neither does that one."

    Now, a' sounds clearly acceptable to me, while b' is still no good.

  65. Schorsch said,

    January 30, 2018 @ 8:23 am

    Adrian said "As others have pointed out, bare "few" and "little" have become relatively rare in general speech, to the extent that many native speakers think they are mistakes for "a few" and "a little"."

    As this was an exam question for students in Taiwan, perhaps it really is simply a case of a mistake for 'a few'. Any TESOL teacher who has taught Chinese students would be familiar with such missing articles.

  66. EndlessWaves said,

    January 30, 2018 @ 9:11 am

    It does seem like the few/a few distinction is a good candidate.

    I wonder how much it's being affected by the s of knows. If we picked a sentence like:

    a) Lydia can jump few hurdles, and so can Peter.
    b) Lydia can jump few hurdles, and neither can Peter.

    would the proportion of a/b responses change? Would anyone go for a?

    As a BrE english speaker few is emphasising how little knowledge she has. Examples using few that spring to mind as things like 'We few, we happy few' and 'Far and few, far and few'. If pressed I'd go for b) for knows few things.

  67. AntC said,

    January 30, 2018 @ 9:13 am

    Thanks Geoff It's quite true that the exam question insists you decide between two sentences neither of which sounds quite right. That is my main reason for not liking it at all as an exam question.

    I can only pity the poor students faced with these questions. As if English isn't already hard enough to learn!

    Lydia knows few things. just doesn't sound quite right; and none of the proffered continuations make it sound righter. It's the vagueness of "things" that for me doesn't allow of polarity or degree: "Lydia knows few answers" is fine; then I'd prefer continuation (b). (Many of the commenters on LinguaFranca mis-read as "Lydia knows a few things, …" and I agree that's more likely.)

    I'm BrE living in NZ FWIW.

  68. Coby Lubliner said,

    January 30, 2018 @ 10:19 am

    I can see something like "Lydia knows few things, but they are important" as a positive statement, while "Lydia knows few things, and understands even fewer" as a (perhaps politely) negative one. But the clause by itself, out of context, is meaningless.

  69. DWalker07 said,

    January 31, 2018 @ 12:30 pm

    "That seems like a solid ungrammaticality, confirming that Lydia doesn't knows anything is negative. "

    What the heck is "Lydia doesn't knows anything"? That phrase appears several times in the original post. Is that deliberate?

  70. ohwilleke said,

    January 31, 2018 @ 11:54 pm

    The thing that makes (a) v. (b) difficult in my mind, is that "few" can be positive or negative.

    In (a), "few" is positive, Lydia and Peter know a few things that make it possible for them to get by, they may not know many things but the things that they know are important. The familiarity of the few important things trope makes this a plausible reading.

    In (b), "few" is negative, Lydia and Peter should know many things but actually know only a few, for example, out of the many test questions. Also a possible reading.

    Few is too ambivalent between positive and negative for it to illustrate agreement between clauses on positivity/negativity grounds.

  71. Keith said,

    February 1, 2018 @ 5:18 am

    I disagree with Hans Adler on the notion that the distinction between "few" and "a few" is disappearing.

    Reading these two examples,
    (l) Few pandas eat meat.
    (m) *Few pandas eat meat, don't they?

    I immediately thought of
    A few pandas eat meat, don't they?

    And there's still a typo:
    That seems like a solid ungrammaticality, confirming that Lydia doesn't knows anything is negative.

  72. Scott P. said,

    February 1, 2018 @ 10:44 am

    Both are stilted, but B sounds better.

    To my mind, the sentence is equivalent to "Lydia doesn't know much, and _______ does Peter."

  73. Jason Merchant said,

    February 1, 2018 @ 4:14 pm

    This American speaker, as well as Edgar Eager, the beloved American children's author, allow "few" and "little" in the split reading of (b). I gave examples and analyzed them in a paper published in 2013, e.g.:

    John has few friends, and frankly, his brother doesn’t really, either. .

    The paper built on Potts 2002 analysis of split negation, and there's a fuller investigation of these and related facts in Tanja Temmerman's 2012 Leiden dissertation ("Multidominance, ellipsis, and quantifier scope"). Many puzzles remain!

  74. Marc-Andre Pelletier said,

    February 1, 2018 @ 5:44 pm

    Interestingly enough, both (a) and (b) sound ungrammatical to my ESL ears; yet become completely clear if you add:

    (a) *Lydia knows few things about football, and so does Peter.
    (b) Lydia knows few things about football, and neither does Peter.

    I think the initial confusing is simply that "X knows few things" without a complement isn't idiomatic, and the brain tries to latch to "X knows a few things" which *is*, and is positive.

  75. stedak said,

    February 1, 2018 @ 9:09 pm

    I don't agree that "about football" makes any difference. To me (AmE) "Lydia knows few things about football, and neither does Peter" is still bad.

  76. Graeme said,

    February 2, 2018 @ 4:38 am

    AustE here. Both sound odd. I'd be more likely to say (a).

    But I read (a) ambiguously. It suggests to me that Lydia and Peter are not just coincidentally ignorant: what they do know may be in cahoots. Whereas (b), clunkier though it is, is clearly just a descriptive statement of their lack of knowledge.

  77. Mike said,

    February 2, 2018 @ 9:58 am

    (Southern) AmE speaker here. Though both 'a' and 'b' sound odd, 'b' sounds better to me. If it were "a few" then 'a' would sound better similar to how "I have little money" sounds negative while "I have a little money" sounds more positive.

  78. skyborne said,

    February 4, 2018 @ 7:26 am

    AmE (WNY), I don’t like either of them, but my first feeling was A. I’d prefer to express it as one of the following:

    Lydia knows few things, like Peter.

    Lydia doesn’t know many things, and neither does Peter.

    There’s something about “knows few things” that makes it a lot weaker than normal, and it doesn’t unambiguously suggest a following structure to me.

  79. Mikey C said,

    February 8, 2018 @ 7:28 am

    Only six examples of “knows few” in COCA, one being “knows few bounds”, which though not commonly used, may be at least familiar. How would a more familiar expression work?

    a) Lydia knows few bounds, and so does Peter.
    b) Lydia knows few bounds, and neither does Peter.

    Would that be less distracting?

RSS feed for comments on this post