Coreference confusion of the week

« previous post | next post »

In stories about Ron DeSantis' expensive PR stunt sending migrants from Texas to Martha's Vinyard back in mid-September, someone named "Perla" played a central role from beginning, described as "a tall, blond woman who spoke to the migrants in broken Spanish".

Recently this person was identified ("Who Is Perla? A Central Figure in Florida’s Migrant Flights Emerges", NYT 10/3/2022), leading to this tweet:

My reaction was "Wait a minute, Perla is a man?"

And I wasn't the only one.

But of course Ron Filipkowski meant "he" to refer to Ron DeSantis, not to Perla Huerta.

That anaphoric relationship is not strictly ungrammatical (in particular, the "binding theory" obviously doesn't apply), but softer constraints are likely to lead a reader (or listener) astray.

I suspect that current automatic coreference resolution programs would also get it wrong, but I don't have time this morning to load one up and try it — perhaps a reader will oblige.


  1. Philip Anderson said,

    October 4, 2022 @ 6:56 am

    “He still won’t talk about or provide any info to the public.”
    I immediately interpreted “he” as DeSantis, but was thrown by the “about” with no object; what is he not talking about?

  2. Trogluddite said,

    October 4, 2022 @ 7:42 am

    @Philip Anderson
    Initially, my inner grammar nerd crowbarred "the public" into place as the subject of both "provide any info to" and "talk about". It then took me a moment to realise that "won't talk [to the media] about the public", though possibly true in a pedantic sense, would be a rather weird accusation.

    Possibly "any info" was intended as the shared object (assuming the quote to be accurate); but for reasons I can't quite pinpoint, I find that parse really jarring.

  3. Trogluddite said,

    October 4, 2022 @ 7:44 am

    * Correction: The sentence in question is, of course, not a direct quote. More caffeine required!

  4. Adriane Boyd said,

    October 4, 2022 @ 8:44 am

    This is a nice chance to spin up the brand-new experimental spaCy coref pipeline for English (, which predicts that "He" refers to "Perla Huerta, former Army counterintelligence,". (Disclosure: I am a spaCy developer.)

    [(myl) Thanks! I'm happy to see that spaCy agrees with me :-)… and I'm looking forward to trying out the new coref pipeline! ]

  5. Terry Hunt said,

    October 4, 2022 @ 11:42 am

    "a tall, blond woman . . ."
    Do US writers no longer maintain the distinctions blonde (woman) / blond (man) and brunette (woman) / brunet (man)? Or is this antediluvian Brit outdated even in his own culture?

  6. Philip Taylor said,

    October 4, 2022 @ 12:38 pm

    This (probably equally antediluvian) Briton would use "blond" for a man but never "brunet". And upon reflection, he may even be more antediluvian, since he would never refer to himself as "a Brit" but only ever as "a Briton".

  7. SP said,

    October 4, 2022 @ 2:27 pm

    What struck me first was “migrants with lies”.

  8. Stephen Hart said,

    October 4, 2022 @ 5:22 pm

    Terry Hunt said,
    "Do US writers no longer maintain the distinctions blonde (woman) / blond (man) and brunette (woman) / brunet (man)? Or is this antediluvian Brit outdated even in his own culture?"

    I've always understood "blonde" used to mean a woman (always a woman), not a hair color,
    So "that blonde has blond hair."
    But not "that man has blond hair" and "that woman has blonde hair."

    “He still won’t talk about…”
    I initially took that to be twitter-speak, sort of equivalent to the (maybe now outdated) tendency to avoid uppercase and punctuation in email to indicate how "busy" you are.

  9. John Swindle said,

    October 4, 2022 @ 11:03 pm

    What Stephen Hart said about Twitter-speak. I don't even use Twitter, but I see enough of it quoted to expect it to be about as coherent as an SMS message. In other words, maybe coherent, maybe not.

  10. Brett said,

    October 5, 2022 @ 10:20 am

    @Terry Hunt: Looking at the OED entry (which is not very up to date in terms of citations), it seems that the distinction between male and female forms has never been consistently applied. Moreover, what spelling distinction was observed has been disappearing on both side of the Atlantic for a long time, just with different results. Regarding the etymology, the OED says:

    reintroduced from modern French in 17th cent., and still sometimes treated as French, as to be written without final e when applied to a man, especially substantively, a blonde; in North America commonly written blond like the French masculine, but in Britain the form blonde is now preferred in all senses.

    As confirmation of that last point, both Shaw and Chesterton were each using "blonde" as an adjective to describe a man all the way back before the First World War.

  11. Barbara Phillips Long said,

    October 5, 2022 @ 8:31 pm

    Bryan A. Garner confirms the confusion over blonde/blond and brunette/brunet in Garner’s Modern American Usage (although I have the third edition, not the latest iteration). He also warns that the blonde/blond has been used in sexist ways. (“Blond(e)” jokes being a prime example, in my opinion.)

    Garner says that British English is more likely to keep the distinction of blonde being f. and blond being m. as in French. He does not provide supporting evidence for his statement.

    He says in American English, “blond” is preferred almost all the time but that sometimes “blonde man” crops up. He shows several citations from the 1990s.

    On brunette, Garner notes that “brunet” is almost never used. I would agree.

    My view is that the blonde/blond distinction is going the way of actor/actress and host/hostess, where there is a gradual default to one term that is shorter and without a vestigial gender marker.

  12. Philip Taylor said,

    October 6, 2022 @ 2:05 am

    If one were to use the word "brunet", would it be pronounced /bru ˈnet/ or /bru ˈneɪ/ ?

  13. LW said,

    October 6, 2022 @ 5:06 am

    as a Br.E speaker, i've always used the gendered forms of blond(e), simply because i assumed everyone else did and it would be one of those cases where if you use the "wrong" form, annoying pedants will pop up to correct you. so i'm pleased to learn that this distinction is either on the way out, or else never existed to begin with.

    but it's interesting that, at least over here, the OED gives 'blonde' as the gender-neutral form, in contrast to other gendered terms where the gender-neutral version is formed by dropping the feminine suffix. i wonder if that's because blond(e) is typically used more often when describing women? (that seems like an obvious conclusion, but if i've learned one thing from reading LL, it's that the obvious answer to any linguistic question is probably wrong…)

  14. Philip Taylor said,

    October 6, 2022 @ 5:51 am

    Not sure where "over here" is, LW, but for me, accessing the OED from the United Kingdom, I see (under "blonde, blond") —

    B. n.
    1. A person with blond hair; one with light or ‘fair’ hair and the corresponding complexion; esp. a woman, in which case spelt blonde.

    so I am not sure what you are seeing that is leading you to believe that the OED regards "blonde" as the gender-neutral form.

  15. LW said,

    October 6, 2022 @ 6:14 am

    Philip: "over here" is the UK – that was probably not as implied as i thought it was from saying i speak Br.E, so i should have qualified that.

    regarding the OED, i'm referring to Brett's post above which includes the following quote attributed to the OED:

    "reintroduced from modern French in 17th cent., and still sometimes treated as French, as to be written without final e when applied to a man, especially substantively, a blonde; in North America commonly written blond like the French masculine, but in Britain the form blonde is now preferred in all senses"

    i understand "the form blonde is now preferred in all senses" to mean that "blonde" is used as a gender-neutral term to refer to a person of any gender.

    i don't have access to the OED to view the full entry, but in support of the claim that "blonde" is (increasingly) used as a gender-neutral term i offer the following Google Ngram plot of "blonde man, blond man" in the British English corpus:

    the trend seems to be much less significant in the American English corpus.

  16. LW said,

    October 6, 2022 @ 7:57 am

    to add to my previous post: i just noticed that Philip's quoted definition of "blond(e)" is for the noun sense. i'd agree that "a blond(e)" is still nearly always used to refer to a woman, but this discussion was started by the adjectival sense of the word – "a blond woman". so when i say that "blond(e)" is becoming gender-neutral, i am referring specifically to that sense.

  17. Barbara Phillips Long said,

    October 6, 2022 @ 8:24 pm

    @Philip Taylor —

    I don’t know how “brunet” is pronounced by English speakers, because I have only encountered it in writing. Men with brunet hair mostly seem to be described as being “dark-haired.”

  18. Barbara Phillips Long said,

    October 7, 2022 @ 3:21 am

    @LW: A quote from the Atlantic:

    In Italy, the far-right Brothers of Italy party proved last month that you can elect the most right-wing government since Mussolini if you find a pretty blond mom to lead it. If installed, that pretty blonde, Giorgia Meloni, will also have the distinction of being the first female prime minister of Italy.

  19. chris said,

    October 9, 2022 @ 3:43 pm

    I find it interesting that Filipkowski apparently sees no need to explain what he means by describing a human being as a coyote. I've heard the usage before but I would not have thought it so widespread that it could pass without explanation.

    P.S. I wonder if the AI parsing engine mentioned by Adriane Boyd considered "Martha" as a possible referent? Most readers probably wouldn't but the reason isn't limited to gender.

RSS feed for comments on this post