*The Haystack's Painting

« previous post | next post »

Appearing in The Paris Review, Winter 2015:

From “A New English Grammar”

Jeff Dolven

*The Haystack’s Painting

The haystack’s painting hangs in the Met;
the painting of the haystack, that is,
the one by Monet, not by van Gogh,

the rose-blue, snow-lit one with the haystack
in it. The haystack has this deal
with many painters, also Millet,

appearing not for a fee, nor a stake,
exactly, but for the sovereign right
to have your eyes back whenever it wants.


*By convention, an asterisk indicates an instance of improper ­usage. All ­titles are drawn from such examples of bad grammar in Rodney Huddleston and Geoffrey K. Pullum, The Cambridge Grammar of the English Language (Cambridge: Cambridge University Press, 2002).

Also *They Resemble Closely, *Either Boy Has a Key, and *An Asleep Child.

For additional linguistic background, see

"A correlate of animacy", 9/27/2008
"The genitive of lifeless things", 10/11/2009
"Mechanisms for gradual language change", 2/9/2014

[h/t Dmitri Tymoczko]



  1. Max said,

    December 9, 2015 @ 6:46 pm

    I'm missing the point of the poem (which is my usual response to poetry), but it's interesting that this linguistic use of an asterisk preceding a word sans space has been picked up, in a mangled form, in online chatting and texting. For example:

    A: Your a wonderfully non-pedantic person.
    B: *You're

    The mangling being that the correction is marked rather than the improper usage, but I wonder if it's a coincidence. I've been seeing asterisks used for disclaimer-like text lately without a referent, which seems like an error to me. "* Odd sizes only" should have a corresponding "All sizes on sale!*" to pair with, as it's fundamentally a footnote.

  2. Ethan said,

    December 9, 2015 @ 7:32 pm

    I would have no problem with a sentence such as "We sat in the haystack's shadow" in any case, but the case at hand is not a generic statement about haystacks. The body of the poem personifies the haystack, so it seems perfectly consistent that the title does also. If the reader is caught short by the title, only to have the tension relaxed by personification in the following lines, this is to the poet's credit.

  3. Vicki said,

    December 9, 2015 @ 8:24 pm

    That has a fantasy/magical realism feeling, with "haystack's" meaning "haystack is": "The haystack's painting its edges" or "The haystack's painting a portrait of the visiting watercolorist."

  4. Rubrick said,

    December 9, 2015 @ 8:41 pm

    I'm not quite sure what's being considered ungrammatical about "The haystack's painting". Haystack is a noun; nouns have possessive forms. It's certainly unusual to consider the subject of a painting to be the "owner" of that painting, but I think it's quite an effective poetic device here.

    I guess I'm missing something?

  5. Ben Zimmer said,

    December 9, 2015 @ 9:32 pm

    I'm reminded of Chomsky's remark about A grief ago, the title and first line of a Dylan Thomas poem: "[A] striking effect is achieved precisely by means of a departure from a grammatical regularity. Given a grammatically deviant utterance, we attempt to impose an interpretation on it." As a further "deviant" illustration, he gave his most famous example sentence, Colorless green ideas sleep furiously — which would become the subject of many poetic ruminations.

    ("Some Methodological Remarks on Generative Grammar," Word 17:2.219-39, Aug. 1961)

  6. Guy said,

    December 9, 2015 @ 9:44 pm


    Sometimes it's suggested that "(genitive) (nominal)" is interchangeable with "the (nominal) of (plain case)", but that's actually generally not true. I believe another example given by CGEL is "the length of silk" vs "*silk's length". There might be some semantically anomalous interpretation of "haystack's painting" in which it is grammatical and even describes the same situation framed differently (as ownership resulting from being the subject of the representation rather than being the subject of the representation), but it's not with respect to that interpretation that the judgment of ungrammaticality is being made.

  7. Guy said,

    December 9, 2015 @ 9:55 pm

    Of course, it should be noted that the semantic signification of a genitive covers many things beyond ownership, it can include many other types of semantic argument such as being a component part or filling a wide range of other "relational" roles (my cousin, his height, the license's expiration date) that go far beyond just ownership, but the semantic roles indicated by of-phrases is a broader class still.

  8. William said,

    December 9, 2015 @ 11:42 pm

    It's clearly genitive: "the painting of the haystack" in the sense of the painting belonging to the haystack. The idea would be that Dolven is taking found example-mistakes and making sense of them. A little like what Robert Olen Butler did in his book of stories Tabloid Dreams. Hence the title of his book (I presume that's the title of the book): "A New English Grammar."

  9. Philip said,

    December 10, 2015 @ 4:20 am

    Could someone who is in New York run up to The Met and see if they have put a sign on the painting with an errant apostrophe s instead of the normal plural s. Haystacks is the title of a series of paintings by Monet of … well, haystacks, actually. The Met may have unintentionally introduced an apostrophe s, and thus unintentionally produced a fairly witty poem about grammar … and haystacks.

  10. Michael Watts said,

    December 10, 2015 @ 4:54 am

    A closely related construction is fully standard. Here's an example, from memory, from an episode of Sabrina the Teenage Witch:

    Friend: Harvey still has a picture of you in his locker!

    Sabrina: He kept my picture?!

    As is made explicit in the context, "my picture" refers to a picture showing me, not a picture that I own. I'd expect "my portrait" to be interpretable in the same way, if I were painted rather than photographed, and in fact I'd also expect "the haystack's portrait" to be accepted as an unexceptional way to describe a picture painted to memorialize the appearance of a haystack. If I used that phrase, I'd be much less surprised by the response "why did you have a haystack's portrait painted in the first place?" than by "huh? How can a haystack own a portrait?"

    I have a hard time saying that "the haystack's painting" is ungrammatical when the 's marker is regularly used to indicate virtually any relationship at all. Would it be better to say that the problem here is actually just the noun painting, and this particular use is precluded by the existing use of "portrait" to refer more specifically to the same thing? If this construction started appearing in The New Yorker for reasons of idiosyncratic house style, how much resistance would it really provoke?

  11. GH said,

    December 10, 2015 @ 5:29 am

    I wouldn't completely rule out that there might be examples where "the haystack's painting" is just odd, rather than outright ungrammatical, but I can't think of any.

    "Portrait", on the other hand, is simply wrong to my ears, because "portrait" only applies to pictures of people (or people-like things), not objects such as haystacks.

  12. Adrian Bailey said,

    December 10, 2015 @ 6:52 am

    "The haystack's painting" is ungrammatical. The correct way of expressing this is "the painting of the haystack". There are many situations in which a form looks grammatical because no "rule" appears to be being broken, but it is in fact ungrammatical because people don't talk/write like that.

  13. Michael Watts said,

    December 10, 2015 @ 8:03 am

    There are also many situations in which a sentence is fully grammatical, but nevertheless strikes people as odd because it's not something a native speaker would say, for reasons of habit.

  14. David Morris said,

    December 10, 2015 @ 8:22 am

    I would not illustrate grammaticality with poetry.

  15. Adrian Morgan said,

    December 10, 2015 @ 8:30 am

    @Max I'd be very surprised if the correction-indicator asterisk had any connection to its use in linguistics. Much more plausibly, it's a development of its traditional use as a footnote marker, indicating that what follows refers to some part of the text preceding it.

    You go on to mention the footnote-marking usage, but without noting the obvious parallels between that and the usage you are curious about.

  16. Theophylact said,

    December 10, 2015 @ 8:54 am

    Dude, it's a poem. The possessive is intentional.

  17. William said,

    December 10, 2015 @ 9:02 am

    @Adrian The note "*By convention, an asterisk indicates an instance of improper usage…" is part of the poem as presented in The Paris Review. I.e. That's Dolven's explanation of what he's doing in the poem. Found phrases turned meaningful.


  18. Adrian Morgan said,

    December 10, 2015 @ 9:11 am

    @William Somehow you misread what I wrote. At no point did I make any reference to the footnote you quote; at no point did I say anything that might give someone the impression I am unaware that it is Dolven's explanation of what he's doing in the poem.

  19. Pete said,

    December 10, 2015 @ 9:31 am


    It's not in general possible to decide if a sentence is grammatical or not by reference to technical terms such as "noun" and "possessive". The only way is to listen to it and ask yourself, "would a native English-speaker ever say this?"

    The rules you find in grammar books about nouns, possessives & so on are just attempts to capture that gut feel.

    Of course this only works if you're a native speaker yourself.

  20. Dan Curtin said,

    December 10, 2015 @ 9:43 am

    Actually I'd say the sign in the Met was ungrammatical, but the poem's title wasn't! Intentions matter!?

  21. Dan Curtin said,

    December 10, 2015 @ 9:47 am

    For the record, I am interpreting the usage of the apostrophe and the context to infer the grammatical intentions of the user. Sorry if this an overly obvious comment. I am a mathematician and at best a linguistic dabbler.

  22. Andrew (not the same one) said,

    December 10, 2015 @ 10:34 am

    There are many things native speakers would not in practice say, because the occasion for them would never arise. It seems odd to call these things 'ungrammatical'.

    If a haystack owned a painting, I think it would be correct to call it 'the haystack's painting'. (The possessive is certainly used for many relations other than ownership, but as far as I can see it can always be used when ownership is in question.) The fact that haystacks do not own paintings is not a linguistic fact. Hence I conclude that 'the haystack's painting' is grammatical.

  23. JS said,

    December 10, 2015 @ 11:03 am

    I find we LLers' dialectical demonstration of exactly the cognitive dissonance the poet intended to invoke insufficiently thorough :P let's go for 50 comments

  24. Dennis Paul Himes said,

    December 10, 2015 @ 11:15 am

    I would probably use an attributive noun there, "the haystack painting".

  25. GH said,

    December 10, 2015 @ 11:22 am

    Is it possible to fully separate out pragmatic and semantic knowledge from grammatical knowledge, though?

    In any case, can't we at least say that it's ungrammatical as a way to express the sense of "the painting of the haystack", as opposed to "the painting owned by the haystack"?

  26. Peter Erwin said,

    December 10, 2015 @ 12:20 pm

    Presumably some light might be shed by someone with a copy of Huddleston & Pullum who can look up the specific context in which "*The haystack's painting" is given as an example of ungrammaticality.

    But one possible use might be to illustrate how some particular grammatical rules have exceptions. E.g., you might imagine a rule for English which allows the transformation:

    The X of the Y. –> The Y's X.

    And then a couple of examples:

    The daughter of the boss. –> The boss's daughter.
    The painting of the haystack. –> *The haystack's painting.

    I can imagine plausible English conversations where the first transformation is perfectly acceptable, if perhaps a trifle stilted — "Is that the daughter of the boss?" "Yes, that's the boss's daughter" — while the second isn't –"Is that the painting of the haystack?" "Yes, that's the haystack's painting."

    In other words, the phrase itself is not universally ungrammatical (as opposed to something like "The man dog bite"), but can be ungrammatical in some contexts.

    (I'd be perfectly happy to accept that there are contexts where "the haystack's painting" could be perfectly grammatical, in a story where haystacks are sentient and have property rights, or where society approves of displaying paintings on haystacks…)

  27. Andrew (not the same one) said,

    December 10, 2015 @ 12:25 pm


    To the second, certainly. 'The X's Y' and 'The Y of the X' are often not equivalent (e.g. 'the enemy's fear' and 'fear of the enemy'), and this is a case of their coming apart.

    To the first; probably not. But in many cases it's clear enough which side something belongs on. Two of Dolven's examples, 'They resemble closely' and 'an asleep child' seem clearly to be ungrammatical, because it's clear enough what they would mean, but they wouldn't actually be used in the relevant circumstance. On the other hand, 'the sky is green' is fairly clearly grammatical, although we would rarely have occasion to use it. It does seem to me that sometimes linguists are too ready to call something ungrammatical when they cannot think of an occasion for it.

  28. William said,

    December 10, 2015 @ 12:50 pm

    @Peter The example is given on p. 477 of Huddleston and Pullum:

    “Certain semantic relations are excluded from the subject-determiner construction but permitted in the of phrase one. There are illustrated in:…

    xiii *the hay stack’s painting || the painting of the hay stack [h is the depiction of d]

  29. William said,

    December 10, 2015 @ 1:48 pm

    @Adrian Morgan — you're right. Apologies: my bad.

  30. DWalker said,

    December 10, 2015 @ 2:01 pm

    @Michael W: "Friend: Harvey still has a picture of you in his locker!

    Sabrina: He kept my picture?!

    As is made explicit in the context, "my picture" refers to a picture showing me, not a picture that I own."

    Of course. If Friend meant a picture that Sabrina owned, he would have said "Harvey still has a picture of yours in his locker".

    That doesn't help the haystack.

  31. Michael Watts said,

    December 10, 2015 @ 3:56 pm

    This is a case of Huddleston and Pullum making a mistake. I'll quote the relevant section more fully:

    We review here a number of factors favouring one or the other; the first involves a structure that excludes the genitive as ungrammatical, while with the others it is a matter of preferences [].

    (a) Semantic relation between the NP and the head noun.

    Certain semantic relations are excluded from the subject-determiner construction [e.g. Mary's sister] but permitted in the of phrase one [e.g. the sister of Mary]. These are illlustrated in:

    i – *Roman coins' collectionthe collection of Roman coins – [h is collection of d]
    ii – *shrub's two kindsthe two kinds of shrub – [h is type of d]
    iii – *red wine's glassthe glass of red wine – [h is quantity of d]
    iv – *gold's colourthe colour of gold – [h is colour of d]
    v – *honour's menthe men of honour – [h has human property d]
    vi – *despair's crythe cry of despair – [h has source d]
    vii – *unemployment's problemthe problem of unemployment – [h has content d]
    viii – *washed silk's dressthe dress of washed silk – [h has composition d]
    ix – *twelve years' girlthe girl of twelve years – [h has age d]
    x – *purple's veilthe veil of purple – [h has colour d]
    xi – *2%'s risethe rise of 2% – [h has size d]
    xii – *the cross's signthe sign of the cross – [h has form d]
    xiii – *the hay stack's paintingthe painting of the hay stack – [h is depiction of d]
    xiv – *all battles' battlethe battle of all battles – [h is supreme example of d]

    There is also no genitive counterpart of the predicative construction the stupid nitwit of a husband [] (cf. *a husband's stupid nitwit), or of partitives like the youngest of the boys (*the boys' youngest).

    (b) Semantic character of the NP

    NPs denoting humans and (to a lesser extent) animals occur with much greater frequency as subject-determiners, e.g. Mary's green eyes and the cat's paw are much preferred to the green eyes of Mary and the paw of the cat []. However, there is also a (less strong) preference for NPs denoting times and geographical entities to be subject-determiners, e.g. October's weather is preferred to the weather of October, and London's pubs is preferred to the pubs of London. Least preferred as subject-determiners are NPs denoting other inanimates, e.g. the roof of the house is preferred to the house's roof [].

    (The brackets around the semantic relations are original; the others involve me cutting out references to various examples in the book or providing examples to define terms that the book defined before the beginning of the quoted section.)

    There are some obvious patterns here, like composition (i, vii, xii) and measurement (iii, ix, xi). The example upthread of the length of silk / *silk's length is of type iii.

    But the semantic relation "h is a depiction of d" obviously doesn't make the genitive subject-determiner construction ungrammatical, because that construction is quite common with the nouns "picture" and "portrait", in exactly the forbidden sense. This has to be a property of the noun "painting" or of the depictee "haystack" in particular.

    (I also don't understand iv [h is color of d]. That construction in that sense seems normal to me. :/ )

    A lot of commenters are voicing strong objections to "the haystack's painting" based on the second factor CGEL lists — haystacks are inanimate and therefore can't participate in the construction "the haystack's portrait" in the same way that a human can, e.g. "my daughter's portrait". But CGEL itself says that only the factor (a) can make the genitive-determiner construction ungrammatical ("with the others it is a matter of preference"), and it is mistaken as to the semantic relation "h is a depiction of d", which poses no problems for the genitive-determiner construction.

  32. Michael Watts said,

    December 10, 2015 @ 4:14 pm

    DWalker: the example contrasts "my picture" with "a picture of me" (they are, in this case, identical). It doesn't contrast "a picture of you" with "a picture of yours".

  33. Mike Grubb said,

    December 10, 2015 @ 4:25 pm

    What about "The Haystack's Landscape"? If, instead of "portrait," we use the more suitable term "landscape," do the same objections continue to apply? Does the polysemy of "landscape" make it more palatable, when the broader term "painting" is less so?

  34. KevinM said,

    December 10, 2015 @ 4:39 pm

    Let's call it the "curator's apostrophe," by analogy to the grocer's (avocado's $2/lb.).

  35. Max said,

    December 10, 2015 @ 5:02 pm

    I'd be very surprised if the correction-indicator asterisk had any connection to its use in linguistics.

    You're probably right, but I (and apparently only I) think it's interesting that the two similarities I mentioned, the lack of referent and lack of intermediate space, match the linguistic usage and not the footnote usage. Not to mention that footnotes are generally not used to mark corrections.

    You go on to mention the footnote-marking usage, but without noting the obvious parallels between that and the usage you are curious about.

    I suppose I assumed that most people would make that connection, but I'll try to be clearer in the future.

  36. Michael Watts said,

    December 10, 2015 @ 5:57 pm

    I can't accept "the haystack's landscape", but that's not very informative since I also can't accept "the landscape of the haystack". I'd have to say "the haystack landscape" (or something equivalent, like "the landscape with the haystack in it"). I guess "landscape" here is supposed to refer to a type of painting rather than a type of scene?

  37. Adrian Bailey said,

    December 10, 2015 @ 8:08 pm

    I find "an asleep child" less ungrammatical. If I said "What did you see in the room?" and the answer was "An asleep child" I'm not sure that I'd be very shocked.

  38. Adrian Morgan said,

    December 10, 2015 @ 11:19 pm


    Of course, interestingness is subjective, and things can be interesting simply as coincidences without in any way suggesting an explanatory hypothesis. But in this case the similarities between the amendment asterisk and the linguists' asterisk strike me as either trivial or contrived, whereas the similarities between the amendment asterisk and the footnote asterisk strike me as a natural development, with plausible intermediate stages between one and the other.

    The footnote asterisk and the amendment asterisk are similar in that they both have a referent, which in the former case is marked explicitly and in the latter case is implicit and left to the reader to identify. In contrast, the linguists' asterisk has no referent at all. A related similarity is that the footnote asterisk and the amendment asterisk both indicate that information is presented in a non-linear order.

    It's worth noting that traditional footnotes do sometimes mark corrections in the special case of corrections to quoted text, but I have no reason to think that's causally relevant.

    An intermediate stage of development would be using the asterisk to indicate a more general class of amendments. For example, suppose someone is typing on a forum in which it is difficult or impossible to edit text once written, and types:

    Language Log is obviously the best blog on the Internet.

    * That should read "best language-related blog". Although it may be the actual best blog as well.

    If you imagine that no-one had ever done this before, an asterisk would be a natural choice of punctuation because, like a footnote, it marks the text that follows as having a referent somewhere in the text preceding it, and alerts the reader to the disrupted linearity. Unlike a footnote the referent is not marked, but that's a consequence of the medium.

    The first typo-correcting asterisks may have been more verbose, for example:

    Spellng correctly is very important.

    * Whoops! I meant "spelling", not "spellng".

    Once such usage becomes common, people will abbreviate it, especially in text messaging. And once it is condensed to the point that the text following the star is typically a single word, the removal of the intermediate space is inevitable.

    I don't know if that's how the amendment asterisk evolved, but I think every step in the hypothesis is plausible. Whereas I dare to submit there's no way to get there from the linguists' asterisk.

    The linguists' asterisk has more in common with other uses of the asterisk that involve setting a subset of items apart. For example, suppose someone gives you a list of possible Christmas gifts and asks you to mark the ones that you would like (or the ones you don't already own). You might well do it with asterisks, indicating that the marked items have a distinguishing property that the unmarked items do not. The linguists' asterisk can be thought of as a special case of that subset asterisk.

  39. Max said,

    December 11, 2015 @ 1:32 am

    The Wikipedia page on the askterisk doesn't shed light on the chat-room origin but mentions a usage I didn't realize was formalized: to correct oneself, the asterisk is placed after the correction, but to correct someone else it's placed before. Either way, no space (unlike footnotes).

    I agree that footnoting seems the most plausible source, but the asterisk rabbit hole is surprisingly deep.

  40. Max said,

    December 11, 2015 @ 2:37 am


  41. Graeme said,

    December 11, 2015 @ 6:36 am

    The Words' Failure.
    (Finish a poem of 3 stanzas under this title, reflecting on why linguistics is a science and poetry an art).

    Without the idealised haystack, no painting.

  42. GH said,

    December 12, 2015 @ 6:11 am

    Regarding the different functions of asterisks to mark footnotes and ungrammatical sentences, it's interesting to note that the poem blends the two uses, using the asterisk in the title both as an anchor for the footnote and to mark the phrase as ungrammatical.

  43. Sally said,

    December 12, 2015 @ 10:15 am

    I'm inclined to agree that "the haystack's painting" is an acceptable way to say "the depiction of the haystack."

    Note that "I took your picture" usually means "I took a picture of you," not "I took a picture of someone else that you own."

  44. William said,

    December 12, 2015 @ 1:04 pm

    @Sally I can't quite get myself to hear "the haystack's painting" and "the haystack's picture" as parallel. I think that's because (for me) painting is a gerund, meaning that in the possessive, it feels as though the possessor has or should have done it. Whereas "picture" or "photo" are just nouns, and I can hear (though a little awkwardly) "the haystack's photo" as okay for the photo of a haystack.

    That can't be the whole story, though, since I can't make this work for "writing" vs. "book" though. Neither "the haystack's writing" not "the haystack's book" works for me; otoh neither does "the writing of the haystack" or "the book of the haystack." Maybe because you write a book but just paint a painting, so in the latter case "painting" contains its object as writing doesn't quite? (You don't write a writing.)

  45. William said,

    December 12, 2015 @ 1:15 pm

    On asterisks: They originally were used to show omissions and elisions. Isidore of Seville said that they showed that all though the words were omitted in the quotation, the original still shone through — hence they are stars.

    Footnote markers, like inverted commas, originated as pointers to the passage in question. IIRC they were originally used to show that a Hellenistic text was quoting Homer, and later that was expanded to the bible. Eventually they were used to show that what was pointed out was extrinsic to the text doing the pointing. Thus they could point to a note in the margin, extrinsic to the flow of the text, just as much as they could point to words from elsewhere. The convention of showing what was not ipsimma verba morphed into the use of italics (a typographical imitation of Petrarch's handwriting, by the way) both for quotation and for clarifying addition — what we would now bracket. The italics in the KJV of the Bible (and others) are translators' interpolations (no, translators aren't themselves interpolated: they do the interpolating). It may well be that the asterisks placed next to defective forms in linguistic analyses and grammar books also suggest that these forms are (so to speak) extrinsic to the canonical use of language. Pointers turned into italics turned back into pointers with the coming in the 18th c. of inverted commas (and similar typographical features, including dashes) to indicate — to point to — quotation. The interplay of italics and inverted commas can still be seen in titling conventions: do you put quotation marks around a title or italicize it? (The title is originally a designation of the work in question: titles as part of the work, rather than simply a reference to it, seem to become more normalized in the 17th c. Poems, for example, really only started getting titles in English, anyhow, around 1600.) Anyhow, I don't know the origin of asterisks as designating defective forms, but I bet there must be some connection.

  46. Jeff Dolven said,

    December 12, 2015 @ 1:16 pm

    The poet here: what an unexpected delight this thread is. "The Haystack's Painting" is part of a series, a few of which are in the Paris Review, a few more forthcoming in the Yale Review. All of them depart from an asterisked phrase or sentence in Huddleston and Pullum.

    I don't think the maker has particular authority over the interpretation, or evaluation, of the made thing, but I'll say a few words about what I was thinking. The Chomsky line above is apropos. Poetry works by one way or another disrupting our accustomed relation to language, and incorrectness is one of the available devices. All I really need, though, is oddness, and I think it is odd enough to say "the haystack's painting" when you mean "the painting that depicts the haystack." Odd enough to provoke some disagreement anyhow! Given that oddness, I have taken my task in the poems to be imagining a world in which the phrase in question is in fact good usage. What would that world have to be like, and what would follow from the people there talking that way? I've found H&P's examples endlessly weird and provocative. As the poems get published, I'll post a notification to the list. In the meantime, thanks to everyone for such close and generous attention. Nothing could make me happier.

RSS feed for comments on this post