Over-reliance on English Hinders Cognitive Science.

That’s the title of a new Open Access paper by Damián E. Blasi, Joseph Henrich, Evangelia Adamou, David Kemmerer, and Asifa Majid whose Highlights section reads:

The cognitive sciences have been dominated by English-speaking researchers studying other English speakers. We review studies examining language and cognition, contrasting English to other languages, by focusing on differences in modality, form-meaning mappings, vocabulary, morphosyntax, and usage rules.

Critically, the language one speaks or signs can have downstream effects on ostensibly nonlinguistic cognitive domains, ranging from memory, to social cognition, perception, decision-making, and more. The over-reliance on English in the cognitive sciences has led to an underestimation of the centrality of language to cognition at large.

To live up to its mission of understanding the representational and computational capacities of the human mind, cognitive science needs to broaden the linguistic diversity represented in its participants and researchers.

The opening section includes this passage:

English has become the lingua franca in most spheres of international interactions, including science, and English-speaking countries are dominant global actors. The cognitive sciences are no exception. This state of affairs has resulted in a homogenous Anglocentric setup: English-speaking scientists explore the nature of the human mind by studying other English-speaking individuals in English-speaking countries (Box 1). In addition, while English itself is constituted of a number of distinct varieties around the world, including regional dialects, vernaculars, and Creoles, it is only a narrow set of these that participate in this near monopoly, most prominently Standard American English and British English.

Needless to say, the idea appeals to me, but I don’t know how reliable their methods and conclusions are. (Thanks, Bathrobe!)


  1. David Eddyshaw says

    The Moral seems sound enough, but the actual paper seems to have been constructed by going on a trawl for unfamiliar (to an English speaker) language features and slinging as many into the stew as possible, regardless of how, or whether, they illustrate any particular point particularly.

    The figures based on WALS seem – uncontributory, even as such things go.
    (But then, I am notoriously WALS-sceptic.)

    A number of individual assertions, in the few cases where I am in a position to check, look a bit off, e.g.

    In Yoruba, other West African, and East Asian languages, indirectness and vagueness are central parts of conversation

    The handwaving overgeneralisation* of “other West African” is naturally highly irritating to me; moreover, this (which is partly true, in certain respects, sometimes, in some places) is to do with culture, not language. While I would be the first so say that the two are intimately linked, attributing vagueness to the languages specifically is pretty sloppy.

    On the other hand, their conclusions are fine …

    * I think one can confidently say that anybody who cites Yoruba alone as representing “West African languages” actually knows no actual West African languages at all, probably not even Yoruba. I’ve seen all too much of this sort of thing, including in papers by actual (non-Africanist) linguists, syntacticians and typologists, who really, really ought to know better. Secondary sources get cited by secondary sources, and nobody in the discussion is actually familiar with the primary material. What’s the html markup for RANT, again?

  2. Lars Mathiesen (he/him/his) says

    You don’t need HTML for SHOUTING IN UPPERCASE!



  3. David Eddyshaw says



  4. J.W. Brewer says

    One of the papers mentioned in the linked piece attracted a series of largely-critical Language Log posts back when it was new. https://languagelog.ldc.upenn.edu/nll/?p=3756 (and follow-up posts linked at the bottom thereof)

    To say that norms in the society that uses a given language about how to use and not use that language in various sorts of contexts/settings are not in some sense part of the language seems debatable. They’re things you need to learn if you want to learn the language and understand what people are saying in it, just like morphosyntax and phonology. In languages with a T/V distinction in 2d person pronouns, are the “rules” (or patterns or loose tendencies) as to which gets used by whom when facts about the language or facts about something else?

  5. David Eddyshaw says

    To say that norms in the society that uses a given language about how to use and not use that language in various sorts of contexts/settings are not in some sense part of the language seems debatable

    Sure, very much so: that’s why I said language specifically; I don’t myself think you can study language adequately divorced from culture at all. Still, I think the way they put it is a feature of an overall somewhat scattershot approach in the paper.

    The indirection thing is quite striking to a European new to West Africa, though of course it turns up when people speak to you in English or French just as much as in Twi or Kusaal or Hausa. And to say it’s a “central part of conversation” is wildly overstating the case (in fact, it’s a distinctly Eurocentric judgment …)

    In my crossness on behalf of West Africa, I missed the fact that they cite the deeply silly Chen paper as if it reflected actual fact. I see they also claim that gender in language is linked to attitudes favouring female inequality. This doubtless explains why women in cultures where the main languages are Niger-Congo have so much higher status than those where it is Afroasiatic, and why Tuareg women are so downtrodden compared with Songhay women, and French women so much more oppressed than Turkish. All these are classics of junk science.

    Bad arguments for a true conclusion are still bad arguments; indeed, they are worse than useless, because they create the impression that there may be no good arguments for the conclusion.

  6. it’s a distinctly Eurocentric judgment

    That can be true, and at the same time it can be true that excessive concentration on English hinders you from noticing things of that nature.

  7. David Eddyshaw says

    True; and, as I say, I think their conclusions are perfectly correct.
    They might have taken rather more care in the way they arrived at them, though.

  8. J.W. Brewer says

    Niger-Congo languages often exhibit the world’s most elaborate systems of grammatical gender. That Africanists unhelpfully call these “noun classes” instead of “genders” just obfuscates things. That said, I’m not sure that the paper’s model even tries to predict what attitudes toward female equality possession of a dozen or more grammatical genders in ones L1 ought to, as it were, engender.

    I guess one question is whether studying cognitive differences between speakers of pairs of languages that are structurally quite different yet spoken in culturally similar and geographically neighboring societies (Hungarian versus Slovak, Marathi v. Kannada, etc.) would be a more straightforward way to investigate the sort of things this paper is concerned with?

  9. David Eddyshaw says

    That’s not “gender” in the sense that they mean: Niger-Congo languages don’t distinguish male from female in the the grammatical gender system (though they very often distinguish human from non-human.) I don’t actually know of any exceptions to this statement, though perhaps there are some out there somewhere …

    I see what you mean with your Hungarian vs Slovak etc; though how would you ever be able to tell with pairwise comparisons of that kind that it was specifically grammatical gender that accounted for any different status of Hungarian and Slovak women, given the great number of other language and cultural differences, except by assuming your pretended conclusion?

    I mean, maybe agglutinating men treat their women worse than inflecting men do. This would, of course, account for the attitude memorialised by Sapir:

    One celebrated American writer on culture and language delivered himself of the dictum that, estimable as the speakers of, agglutinative languages might be, it was nevertheless a crime for an inflecting woman to marry an agglutinating man.


  10. J.W. Brewer says

    Thinking that “gender” in a morphosyntactic sense presumptively means systems that distinguish masc. from fem. is a sign of ignorance of the considerably variability of the world’s languages in this regard.

    Here’s an unpaywalled copy of the paper (although without the appendices that would make it more easily understandable to me) that purports to find that:

    “[M]arried female immigrants [in the U.S.] speaking a language with sex-based distinctions in its grammar are less likely to participate in the labor market. This is true even after controlling for observable characteristics such as traditional household measures, husband characteristics, and bargaining power measures, as well as when controlling for a vast set of unobservable cultural forces through country of origin fixed effects. When we empirically decompose the relationship between gender in language and gendered behavior in this manner, it suggests that roughly two thirds of this relationship can be explained by correlated cultural factors, with about one third potentially explained by language having a causal impact.”


    But I agree with the point that the Blasi et al. paper alludes to a truly random-seeming grab-bag of studies (of perhaps different levels of plausibility), some of which get into topics (like rates of exchange of labor for cash by married women) farther afield from what I take to be the core domain of cognitive science than others.

    “Right-branching phrase structure [in the research subject’s language] is associated with better recall of the last (vs. initial) item in non-linguistic sequences” by contrast sounds like an empirical claim that ought to be testable and where it ought to be easier to control for cultural factors.

  11. David Eddyshaw says

    Welsh used not to have a future tense, but now does; this explains our relatively high rate of “obesity, smoking, drinking, debt, and poor pension provision” compared with the English (English, of course, has no future.)

    If only we had known! No doubt it seemed like a useful innovation at the time …

    There may very well be correlations between whether your L1 has sex-associated grammatical gender and your employment prospects as a married female immigrant to the US. Spanish-speakers from rural Mexico conceivably might be less successful, on average, than Finns, for example, and no doubt it would be easy to prove with sufficiently sophisticated mathematical techniques that this was in a large part due to Spanish gender marking.

  12. jack morava says

    @ DE,

    xlnt Sapir citation, kudos!

  13. I am sympathetic to the general conclusion, but would they all step off their helicopter and check things on the ground? WALS and typology are fine, but how about asking some actual translators or bilingual writers to confirm or elaborate on what the typological questionnaire implies?

    Blasi et al.’s method reminds me of those of Everett, Blasi, and Roberts (2014), who concluded that cold, dry climates suppress the existence of lexical tone through difficulties in phonation; or of Blasi et al. (2019), who concluded that [f] and [v] arose, right up to the present, through changes in subsistence leading to changes in jaw geometry. In both cases, the statistical methods were modern and computation-intensive; the diagrams were clear and colorful; but the authors never thought to collaborate with an actual speech therapist.

  14. J.W. Brewer says

    In absolute numbers the U.S. has a LOT more employed female immigrants from Mexico than employed female immigrants from Finland. The latter may earn more on average and/or have more “prestigious” jobs (to the extent “prestigious” still means something independent of “earning more”), but that’s not what the study is claiming. The U.S. still has plenty of low-paying jobs for folks with limited formal education and limited command of English! In any event, I don’t think the study is distinguishing the formally-employed from those seeking employment but failing to find it as opposed to from those who defy the dictates of neo-liberalism and goosing the GDP stats by deliberately living (on an individual basis, not even a household basis) outside the cash economy and specializing in household production, to use the proper economists’ jargon.

    I was once told by a Bengali-American friend that the propensity of Bengali-American married women to work in the cash economy was strongly correlated with whether their family was Muslim or non-Muslim, with both subsets of course having the same non-English L1. I have not tried to verify this independently. Bengali apparently has less grammatical gender than Hindi and many other Indic languages, but I don’t have time to increase my knowledge of exactly how much less.

  15. David Eddyshaw says

    Bengali apparently has less grammatical gender than Hindi and many other Indic languages

    Indeed; very much less …


  16. the actual paper seems to have been constructed by going on a trawl for unfamiliar language features and slinging as many into the stew as possible, regardless of how, or whether, they illustrate any particular point particularly

    You confirm what I felt. And they do manage to allude to a lot of suspect hypotheses, verging on pseudoscience. The lack/presence of future tense was once said to be correlated with the propensity to save, which seems to have been debunked.

    Of course, Chomskyanism gets off scott-free in this, despite the huge claims it makes as a correct characterisation of the human ability to use language, because it’s only concerned with syntax in an extremely narrow sense, even though its vision of syntax is heavily influenced by English.

    I agree with you on WALS. Even between authors there is disagreement on fundamental points (which has no doubt been exacerbated by the fact that the online version can be easily updated in a way the book couldn’t).

  17. I see that the future tense issue had already been alluded to by JWB.

    On gender, I think Lakoff’s Women, Fire and Dangerous Things deserves a mention.

  18. David Eddyshaw says

    Must read that again some time (I vaguely remember it being hard going and not as persuasive as I expected, but vaguely is the operative word.)

    It’s an interesting question in its own right, the interaction between meaning and grammatical gender. Although there seem to be few if any examples of systems with no correlation whatsoever, the details seem to vary so greatly from language family to language family that it looks to me like a mug’s game to seek much in the way of universal principles behind it all. Soap, porridge and war … as the Kusaasi say …

    There’s quite a cottage industry of papers trying to do this for various Bantu systems, typically pushing the various undoubted real correlations much farther than any actual facts seem to warrant. It is mandatory to mention Wittgenstein at some point in such papers. The most exciting ones are by L1 speakers appealing to their native intuitions.

  19. it’s pretty ridiculous – and a sad comment on linguistics as a field – that “grammatical gender” is anything but a tired punchline in the 21st century. among other things, because it makes it a lot harder to understand how noun class systems work in the very european languages taken as ‘standard’ when you refer to the entire set of systems by one idiosyncratic feature of one subset of such structures.

    the only function of using “grammatical gender” to describe noun classes is to pretend that the norm for human language is defined by a small number of european languages (whose outsized current political and academic importance is the direct result of colonialism).

    it’s like calling verb conjugation systems “grammatical age” because standard average european does an age-linked differentiation in the 2nd person.

    the persistence of this phlogiston crap is, of course, an example of the fundamentally correct conclusion of the paper, whatever critiques we can make of its arguments (i’d’ve ended the title with “cognition”, myself).

  20. Grammatical gender systems based on biological sex are common, not just in Europe (see here).

  21. J.W. Brewer says

    I should perhaps first note that it’s hard to dispute the high-level claim of the paper — although maybe focusing on not drawing cog sci conclusions exclusively from the WEIRD subset of humans will by itself lead to more linguistic diversity in the humans from which conclusions are drawn? (Although of course these days Anglophony is not an exclusively WEIRD trait although it tends to correlate with checking at least some of those five boxes.) And the stuff in the paper re leaning to read in a Latin-scripted language written left-to-right having a non-random impact on other sorts of cognition/perception that would not be the same for humans who became literate in other sorts of writing systems seemed at least superficially plausible.

    But separately, I wonder if rozele has the arrow of causation backwards here. Sort of stepping back in time and panning out in space, the broadest-scope core meaning of “gender” is something like type, kind, class. An etymological doublet of genus and genre, with originally similar semantics. It would then be the fortuitous fact that grammatical gender in the languages with which Anglophones tended to be familiar involved a masc./fem. or masc./fem./neut. schema, rather than any of the other ways of organizing grammatical gender found cross-linguistically, that enabled (firstly) “gender” to become a somewhat prissy euphemism for “sex” in English and then (secondly, over the last few generations) to become a word that could be used for talking about a bundle of characteristics that correlated heavily with sex but which it was thought analytically useful (first in academic circles and then in the broader culture) to distinguish from quote unquote biological sex because they were culturally variable and/or “socially constructed” and/or did not in fact correlate with quote unquote biological sex in 100% of the population. It is I suppose possible that what we might call the sex-adjacent senses of “gender” have now become so dominant in English that the old “genus, genre” sense has been irretrievably lost or skunked and trying to hold on to it in specialized-jargon domains will just end up confusing The Young People.

    Note also that the fairly impoverished use of grammatical gender in English can perhaps ironically make it feel more quasi-essentialist, what with inanimate objects (modulo the use of “she/her” for ships and balky mechanical thing-a-ma-bobs etc.) being uniformly neuter. Even within pretty English-adjacent languages with stronger gender systems, once you know that e.g. tables are masculine in German but feminine in French it becomes harder to take the conventional labels of the “noun classes” quite so literally.

  22. Writing system influence on cognitive function can be readily researched on Jews, some of whom made an aliya and others remained in galut. Obviously, almost all second and further generation Ashkenazim who ended up in Israel can write left to right as well, but if the effect is robust enough, it should be detectable.

    I am not sure why everyone is sold out on the main thesis of the article, in principle. Obviously, if someone wants to research cognitive patterns of humans in general, selection of English speakers is too narrow. But the paper is focused on a more specific claim “The over-reliance on English in the cognitive sciences has led to an underestimation of the centrality of language to cognition at large” is not obvious at all. As authors themselves purport to show, this over-reliance led to a situation where differences in language and in broader culture are not sufficiently separated from each other, which leads to all sort of idiotic claims of “no word for X” type. It might very well be, that popular notion of language shaping mind leaks into more scientific investigations and prevents researchers from seing clearly what depends on what.

  23. David Eddyshaw says

    Grammatical gender systems based on biological sex are common, not just in Europe

    Indeed: it’s the norm rather than the exception.

    If fact, all the examples I can think of of grammatical gender systems which do not pay any attention to biological sex are either secondary cases where originally distinct “masculine” and “feminine” genders have fallen together, or belong to just two language famiies: Niger-Congo (which for rhetorical purposes I will here pretend is a real thing) and Algonquian.* The opposite is found all over: Northwest Caucasian, Arawa, Lavukaleve, Yimas, Afroasiatic, Gunwinyguan, Eastern Nilotic …

    Even accepting the fact that there is a tendency for sex-indifferent systems to be taken as classifier systems rather than gender systems (and the distinction is not absolutely clear-cut) the general trend is clear. You’d think people were obsessed with sex, or something.

    * And, come to think of it, PIE itself in its earliest incarnations. Ironically …

  24. In Papua New Guinea, Austronesian languages almost never distinguish ‘he’ from ‘she’ in their 3rd person pronominals, but many Papuan (= non-Austronesian) languages do, so the influence of non-Austronesian neighbors explains those few that do. Papuan languages, for their part, rarely distinguish inclusive (we and you) from exclusive (we not you) pronominals, so Tok Pisin speakers in the Highlands often use mipela (me plural) in place of yumi (you and me), as naive English-speakers would be inclined to do.

    On one long stretch of coastline whose languages I researched, there are no Austronesian languages nowadays, but every subgroup of Papuan languages along that coast has innovative inclusive vs. exclusive distinctions in their pronouns, some even in their subject-marking enclitics. Those innovations, along with many Austronesian words (including for ‘canoe’ and ‘sibling of the opposite sex’) that can be reconstructed at the highest level of that subfamily, suggests that the present dwellers along that coast once had Austronesian-speaking neighbors.

  25. David Eddyshaw says

    Interesting …

    It doesn’t look like Proto-Volta-Congo distinguished inclusive/exclusive “we”, but quite a lot of individual Volta-Congo languages do. It seems a pretty borrowable feature, contrary to Johanna Nichols’ claims.

    Do you know of any Papuan languages that have grammatical gender not distinguishing male/female? (I’d have guessed that there must be some, PNG being World Linguistic Diversity Central and all …)

  26. Nichols seems to have changed her mind later. In an article with Bickell on clusivity, they admit that the concept of inclusive vs. exclusive ‘we’ is easy to understand and create. In the Binanderean languages I examined, the inclusive forms were innovated, mostly from emphatic forms, but in one case from the equivalent of yumi. (In moribund Rabaul German Creole, uns served as exclusive and wir served as inclusive.)

    I’ll have to look into Papuan gender/noun classes a bit more, but there is a moribund Oceanic language in West New Britain, Bebeli, that has gone crazy with male/female distinctions in its pronouns, adding male vs. female articles for its nouns, distinguishing male and female even for 3rd person plurals. Marital status also affects gendered pronoun usage! Unique among Austronesian languages as far as I am aware.

  27. @DE: Gender/noun classes are pretty common in Papuan languages, but less so in the most widespread Trans-New Guinea family. There are a few TNG languages with gender in the Bird’s Head region (western end of New Guinea island). Sound at all similar to Niger-Congo noun classes?

    Mian (Ok subfamily): “Masculine nouns include most male animates and, in the case of inanimates, nouns that are singular or small in size or quantity. Feminine nouns include female animates, some animals that are round and squat in shape (e. g. turtles, crabs) and in the case of inanimates, nouns that are plural or large in size or quantity. There is concord for gender between a head noun and its adjuncts.

    “In Marind (Drabbe 1955) four genders are distinguished by the final vowel of the noun stem. One class consists of male humans, which if singular are usually marked by the vowel e. A second class consists of female humans and animals, which if singular are marked by u. When the noun is plural both genders are marked by i. A third class, marked by e, a or o, consists mainly of trees and other plants. A fourth, residual class includes clothing and decorations, body-parts and some plants. Modifiers agree with the gender of the head noun, agreement being marked by ablaut of the last vowel of the adjective, making the vowels the same as the final vowel of the noun.”

  28. David Eddyshaw says

    Those sound like the usual sort, where biological sex is a major part of the grammatical gender system.

    I must admit I hadn’t really given much thought to the relative cross-linguistic oddity of Niger-Congo noun class systems in this particular regard, probably on account of being so used to them. I do remember that Greenberg thought that such systems were so peculiar that mere possession of something like that more or less guaranteed genetic cognacy (this was pretty much the sum total of his evidence for adding Kordofanian to the mix); I don’t remember whether the sexlessness was an important part of that in his view. (In any case, it seems an odd view for Mr Typology himself to have held, and anyway there seem to be fairly clear instances of such systems having demonstrably been borrowed.)

  29. PNG is one of the few places outside Africa where you can find (in only one area that I know of) coarticulated kp, gb unit phonemes (in the Papuan (TNG) Kâte). I read quite a bit of the Africanist lit in the 1970s while wrestling with verb serialization and word order change (SVO to SVOV to SOVV) in PNG. PNG Austronesian verb serialization is halfway between African-style (IIRC) same-subject across the verbs, and Papuan-style switch-reference constructions (where O1 = S2, as in ‘I chopped the tree toppled’).

    I believe there are some typologically Central African languages (Gbaya, Mangbetu) that mark both ends of their postposed relative clauses, a pattern widespread among PNG Austronesian languages, the first marker typical of Oceanic languages with N + RC word order and the final one presumably to help mark the ends of the NPs as they moved toward SOV word order. Lots of the SOV languages have proliferated sg/pl pronominal enclitics to mark the ends of NPs of all types.

  30. ‘I chopped the tree toppled’

    ‘the tree’ is actually shared between two clauses, as different arguments?

    (Shades of “If I had some ham, I’d have a ham-and-cheese sandwich, if I had some cheese.”)

  31. A bunch of Papuan languages around PNG have k͡p and/or g͡b, including Nen (which also has ᶰɡ͡b) and a number of TNG languages, as does the Austronesian Toqabaqita, of Malaita in the Solomons.

  32. David Eddyshaw says

    African-style (IIRC) same-subject across the verbs

    Yes, that’s the norm.

    Kusaal actually can do both same-subject and different-subject, though strictly speaking the relevant construction is not proper verb serialisation (according to Alexandra Aikhenvald’s criteria, anyhow), though it’s always been wrongly identified as such in descriptions of the language other than my own.

    The different-subject type actually can be predicative in Kusaal too, e.g.

    M daa pʋ nyɛ dau la ka o an na’aba.
    I TENSE NEGATIVE see man the and he be chief.NEGATIVE
    “I didn’t see the man as a chief.”

    where, despite my gloss “and”, the particle ka is actually subordinating, and is the different-subject counterpart of the same-subject linker n, as in

    M daa kuos lɔri tis na’ab la.
    I TENSE sell car.N give chief the
    “I sold a car to the chief.”

    where the pattern is very like the familiar canonical serial verb construction. (It isn’t, though; there is an explicit linker particle, and you can put adjunct NPs and even other subordinate clauses before the linker particle; moreover, the main clause can actually be non-verbal, so there is no way that “serial verb construction” can be a correct description of what’s really going on.)

    Kusaal relative clauses are internally headed and can occur anywhere NPs do within clauses: they aren’t of that extraposed adjoined type. (The Senoufo language Supyire has that, though.) The Kusaal construction must be a relatively recent development though; it’s not reconstructable even to Proto-Western Oti-Volta. I suspect it began life as a repurposing of the (pseudo) serial verb construction, with which it shares some tantalising formal similarities.

  33. @Y: I think the canonical version is, “If we had some ham, we’d have some ham and eggs, if we had some eggs.” (See here for some discussion of the logical structure.)

  34. gendered categorization of humans into various noun classes has the exact same relationship to noun class systems in general as the age-based categorization of human actors into different verb conjugations does to verb conjugation systems. the fact that they both mark socially significant categorizations does not change that. each is at most one part of some languages’ versions of those systems*.

    more to the point, even the noun class systems of languages whose noun class systems do include that feature cannot be understood based on it as a starting point. if someone wants to try to give it a try, may i suggest using the seven noun classes that uriel weinreich identified in yiddish as a test case (or even just the four he saw as active in NE yiddish)? please! tell me how you get from the specific gendered social system of eastern european jewry (or even the surrounding christian/roman gender system) to פֿוס falling in what weinreich** calls the “intermediate” noun class, or to a set of parameters for the words that fall in his “mass” noun class***! i’ll be over here laughing.

    this is week one of linguistics 101 stuff, people! (and one of the first moments when linguistics as a field intrudes on french, spanish, german, etc language classes for anglophones – generally on day one) y’know, the part where the embarassed professor makes incoherent excuses for still calling it “grammatical gender”, while pointing out at length that you can’t understand a noun class system if you’re thinking about it as being in some way based on human social categorization! (i recall niko besnier being particularly snide that day, but his overall snarkiness was part of his appeal.)

    * the number and distribution of the languages involved doesn’t matter to this linguistic point in any way. it does matter to the historical reason why linguistics as a field is still talking about phlogiston: that’s because the colonial european languages have this feature, and the field was made by people who based their ideas about what language-as-such is on those languages. this is about as controversial a historical point as victoria having married albert.

    ** who said: “the very imperfect correlation of the four genders [of NE yiddish; also: sic] with formal or semantic features appears to be typical of most gender systems [sic] in the world.” this is of course true as stated, of the social category of gender – but does also apply, as cousin uriel intended, to noun classes. and what it amounts to is a remarkably polite way of saying that you can’t understand these linguistic categories based on social categories (though there might be some insights to be gleaned the other way around: it may illuminate the social category of פֿרויען to know that a מײדל takes דאָס and not די – or it may not, given that all yiddish diminutives do****!).

    *** pertinently, weinreich identified the most fundamental distinction among NE yiddish noun classes “based on the morphophonemics of the article” as the one between the mass class and the other three noun classes.

    **** except in the dialects where they don’t.

  35. David Eddyshaw says

    this is week one of linguistics 101 stuff, people!

    It’s really not.

    Most languages with noun class systems do in fact show highly significant correlations between noun class and biological sex. Quite often sex is the only feature which has much value in predicting the class membership of a noun. That may be regrettable, but it is true nevertheless. European languages are not exceptional in this regard. They are typical.

    It is perfectly true that sex alone cannot explain any grammatical gender system: by definition, true. A system which can be so explained is not grammatical gender. It’s “natural” gender.

    It does not by any means follow that biological sex is irrelevant to grammatical gender. Systems in which it is completely irrelevant are the exception, not the rule, though the count is slightly confused by the fact that there are (happily) a lot of Niger-Congo languages, which do collectively constitute a grand exception.

    You can make a case that the word “gender” should be eliminated from linguistics (in which it originated) because it has now acquired other meanings which make the linguistic use confusing; but that is quite a distinct question from the actual facts of how noun class systems actually are in the world’s languages, and the cause is not advanced by misrepresenting this.

  36. I wanted to note that genus, genre and “gender” did not originally refer to the skirt-pants-beard-boobs complex (it is a relatively recent idea) and that in Russian it is still род “kind”.
    And also that studies in English may be focus on WEIRD cultures rather than just English.

    But JWB made both points.

    I can say about myself: it is tempting for me to look at African classes as a system similar to our рода.

    And as a Russian school graduate I think of our рода as an element of the declension system.
    “table m, window n.” rather than “girl f. boy m.”.

    But of course birch trees and mice in Russian fairy tales are going to be female (indepndently of school, just because they are “she”).

  37. J.W. Brewer says

    Back in the day (it was, may the Lord bless and keep us, the 1980’s …) I went to a university so sophisticated that what other places might call Linguistics 101 was Linguistics 110a — nine notches further up the pecking order! Somewhat regrettably, and probably due to the lingering effects of Chomskyanism, I don’t think we got very much into non-English (much less non-IE) gender/noun-class systems in that first class. It took another few semesters before we were having rambling conversations about how things supposedly worked in Dyirbal and Basque and whatnot, and it’s possible that the wacky gender/noun-class systems of the Bantu languages escaped my attention until a different class one semester past the one I was just adverting to.

    It is probably simultaneously true that linguists of certain prior generations started out with an insufficiently examined tacit assumption that Standard Average European defaults might well be universal defaults, but also true that they presumably learned their lesson and course-corrected sooner as to structural aspects of language where they didn’t have to get all the way to Algonquin-or-Niger-Congo to be contradicted by the empirical evidence.

  38. “…the very imperfect correlation of the four genders [of NE yiddish; also: sic] with formal or semantic features appears to be typical of most gender systems [sic] in the world.” this is of course true as stated, of the social category of gender…”

    Oh. It did not occur to me to apply the distinction “form vs meaning” to people:)

    (not making any point, just appreciating a funny comparison)

  39. The first English attestation of gender in the grammatical sense specifically mentions masculine and feminine. That it was borrowed from a French term which came from a Latin term with a neutral meaning is not germane. Gender is not used in current English to mean ‘kind’ or ‘type’ generically.

    In Hebrew, מִין min signifies alike ‘grammatical gender’, ‘biological sex’ (plus ‘sex’ as in ‘sex’), ‘species’ (in the biological sense) and ‘variety, type’, as in אַרְבַּעַת הַמִּינִים arba’at haminim ‘the four species’, four plants used ritually in the holiday of Sukkot. But even then, when you speak of מִין in the context of grammar, the association (at least to me) is with ‘biological sex’, not with ‘species/variety’. (The two genders of Hebrew are called male and female, and prototypically attach to the biological sex.)

    I’m saying that calling on the etymological fallacy doesn’t work to address rozele’s argument, whether you agree with it or not. I agree with her to a point, that the term encourages conforming to a particular model of nominal classification (however common). But it’s not as bad as other examples of forcing languages to look like Latin, etc.

    BTW, Modern Hebrew has a separate word, מִגְדָּר migdar, referring to social or personal gender, distinct from biological sex. That’s one of those clever innovated words, made to resemble a well-known foreign word, but with a perfectly good native derivation (from גדר gdr, ‘border, limit, fence’ etc., as e.g. לְהַגְדִּיר lehagdir ‘to define’).

  40. Back in the 1970s I attended a visiting scholar lecture by Charles Hockett (!) titled “F stops and cereal diet” in which he argued that people who ate pap (esp. rice) tended to have labiodental /f/ phonemes while those who ate firmer starches tended to have bilabials, being less weak-jawed. Well, I was then cutting my teeth on Yapese, whose principal starch consisted of hard, fibrous swamp taro (you can almost feel it hit bottom when you swallow) and who consume copious quantities of betelnut (in husk) between meals. Yapese has not only /f/ but also ejective /f’/ (along with other ejectives /p’/ /t’/, /k’/, and even /m’/, /n’/, /w’/ etc.). People listened politely, but none of the Austronesianists present were buying it, and I don’t think anyone else was either.

  41. I never ate swamp taro. I’m surprised it’s not prepared, like other difficult/tough/poisonous foods, in a way that would make it pleasant to eat.

  42. ” Gender is not used in current English to mean ‘kind’ or ‘type’ generically.”

    Of course it does not, but

    – the terminology for IE/AA “genders” and Bantu “classes” is shared by several European langauges. Not just English.
    – linguistics is older than modern gender studies.

    Do you mean that there is a certain English-specific (as opposed to European linguistics) bias in analysis of gender systems?
    I do expect such a bias, but more because English does not have genders. Yes, the term and its modern use also must contribute. But then we maybe should state more clearly what point we are making.

    (I was not really making any, just noted that we should keep in mind that gender did not mean boys-girls originally)

  43. @Y: There are many reasons to consider serial constructions like ‘I chopped tree toppled’ to be one clause in many PNG languages: scope of TAM, negation, adverbs, etc. In languages that have migrated their word order from SVOV to SOVV, you can find clauses that parse etymologically to ‘I pig spear-killed’ (same S, same O for each verb) and ‘I pig spear-died’ (speared-O = S-died). Most of the SOVV languages have relegated their manner-verb (V1) to the status of a ‘classifying prefix’ indicating the manner in which the result (V2) was achieved: ‘by hitting’, ‘by stabbing’, ‘by handling’, ‘by kicking’, etc. Meanwhile, the languages that have remained SVO have turned the second V into a class of very productive Resultatives (SVOR): ‘broken’, ‘dead’, ‘split’, ‘failed’, etc. That kind of summarizes one chapter of my dissertation.

  44. drasvi:
    gender did not mean boys-girls originally

    The earliest usage of the word gender in English, per the OED, is Hire name, þat was femynyn / Of gendre, heo turned in to masculyn. That from the 14th century story of St. Theodora (here), where Theodora, disguised in male clothing, is asked for her name, and says “Theodorus” (because future saints don’t lie, only fib).

  45. @Y, I think each subsequent generation of Europeans used this word just because authors of their grammar books did. The first attestation in English is not an interesting point in this respect: the first English speaker to use it only used it because French and Latin authors did. She just code-switched.

    I don’t know the history of its appearance in Greek (γένος) and Latin (pl. genera) but I suspect the person who coined it was thinking of “kinds”. I am not sure.

    What it means: we can’t say “this form is used in English because of the sex-gender association”. It is used because it is used….

    It specialised and became distinct from the doublets (or triplets?) “genre” and “genus” which mean something else in English. It seems this specialisation was motivated by its grammatical rather than social meaning. On the other hand, it expanded to cover boys and girls. Yet this expansion went in the direction grammar > sex rather than sex > grammar, and I think the grammatical meaning remained the core meaning for grammarians.

    But I agree that likely, at any stage (including the starting point in Classical languages), the grammatical cathegory and sex interacted in heads of its users, just because of how gender works in those languages.
    Even if the person who coined the word not mean it, likely she thought about it or felt it.

    And this interaction must be stronger in modern English, because
    – gender is more abstract to English speakers. They don’t know how Greek gender feels from inside. (Neither Greek speakers fully understand Semitic or Bantu systems, of course)
    – a word now is strongly associated with skirts and kilts.

  46. I am not sure whether I am merely repeating one of drasvi’s points, but it seems to me that that Theodora/Theodorus quote doesn’t prove anything. The word gender there is used as a term for grammatical gender of Theodora/us name, not person. Interestingly, OED also cites a 1398 text (admittedly full 8 years after 1390) with Byshinynge & liȝt ben diuers as species and gendir; for eueryche bischinynge is liȝt, but not aȝenwarde. which is not in a language known to me, but OED doesn’t classify it either under grammatical gender or biological sex which makes me to conclude that they (who apparently know what “bischinynge” means) think this usage of gendir means kind.

  47. Right, my point was that gender meant grammatical gender in the earliest usages, not ‘kind’ as with Latin genus.

    Byshinynge: see here.

    The meanings ‘grammatical gender’ and ‘kind’ came to be simultaneously, see examples here.

  48. Well, thank you. But isn’t it the whole point that grammatical gender always meant “kind” not “sex”, which in appropriate cases and by an obvious synecdoche were named after most prominent representatives, women and men, whenever the words associated with them happened to be a part of that particular kind.

  49. Lars Mathiesen (he/him/his) says

    FWIW, both boy/girl and masc/fem (in other languages) are called køn in Danish. In ON and as late as 1918 it could just mean ‘sort’, and I suspect that the grammatical use is a loan translation.

    Danish itself, otherwise quite standardly European, only has grammatical gender for nouns: intetkøn ‘neuter’ and fælleskøn ‘commune’. High-animacy pronouns are biologically gendered (han/hun) and it is pragmatically wrong to coordinate one of those with a neuter noun. (This since 1200 or 1400 or some such). All of Scandinavian too, except for some dialects and at least some of Nynorsk I believe, but I don’t know the timing there.

  50. I was thinking about possible synonyms for “m./f. [sex, gender, what?]”, e.g. kind – it occured to me that “half” can be used. Rather poetically.

    But…… This exact word (its Russian translation, pol) is used in Russian. It just does not feel as “half” anymore and honestly it sounds bad. If pol for me is just cold and official, while a phrase voprosy pola is mildly disgusting, the adjective polovoj (unlike pol used to hint on relationships between men and women, especially sex) and all phrases that include it are strongly disgusting. I don’t know if other Russian speakers feel the same.

    Fortunately, there is a word “fuck”… Unfortunately, some people dislike it too.

  51. Jen in Edinburgh says

    that they (who apparently know what “bischinynge” means)

    Shining, I think. Shiningness? Every shiningness is light, but not vice versa – although I have no idea what the distinction is supposed to be.

    It’s the kind of Middle English that you can half figure out by looking at it in Scots, but that’s not the same as knowing what they were trying to say!

  52. David Eddyshaw says

    weinreich identified the most fundamental distinction among NE yiddish noun classes “based on the morphophonemics of the article” as the one between the mass class and the other three noun classes


    Part of the reason why I am sceptical about noun class systems “proving” the genetic unity of “Niger-Congo” is that for this to work you evidently need not only for the classes to exist and, to a significant extent, correspond, but for the actual class affixes to be plausibly cognate. In fact, once you get outside Volta-Congo, the only cases that really look like more than wishful thinking* are the “human” class, at least in the plural (ba or the like) and the “liquids” class (in ma), exactly the classes with the most evident semantic associations of the whole lot, and thus the very classes where borrowing would be most plausible as an explanation of the similarities.

    In fact, even within Volta-Congo, where the protolanguage must undoubtedly have had noun classes more or less of the Bantu type, there is only one other noun class apart from the “human” and “liquid” classes which is absolutely and unequivocally reconstructable for the whole family: singular, ŋa plural (though another one with ŋʊ singular, ŋɪ plural also looks hopeful.)

    It works the other way too: the “liquid” class is very robust when it comes to resisting simplification of the inherited system. Mampruli, which has abandoned agreement by noun class, still has a three-way distinction of animate, inanimate, and liquid, and even in Kusaal, which has given up agreement altogether as a bad job and has a “natural” animate/inanimate system in its pronouns, some speakers still require agreement from adjectives modifying a noun referring to a liquid.

    * With morphemes of the form CV or V, a fairly restricted range of possibilities for both C and V, and a huge number of potential classes in each branch to pick and choose from (Nateni has twelve noun classes with grammatical agreement, nine of them with singular and plural suffixes), if you can’t find some lookalikes you just haven’t been trying hard enough. And one paper I saw purporting to show that the Kordofanian systems were cognate to Volta-Congo unblushingly shows bu and gu as “cognates.” You can prove whatever you like that way.

  53. Shining, I think. Shiningness?

    Y’s MED link says “Radiation of light; illumination.”

  54. J.W. Brewer says

    The good Anglo-Saxon descended word “kind” is in fact a cognate-cousin of the borrowed-from-Italic-via-French word “gender.”

    If you look into the etymologies of e.g. Latin “sexus” or German “Geschlecht” it feels as if the Euphemism Treadmill has been around for millennia and perhaps all IE words referring to the boy-girl-etc. distinction were generated by it.

  55. drasvi, I wouldn’t say that “voprosy pola” is disusting to me, but the word “pol” as applied to sex feels very dated and very formal.

  56. @D.O. maybe “midldly disgusting” is imprecise. What I mean is: already unpleasant, not just “formal”.

    But половой is worse. (Except in МПХ and половые органы which I find more or less neutral when in plural )

    but the word “pol” as applied to sex feels very dated and very formal.

    I meant specifically the adjective. If женский пол is just female gender, then phrases with половой refer to anything from dating to genitals. Maybe as a child I processed them as “adults mean sex, and see it as central to relationships but they also think it is obscene, so they use a medical word that does not mean sex”. Which is of course not the actual reason why these phrase were invented, but may be something like that was in the contexts where these phrases appeared.

  57. МПХ is (ironical) мужской половой хуй.

    A formal name of penis where the word “member” is replaced with the colloquial and highly obscene word for penis (thus preserving the worst of both words….)
    Lit. “male sexual dick”.

  58. David Eddyshaw says

    It’s good to avoid ambiguity.

  59. * And, come to think of it, PIE itself in its earliest incarnations.

    There are an awful lot of languages and families with animate/inanimate (aka human/non-human, rational/irrational, socially active/inactive, etc.): Basque, Elamite, Georgian, Mapudungun, Sumerian, Vietnamese; Algic, Chukotko-Kamchatkan, Siouan, Uto-Aztecan.

    (I note that in Ojibwe, the word for ‘raspberry’ is animate, but the word for ‘strawberry’ is inanimate. Then again, in Lojban, semantics is 100% irrelevant to gender: there are 18 genders, for words in b-, c-, d-, f-, g-, j-, k-, l-, m-, n-, p-, r-, s-, t-, v-, x-, z-, plus one for foreign words beginning with a vowel, which don’t have a coreferential pronoun.)

    “If we had some ham, we’d have some ham and eggs, if we had some eggs.”

    “If the dog would catch a rabbit, we’d have rabbit stew for dinner — if we had a dog.”

  60. Is that an actual saying? I can find no trace of it.

  61. Lars Mathiesen (he/him/his) says

    half: moiety if it’s not tied to sex.

  62. David Eddyshaw says

    There are an awful lot of languages and families with animate/inanimate (aka human/non-human, rational/irrational, socially active/inactive, etc.)

    Sure: but (except for the Algonquian part of Algic) those are natural gender systems, i.e. the sort where you can accurately predict class membership from the nature of the referent alone. Kusaal in its present-day incarnation actually has such a system, with an animate/inanimate gender system; trees are animate, which is odd from a SAE viewpoint, but is clearly for cultural rather than synchronically-unmotivated purely grammatical reasons.

    One mark of such systems is that synonyms can’t differ in gender/noun class, for example, although the same noun may have a different gender from usual if it’s being used metaphorically. (In the Kusaal version of 1 Corinthians 12:15, where body parts are represented as speaking, they are construed as animate gender, for example.)

    I was trying to think of grammatical gender systems (i.e. those where you can’t predict class membership, in the general case, from the semantics of the noun involved) where biological sex is irrelevant; PIE had such a system before the feminine (grammatical gender) developed (as in Hittite.) As I was saying, such systems are a minority cross-linguistically (or at least, they would be, if there weren’t so many Niger-Congo languages to tip the scales.)

    I must admit this is a cline rather than an absolute dichotomy. I’ve seen efforts (implausible, IMHO, but not stupid) to make out that in Algonquian, some inanimate objects are treated as grammatically animate for cultural reasons. (Actual L1 speakers seem keen on this interpretation.)

  63. David Eddyshaw says

    And there are a good many Women-Fire-and-Dangerous-Things-inspired efforts to make out that various Bantu systems can be more or less “explained” completely via semantics; they don’t work, at the end of the day, but the reason people keep on trying is because the system is very far from completely arbitrary semantically.

    I was reading some time ago about a gender/noun-class system where assignment apparently is pretty much completely arbitrary from a semantic POV; the author flagged this up specifically as a linguistic wonder of the language. Unfortunately, I can’t for the life of me remember what language it was … (it wasn’t Lojban, it was a real language [runs for cover])

  64. If I drink [colour]-N, then I drink [violet] wine.
    If I drink [colour]-F, then I drink [violet] vodka.
    Substantivated neuter “red” is commonly used to refer to red wine, and diminutive of substantivated feminine “white” is a nickname for vodka. Gender allows one to imply/hint at a noun.

    Similarly, both in Latin, Greek and Russian the neuter gender can mean “some sfuff with this property”.

    Also the neuter gender has a grammatical function in Russian. For one thing, “it” (as in “it’s cold today”) is third person and neuter. сегодня холодно – today cold-N.

  65. David Eddyshaw says

    Gender allows one to imply/hint at a noun

    It can do much more than that, if your gender system is less impoverished than in poor old Indo-European: in his grammar of Yimas (ten genders, plus six more comprising one noun each) William Foley reports that in a Yimas creation legend an important prop is introduced, ignored for about three pages, and then reintroduced purely though a verb agreement prefix.

    Niger-Congo genders/classes often have secondary uses marking particular grammatical functions, like your Russian example. In Kusaal, the old “liquid” agreement suffix -m is used with many adjective stems to make adverbs of manner (paalig “new”, paalim “recently”), and Swahili uses the singular class prefix ki- to make manner adverbs (mtoto “child”, kitoto “childishly”); presumably this is historically the reason why language names are ki-class too (“I speak Frenchly.”)

    Come to that, in Bantu languages, “locative” is usually a gender/class, taking verb agreement just like any other class. Nahuatl has something a bit like that, too.

  66. Yes. But what I meant here: Latin-Greek-Russian neuter seems to have something like a meaning of its own. And use of epithets (as with vodka) is a step in this direction.

    P.S. I wrote this before you updated your comment.
    Niger-Congo genders/classes often have secondary uses marking particular grammatical functions, like your Russian example.” – I didn’t know, thank you!

  67. Lars Mathiesen (he/him/his) says

    You could argue that “feminine with count number” is a sort of time adverb in Spanish: a la una/a las dos. I would be very surprised if historically it’s not just the word hora(s) that’s being assumed. The point being, would Kusaal paalim historically have been agreeing with some word meaning ‘manner’ that happened to be in that suffix class and has since been economized away?

  68. David Eddyshaw says

    I don’t think so.

    The -m class, which I have called “liquid” for convenience, comprises liquids (ku’om “water”, daam ‘beer”), “substances” (yaarim “salt”, bugum ‘fire”) and abstractions (sʋ’ʋlim “property”, dama’am “deception”); the handful of exceptions are historically refugees from defunct obsolete noun classes, e.g. piim “arrow”, which was originally a member of a u/i “long thin things” class which is no more in Western Oti-Volta, though alive and well elsewhere in the family (Nawdm fiimu “arrow”, fiimi “arrows.”)

    The primary effect of sticking the -m class suffix on an adjective stem is to create a derived abstract noun, e.g. pielig “white”, pielim “whiteness.” But Kusaal manner adverbs are basically a subset of nouns syntactically, so the further extension of meaning is natural enough in-language.

  69. John Cowan says

    “I speak Frenchly.”

    This seems to be a common idea: cf. Latine loquor ‘I speak Latinly’, where there is no question of an elided noun.

  70. David Eddyshaw says

    Same in Eskimo languages: one speaks “like a (real) human”, “like a Greenlander” (Kalaallisut, with the equalis case.)

    The m-class-as-adverbial sense probably explains why most Oti-Volta language names belong to that class, in fact: Yom, Nawdm, Mbelime, Waama, Gulmancema … the -lɪ “language” class seen in Kusaal, Mampruli, Buli etc seems to be a Western-Oti-Volta/Buli-Konni innovation, though it is eerily reminiscent of Luganda, Lingala etc. at the other end of Volta-Congo.

    Though I suppose you could argue that a “language” is an abstraction, after all (even without going Full Chomsky.)

  71. @DE, yes, in another thread I suggested that prototypically a langauge is what you refer to in a “Do you speak …. ?” question.

  72. Trond Engen says

    I speak Frenchly

    I speak Frenk-sh.

  73. David Marjanović says

    The Hittite genders have been called “common”/”neuter” and “animate”/”inanimate”; semantically, “common”/”inanimate” would perhaps be best, because while all animates have common gender, so does a large random sample of inanimates, while conversely all grammatical inanimates really are semantically inanimate.

    In languages with a T/V distinction in 2d person pronouns, are the “rules” (or patterns or loose tendencies) as to which gets used by whom when facts about the language or facts about something else?

    To me, in German, it feels nothing short of grammatically wrong to, say, use V forms and first names together. (…Even though there are regions where that’s a common combination.)

    I speak Frenk-sh.

    “I must be Frank.”
    – Emperor “Frank” Palpatine

Speak Your Mind