What Exactly Is Universal Grammar?

What exactly is Universal Grammar, and has anyone seen it?” (Front. Psychol., 23 June 2015), by Ewa Dąbrowska, Professor of Linguistics at Northumbria University, begins “Universal Grammar (UG) is a suspect concept” and goes on to back that up in a thoroughgoing manner. As a sample, see this devastating paragraph from the Conclusion:

Is it a fruitful approach? (Or perhaps a better question might be: Was it a fruitful approach?) It was certainly fruitful in the sense that it generated a great deal of debate. Unfortunately, it does not seem to have got us any closer to answers to the fundamental questions that it raised. One could regard the existing disagreements about UG as a sign of health. After all, debate is the stuff of scientific inquiry: initial hypotheses are often erroneous; it is by reformulating and refining them that we gradually get closer to the truth. However, the kind of development we see in UG theory is very different from what we see in the natural sciences. In the latter, the successive theories are gradual approximations to the truth. Consider an example discussed by Asimov (1989). People once believed that the earth is flat. Then, ancient Greek astronomers established that it was spherical. In the seventeenth century, Newton argued that it was an oblate spheroid (i.e., slightly squashed at the poles). In the twentieth century, scientists discovered that it is not a perfect oblate spheroid: the equatorial bulge is slightly bigger in the southern hemisphere. Note that although the earlier theories were false, they clearly approximated the truth: the correction in going from “sphere” to “oblate spheroid,” or from “oblate spheroid” to “slightly irregular oblate spheroid” is much smaller than when going from “flat” to “spherical.” And while “slightly irregular oblate spheroid” may not be entirely accurate, we are extremely unlikely to discover tomorrow that the earth is conical or cube-shaped. We do not see this sort of approximation in work in the UG approach: what we see instead is wildly different ideas being constantly proposed and abandoned. After more than half a century of intensive research we are no nearer to understanding what UG is than we were when Chomsky first used the term.

Anyone interested in this influential delusion of Chomskyism should read Dąbrowska’s paper. Thanks, Stan!

Addendum. ‘New mathematical methods’ in linguistics constitute the greatest intellectual fraud in the discipline since Chomsky, by Roger Blench: “For a method or disciplinary procedure to be deemed scientific it seems it should meet some minimum criteria […] It is relatively easy to show that on present showing none of these conditions are, or possibly can be met. If this is so, then the editor of Science has presumably been bamboozled.” Response by Sean at Replicated Typo, who disagrees with Blench but admits: “Those using mathematical models may have to spend more time justifying and clarifying their work.” Thanks, Yoram!


  1. Ken Miner says

    The reason we are no nearer to understanding what UG is than we were when Chomsky first used the term is that UG was an _assumption_. Remember the arguments? “What _else_ could explain how kids learn their languages so quickly from random data different for each kid?” Etc. You can’t hone in on an assumption as you can on the shape of the earth.

  2. John Roth says

    What I find amusing about this is that there is, in fact, a proposal for a universal grammar in the sense of a subset of grammar that’s shared by all humans. It’s part of the Natural Semantic Metalanguage in the sense that the 65 semantic primes have an accompanying grammar. A quick look at the references for this article shows that there are no references to either Wierzbicka or Goddard.

    See: https://en.wikipedia.org/wiki/Natural_semantic_metalanguage
    and: https://www.griffith.edu.au/humanities-languages/school-languages-linguistics/research/natural-semantic-metalanguage-homepage

  3. David Eddyshaw says

    It’s hard not to like a paper that references Isaac Asimov.

    In the context of pointing out the wholly unscientific nature of the way UG has shifted its ground to evade refutation over the years she cites “The Relativity of Wrong”, from which I particularly like

    “So, although the flat-earth theory is only slightly wrong and is a credit to its inventors, all things considered …”

    As with a fair bit of Asimov’s entertaining pop-science stuff, especially in the more philosophical areas, I’m not altogether persuaded that he hasn’t made a complicated issue just that little bit too simple to be quite correct. Fun, though.

  4. Ken Miner says

    “in the sense that the 65 semantic primes have an accompanying grammar”

    I know Wierzbicka’s work as an approach to _semantics_, but have not run across this idea before. Any refs?

  5. John Roth says


    It’s mentioned in several of their papers, but I haven’t seen it laid out systematically anywhere. I may have missed the appropriate paper, it may be paywalled, or they may just not have gotten around to documenting it. Frankly, I find the omission frustrating.

  6. David Eddyshaw says

    Makes all the difference – it would be scarcely surprising if all human beings shared concepts like “body” and “dying”; the claim that there are grammatical phenomena universally linked to them is much more difficult to swallow (unless “grammatical” is used in a sense so vague that the statement becomes true but vacuous.)

    There are languages in which family relationships are normally expressed by transitive verbs. Less exotically, it’s a linguistic commonplace that “good” “bad” “big” “little” (to take examples from the Wikipedia page) may be expressed in different languages by adjectives; by nouns; by verbs; by word-internal affixes; even by change of grammatical gender.

  7. NSM explications are normally written in a stereotyped subset of English, but as far as I know no theoretical claims are made for this subset, though it is obvious that because it is English and the NSMs are named in English, this affects the form that explications normally take.

    In any case, I would admire (as Mark Twain says) to see an explication of “broccoli”, “etymological nativization”, or “the United States of America”. It seems clear that these would all be encyclopedic in character! Mark Shoulson pointed out (probably on the Lojban or the Conlang list; I can’t find it now) that conlangs tend to use compounds like “small-green-tree-vegetable” for ‘broccoli’; while this is perfectly fine in itself, nobody would be able to recover the meaning ‘broccoli’ from such a compound. Lojban, indeed, has 1500 root predicates, ten times as large as the NSM list, and does not even attempt to cover such broad but shallow domains as species names, foodstuffs, and languages/cultures: the Lojban idea is to have a sufficiently broad list to make it possible to talk about anything.

  8. John Roth says

    Just got back from church, so I’ve got time to do a bit of searching. Some notes on the syntax of NSM primes are in: https://www.griffith.edu.au/__data/assets/pdf_file/0010/347527/on-going-development-of-the-nsm.pdf . It’s hardly complete, but it does give flavor of what the syntax is like.

    John Cowan: as far as the language of the explications, there is an explicit claim: valid explications can be translated to any language without losing any meaning. They’re in English because, at this time, English is, for better or worse, the working language of science.

    For “broccoli”, the approach in chapter 7 (Animals and artifacts) of Cliff Goddard’s textbook “Semantic Analysis, A Practical Introduction,” would seem to be appropriate. As you suggest, the explication, even with appropriate semantic molecules, would be rather long (he takes several pages to work though the description for ‘cat’.

    For a really loaded phrase like “the United States of America,” the approach in the papers in List C: Ethnopragmatics and Cultural Scripts, might be more appropriate. The discussion on pages 220ff in “Imprisoned in English” mentions a 21 page (!) discussion of the difference in German between Heimat and Vaterland. I seriously doubt that an in-depth analysis of “The United States of America” would be any shorter.

    In any case, explications use intermediate forms where appropriate: these are called “semantic molecules.” As a simple example, it takes a stack of five or six explications to get to what most people would regard as the simple concepts of brother and sister. Trying to do anything reasonably complex using just the primes would be horrendous, and wouldn’t be at all easy to comprehend. The explication for “brother,” for example (from “Imprisoned in English: the Hazards of English as a Default Language” p37) is:

    Someone can say about a *man: “This is my brother”
    if this someone can think about him like this:
    “His *mother is my *mother, his *father is my *father.”
    Someone can say about a *child
    if after some time this *child can be a *man.

    I’ve put an asterisk by the molecules – trying to expand this through several levels to just primes would render it unreadable.

    As you point out, a practical language would have a reasonably large vocabulary to make it possible to say things concisely. NSM is a fundamental basis, not a practical language.

  9. John Roth says

    I see I forgot to mention where List C can be found. It’s at: https://www.griffith.edu.au/humanities-languages/school-languages-linguistics/research/natural-semantic-metalanguage-homepage/resources .

    (This might get hung up in the spam filter for a while. Sorry.)

  10. David Eddyshaw says

    @John Roth:

    Very interesting – thanks.

    In using the word “syntax” with regard to NSM primes, the first paper seems to be really using a sort of metaphor, describing the sort of linkages that might characteristically occur *between* primes. It is logically conceivable that future work might lead to an actual mapping of this sort of thing onto what is usually thought of as syntax in language. At present, I wonder if calling it “syntax” is rather begging the question.

  11. Ken Miner says

    Excellent comments on NSM. I’ll make one last point and then I’ll shut up: Wierzbicka’s approach is relentlessly and assertively empirical. Thus it has the serious problem that by far most of the languages humans have ever spoken are gone forever. Rationalist approaches like Chomsky’s don’t have this problem. (Of course as Dabrowska wrote, they don’t seem to get anywhere either…)

  12. John Roth says

    @Ken Miner:

    This is only a problem if we assume that languages are independent of each other. If we assume that they are not independent (see, for example, the similarity of creole syntaxes) then it is not really an issue that we don’t have a comprehensive inventory of all languages that have ever been spoken.

    Apropos of that, I turned up another rather interesting paper: “Conceptual Primes in Early Language Development,” by Cliff Goddard (2001.) This is a diary study of one child’s language development during an approximately 18 month period. Early language development is, of course, one of the major approaches in trying to suss out basic elements of language. The paper is the next-to-the-last item in list D on Goddard’s NSM resource page.

    Taking the discussion in a somewhat different direction, there is a very suggestive chapter in “Imprisoned in English” (13: Chimpanzees and the Evolution of Human Cognition) about attempts to suss out a set of primes for chimpanzees – who, rather obviously, don’t have language in any human sense at all. There are a couple of papers that justify a set of approximately 70 primes (Goddard, et al, Forthcoming). That “Forthcoming” is ultimately frustrating.

  13. David Eddyshaw says

    You would not set out to discover the nature of fish in general by starting out from the premise that as they are all obviously related, you might as well confine your study to the goldfish in your pond.

    The “independence” of languages is one of the most interesting questions at issue. It should not be assumed as a premise.

  14. George Gibbard says

    I didn’t follow the link, but it seems wrong to think that everyone will have the same semantic decomposition of the concept “cat”, and it seems for some it will just be something like “animal a lot like the one I saw that one time that mommy called a cat”, about which fairly little may be remembered.

  15. George Gibbard says

    And I suspect if you asked people whether mountain lions were cats, you’d find both disagreement and uncertainty. I’m hoping someone tells us if Wierzbicka or Goddard handles this.

  16. John Roth says

    George Gibbard:

    Yes, they do. The only things they think are universal are the 65 semantic primes and 30 or so semantic molecules. Everything else is language and culturally contingent. The specification for “cat” in Goddard’s textbook should be understood as applying to the word in the Anglo dialects of English, and not otherwise.

    The utility of that specification is that it can be translated into any other language, and someone who speaks that language can then understand what a speaker of Anglo English would understand by it – with an appropriate amount of study, of course.

  17. David Eddyshaw says

    Says Blench:

    “Classification [of languages] is something that particularly appeals to middle-aged white males”

    Ouch. Direct hit.

  18. Blench carefully takes aim at Atkinson, Gray, and friends while ignoring the Philadelphians, who accept the comparative method in full: the problem of unusual approaches being stuffed under the rug. In addition, he does not seem to grasp how the historical sciences work (though he does say that cosmologists can’t do experiments): a tree diagram is not a classification whose utility depends on its purpose, but a hypothesis about how things once were based on inference from observations made today.

  19. Eli Nelson says

    The whole concept that words in natural languages can be decomposed to semantic primes seems implausible to me. I’m looking at some of the links from the Wikipedia article on “natural semantic metalanguage,” and I just encountered a spot in Cliff Goddard’s “The Natural Semantic Metalanguage Approach” where he writes on page 468:

    It would be incorrect to assume that shape descriptors are more basic than all body-part terms, however, because one human body part, namely *hands*, is necessary in the explication of shape descriptors themselves. This is because shape descriptors designate properties which are both visual and “tangible,” and to spell out the latter concept requires the semantic prime TOUCH (contact) and the semantic molecule ‘hands (M)’. For example:

    [G] *something long (e.g. a tail, a stick, a cucumber)*
    when someone sees this thing, this someone can think about it like this:
    “two parts of this thing are not like any other parts,
    because one of these two parts is very far from the other”
    if someone’s hands (M) touch this thing everywhere on all sides,
    then someone can think about it the same way

    To me, the above just seems ludicrous as a serious attempt to explain the semantic meaning of “length.” I don’t believe my own conception of “length” is semantically derived from the concept of “hands”. There are people born without hands, and as far as I know they are just as capable of understanding the concept of “length” as handed people. I have similar problems with explanations of “red” as “the color of blood” or whatever.

  20. JC: I agree with you that Blench is cheating, but only a little bit. Papers such as Ringe et al. and Chang et al., which are transparent and follow careful historical research, can be counted on the fingers of one hand holding a teacup. Moreover, the very term ‘linguistic classification’ reflects a view that good binary trees are a major goal of historical linguistics. I suspect that in the mind of many that is still true.

  21. As with a fair bit of Asimov’s entertaining pop-science stuff, especially in the more philosophical areas, I’m not altogether persuaded that he hasn’t made a complicated issue just that little bit too simple to be quite correct. Fun, though.

    In Asimov’s two-volume and entertaining commentary on the bible he somehow managed to forget to say anything at all about the Ten Commandments. Quite the accomplishment. I guess he didn’t like taking orders.

  22. John Roth says

    @Eli Nelson

    Please note that he says explicitly that shape descriptors are both visible and tangible. If you want to define a shape descriptor that is only visible, or one that is only conceptual, or one that conforms to some system of measurement, it would need a different explication.

    A lot of words have multiple meanings, depending on the approach you take and the context in which you find them. The NSM approach is to focus in on exactly one of those meanings, leaving the others to the side to be dealt with elsewhere.

  23. J. W. Brewer says

    I’m a bit puzzled by what role the “natural” in NSM is supposed to be doing. It seems like rather an extreme version of projects like “Basic English”: that it may be possible through sometimes clumsy and lengthy paraphrase to more or less express the approximate meaning of most pragmatically-plausible sentences in a given natural language using only a highly-restricted and carefully-selected lexicon (and a smaller one the more arbitrary/artificial you’re willing to be and the more elaborate and inefficient the resulting artificial compounds you’re willing to tolerate) is an interesting party trick, and it may even be helpful when trying to figure out how to draft instructions or warning labels for foreigners with only minimal command of the language, but it doesn’t seem like it tells you very much about the language itself in its fullness.

  24. J. W. Brewer’s reactions are mine as well.

  25. The idea, I believe, is to determine what synonyms (whether in the same or in different languages) have in common (or not) by breaking them down. Thus French empirique vs. English empirical:

    someone thinks like this (about something):
    “I want to know some things about this
    maybe I can know these things if I do some things
    because of this, I want to do some things
    I don’t know well what I can do
    I don’t want to think about it for a long time
    I know that before, when people did some things
    they could know some things about things like this because of this
    maybe the same will happen to me now”
    it can be bad if someone thinks like this

    someone thinks like this (about something):
    “I want to know some things about this
    I know that people can’t know things like this about something
    if they don’t do things of some kinds to some things
    if people do things of these kinds to some things, they can see some things because of this
    at the same time some parts of their bodies can touch some things
    after this, they can know some things because they have done these things
    I want to do some things like this now”
    it is good if someone thinks like this

    Note that the two explications irretrievably diverge after the first two lines.

  26. Trond Engen says

    Heh, I didn’t understand that “universal semantic molecules” was meant to say something about real language. That sounds even more phony than “universal grammar”.

  27. Trond Engen says

    But this sounds more like an aid to automated translation. That might well be useful. But it’s still bound to lose a lot of nuance.

  28. The idea, I believe, is to determine what synonyms (whether in the same or in different languages) have in common (or not) by breaking them down.

    But that presumes that word meanings can be broken down in a unique and exhaustive way, like breaking a physical body down to its constituent molecules, and that strikes me as utterly insane. I mean, I thought about trying to do that kind of thing when I was in the painful process of transitioning from a math major to a linguistics major — “Hey, let’s analyze language in mathematical terms!” — but I quickly realized what a sophomoric idea that was. Which was fine for me (I was, after all, a sophomore in college), but seems like a strange thing for adults to spend energy on.

  29. No offense to those of you who work on that stuff, I don’t think you’re either insane or sophomoric, I’m just trying to convey my instinctive reaction to it.

  30. It doesn’t have to be unique, just exhaustive. The logic behind the term “natural”, by the way, is that in principle the explications can be translated into any natural language without losing information: therefore, any language can serve as its own metalanguage (or indeed the metalanguage of any other language). Of course, as with Newspeak, you have to keep firmly in mind that the words of the metalanguage have narrow, specific meanings, not the broad meanings they have in a natural (use of the) language.

  31. The logic behind the term “natural”, by the way, is that in principle the explications can be translated into any natural language without losing information: therefore, any language can serve as its own metalanguage (or indeed the metalanguage of any other language).

    Same idea as “universal grammar,” I guess. I say it’s spinach and I say the hell with it.

  32. I mean, one speaker’s English is not the same as the next speaker’s English; how can anyone imagine… oh, never mind, it’s too hot and muggy to finish the sentence.

  33. David Eddyshaw says

    The idea that the world of ideas can be decomposed into “atoms” is a hardy perennial reminiscent of all those lovely renaissance ideas for universal scripts capable of uniquely classifying all concepts and assigning unique symbols accordingly.


    Not even language really works that way, let alone the entire conceptual world of humanity.

    Wittgenstein. Lakoff.

    I’m struck however by Ken Miner’s remark that “Wierzbicka’s approach is relentlessly and assertively empirical.” It seems to me that there is considerable danger in enterprises like this of inadvertently imposing your own mental categories on the data unless you can somehow find a hook outside your own preconceptions. A priori assumptions as to the fundamental similarity of all human languages and thought worlds seem an unpropitious place to begin. But I speak in complete ignorance of Wierzbicka’s actual methodology (which may indeed be truly empirical) and may therefore be utterly wrong.

    Even if this grand project of atomistic analysis of concepts is possible, it does not amount to anything you could call a “universal grammar” (except by altogether divorcing the word “grammar” from its accepted sense in linguistics.) To do that for real you would need the further step of being able to consistently map the concepts into actual utterances of existing natural languages. You’d need to make generative semantics *actually work*.

    Or as Hat rightly says – spinach.

  34. Trond Engen says

    I think what bugs me most is the level of precision. For every new semantic atom you add to a definition, there’s a wide variety of new nuances, potential additions of other atoms. Semantics is fuzzy. And the borders of semantics itself are fuzzy. Pragmatics keep leaking into it, and it spills over into grammar. Or was it the other way around.

  35. Trond Engen says

    Or: Yeah, what David said.

  36. Bathrobe says

    The example of amae at Explications left me scratching my head. Is this really how amae should be understood?

    someone X feels amae (towards Y) at this time:
    someone X thinks like this at this time (about someone Y):
    “this someone can do good things for me
    this someone wants to do good things for me
    when I am with this someone, nothing bad can happen to me
    I want to be with this someone”
    because of this, this someone feels something good at this time
    like someone can feel when they think like this

    But there are probably advantages with this approach. One that occurred to me is the Japanese word 悔しい kuyashii, which can be hard for English speakers to understand. If you look on the Internet you’ll just find the old standbys from ancient dictionaries, which are extraordinarily unhelpful:

    悔しい: vexing; annoying; frustrating; regrettable; mortifying

    Explanations at Italki are:

    Best answer: Regrettable (I’m not sure how this got best answer.)

    Other answer (obviously by a Japanese native speaker):
    1. means go through a failure or shame, and can’t give up or forget it. it can translate to mortifying
    2. regrettable

    The second answer is actually more illuminating, despite the non-native English.

    How to Japanese explains 悔しい as the feeling he experienced when “the Seattle Seahawks ended the Saints’ hopes of repeating their championship last year”, remarking that “in English it would take a lot more to describe my feelings. I was totally broken, exasperated, depressed. It sucked.”

    The Natural Semantic approach could be useful in unravelling the semantics of this kind of expression since it attempts to break it down into its essential elements.

  37. But how is that an improvement over just figuring out how to explain it in terms an English-speaker can understand (e.g., “go through a failure or shame, and can’t give up or forget it”)?

  38. John Roth says


    The acid test of their methodology is this: the 65 primes are said to exist in all languages with the same meanings and to be intuitive; that is, there is no way of defining them in simpler terms. That not only appears to be true in all languages that have been examined, but finding a language where one or more doesn’t exist would challenge the notion that the missing concepts are primes.

    To give an extreme example, consider Piraha – a language that is notorious for not having a concept of number. Looking at the list of primes, number is not on it. It’s not, according to NSM, a basic, irreducible concept. One, two, some, many and few are on the list, all of which I think can be found in Piraha. Interestingly, none is also not on the list of primes, which makes one wonder about the arguments in the Renaissance about whether zero was a legitimate number.

    The same is not true of semantic molecules: there are possibly as many as 30 that appear to be pervasive in all languages studied, but finding some of them missing wouldn’t be a challenge; it would simply be one more example of cultural diversity.

  39. David Eddyshaw says

    Happy to accept that Natural Semantics (or similar) may have practical utility; indeed anyone who believes that real communication is possible across linguistic and cultural barriers (and it plainly *is* possible) probably needs to invoke something rather like it as an intermediate form of communication at some point. Something with lowest-common-denominator words, so far as possible bleached of the manifold nuances and distracting second meanings and penumbra of associations real words have in real languages.

    Though I’m not so sure that what is going on in these cases is actually always *analysis* in the strict sense of breaking a complex into simplices; often it is really exemplification or analogy or appeal to presumed common experience over part of the domain of the concept to illustrate the other parts – to call that “analysis” is at best a metaphor, at worst an attempt to claim as a discovery about the structure of concepts what is in fact an assumption that concepts are like molecules and can be broken into atoms, and that to do this will explain them.

    There also seems to be a danger of falling into the Basic English trap; limiting your vocabulary is all very well, but as soon as you try to say anything non-trivial the complexity immediately reappears, but now in combinations of words whose overall meaning is not deducible from the components. “Put up with” vs “tolerate.”

    In any case it’s a big step from “[Esperanto] is very useful in cross cultural communication” to “[Esperanto] is the essence of Language itself.”

  40. David Eddyshaw says

    @John Roth (previous reply was to Bathrobe. I really must learn.)

    Thanks, that’s illuminating; in particular, I think it clarifies that my comment about attempting to atomise the entire conceptual human world is beside the point, as (if I understand you right) the project is on the contrary very much a minimalist one, looking for the most number of concepts which can be plausibly attributed to *all* human groups. Presumably it need not follow of necessity that all concepts in all cultures need necessarily to be ultimately reducible to these 65?

    Though to call all concepts other than the 65 “molecules” does seem to imply this?

  41. David Eddyshaw says

    The undoubtedly basic verb “ji” in Hausa means “hear, feel, taste, smell”, i.e. “perceive by a sensory modality other than sight.”

    Hausa speakers are (of course) perfectly able to distinguish between hearing and tasting, and to explain the difference in Hausa without any trouble. There is no possible doubt about them having the separate *concepts.*

    But this is not compulsorily encoded in language as it is in English. We can be confident that Hausa speakers share similar concepts to ourselves in this domain, but surely from the Hausa language itself studied without preconceptions, one would have to say that the concept “see” is atomic, but “hear” is not: it is to “ji” with one’s ears.

    You could say (and I would!) that “hear” is the “core” sense of “ji”, but personally that gives me the uneasy sense that I am imposing my own Anglophone categories surreptitiously in doing so. How can one guard against that systematically? Or is the prime really better called “ji” than “hear”?

    Thirty languages is a tiny subset. But admittedly one needs to start somewhere …

    Talking of “ears”, I know no language which lacks a word for “ear” and can conceive of no human culture lacking the concept. Why is that not a prime? Is it a composite concept? Offhand I can think of no language in which the word for “ear” is not basic underived vocabulary (though it would not surprise me if some erudite Hattic can.)

    This suggests to me that I must have missed something fundamental.

  42. John Roth says


    Yes, it does imply that all concepts must be reducible to those primes. If something isn’t reducible, then either they’ve found another candidate prime, they’ve been insufficiently clever, or they’ve found a mutant strain of humanity that has an extra built-in concept. (I wouldn’t, by the way, rule this out. We know so little about how language actually works in the brain that there might be all kinds of similar oddities out there).

    I’d also like to mention that Anna Wierzbicka is writing a children’s book titled “The Story of God and People” using the NSM approach; it includes semantically equivalent translations in English, Tok Psin, Ewe, Russian and Chinese. (I presume it’s MSM, “Chinese,” could describe a number of different languages.) The first chapter in all five languages is here, in a lecture she gave on receiving the 2010 International Dobrushin Prize. https://www.griffith.edu.au/__data/assets/pdf_file/0015/453201/Wierzbicka_2011_Common-Language-of-All-People.pdf .

  43. David Eddyshaw says

    Indeed yes:


    So something can indeed be universal to all languages and cultures and yet be a molecule rather than an atom, even if the word expressing it is typically monomorphemic or at least underived within that language.

    It seems to me that the gulf between this kind of semantic analysis and the study of the syntax of actual languages is even wider than I thought (a fortiori morphology and phonology.)

    A lot of this is highly reminiscent of generative semantics.

  44. David Eddyshaw says

    @John Roth: (see, I do learn)

    Much appreciate the linked pdf of Anna Wierzbicka’s ipsissima verba, which are most clarifying (and very interesting to boot.)

    Interesting to see the references to Leibitz and Wilkins (though of course AW believes she is discovering rather than creating.)

    It won’t astonish you to learn that I have major reservations about this; but it’s nonetheless extremely interesting and thought-provoking. Unlike Chomskyism it looks to me like the sort of wrongness (sit venia verbo) which could at any rate be intellectually fruitful. That can sometimes be more useful than being right in the end.

  45. John Roth says

    @ David Eddyshaw

    Yes, we credit the pioneers because they explored a direction, not because they were right in all particulars. I had to look up generative semantics; it does seem to have a certain similarity in that “deep structure” is placed where it belongs; in the person’s head. Otherwise I find NSM to be much more minimal and pragmatic. The absence of highly abstract theorizing is refreshing.

    There’s a question that arises quite naturally: if humans have 65 irreducible concepts (primes), then when did the different ones arise in human ancestors. Is there any way to know, or even give a reasonably grounded speculation? Wierzbicka jumps off into that territory in chapter 13 of “Imprisoned in English,” “Chimpanzees and the Evolution of Human Cognition.” I’ll have to admit that, since I’ve got a slight background in paleoanthropology, I found the speculations interesting but hardly convincing. The experimental program leading to a highly tentative list of about 60 chimpanzee primes was most interesting.

  46. George Gibbard says

    Language communities that have cilantro have a semantic prime that other language communities lack. The same is true of Sichuan pepper. But it still is true that word meanings usually involve similarity to a prptotype rather than necessary and sufficient conditions.

    As for “The Story of God and People”, Heaven help us.

  47. J. W. Brewer says

    If we’re being parsimonious, the irreducible minimum number of conjectural semantic primes needed to account for the full lexical inventory of the natural languages typically spoken by chimpanzees would be zero, wouldn’t it?

  48. forget to say anything at all about the Ten Commandments

    Since Asimov’s Bible commentaries are firmly in the aggadic rather than the halakhic tradition, that’s really not too surprising.

    relentlessly and assertively empirical

    Indeed, and this is a Good Thing, since that’s what proper scientific theories are. If a given prime turns out not to be expressible in some language not yet looked at, or if it turns out that a given prime can be explicated in terms of other primes, or a concept is found in some language that cannot be explicated using the existing primes, then the theory needs to be changed. Since in NSM terms the atoms are truly atomic, and not generated from some underlying set of linguistic quarks (or quirks), there is no theoretical problem with expanding or contracting the list, and indeed all of these cases have happened. NSM can never be proved, only disproved, and this again is what actual scientific theories look like: see almost any David M comment these last N years. (In particular, Everett’s most recent investigation of Pirahã says it does not have words for ‘one’ or ‘two’, and this indeed is a problem for NSM.)

    I don’t believe my own conception of “length” is semantically derived from the concept of “hands”.

    No, indeed. NSM is not a theory of language production or language comprehension or language acquisition. Children learn cat (if they are in an environment with cats) well before they learn for some time, and the former is clearly more psychologically basic. The claim is that from a small set of primes it is possible to state the meaning of any of the world’s 100 million (at a guess) concepts in a precise way that can be rendered into any language without deviation of meaning. That’s a strong claim and requires strong evidence.

    one speaker’s English is not the same as the next speaker’s English

    Disclaimer: this is a theory of semantics, not pragmatics, and I think the distinction is essentially artificial, semantics being a theory of what sentences don’t in fact mean. But let’s say that semantics is a Newtonian simplification of meaning that discards pragmatics for the time being.

    It’s not clear if this is a complaint about the object language or the metalanguage. If you and I use cat in different ways (that is, if there is some object X such that you’ll assent to “X is a cat” and I won’t, or vice versa), then we speak (slightly) different languages and our words need different explications. Similarly, polysemy requires multiple explications for the same forms.

    But if it’s a complaint about the metalanguage, then I don’t think it’s relevant. Do you really think that you and I, both speakers of English, use two in different ways? And after is polysemous all right, but the sense of the prime is always temporal, never merely sequential, and anyone who speaks and understands English has no trouble differentiating the two.

    But how is that an improvement over just figuring out how to explain it in terms an English-speaker can understand?

    Because it’s one and done. Instead of figuring out what the rendering of each language-specific concept in each other language is (billions of translations), you figure out an explication for it using the primes, and then translate the explication without semantic loss into whatever language you like.

    The undoubtedly basic verb “ji” in Hausa means “hear, feel, taste, smell”

    Sure. Again, there is no reason why the translation of a prime can’t be polysemous, as long as you grasp that in interpreting explications only one particular sense is meant (see the discussion of after above). Nor is it a problem if the translation of a prime is a phrase (or a mere bound morpheme, for that matter). If ji is polysemous, then it can get several explications; if it is merely broad and vague, then an explication for it would express that broadness and vagueness in the same style as “perceive by a sensory modality other than sight”.

    Why is [ear] not a prime?

    Because an explication of it, presumably in terms of hear and body can be found.
    That is quite independent of it being basic vocabulary in the usual sense.

    believes she is discovering rather than creating

    In a sense. I don’t think there’s any claim that this set of primes is a unique basis: it may well be possible to formulate a different set that meets all the constraints, so in that sense there is some creativity here. I’m thinking of the various formulations (as opposed to interpretations) of quantum mechanics: wave mechanics and matrix mechanics appear to talk about different things, but the use of one can always be translated into the use of the other.


    Species, foodstuffs, and cultures are always a big challenge for this sort of theory, as they shade off gradually into names, which are not explicable in terms of primes (I suspect that “the United States of America” does not need an explication any more than “John Woldemar Cowan” does). But with enough biological knowledge it should be possible to give an explication of cilantro on the lines of the one for tomato — and the claim is only for possible decomposition, not for natural or plausible decomposition.

  49. David Eddyshaw says

    So what, exactly, would constitute a refutation of NSM?

  50. “…extremely unlikely to discover tomorrow that the earth is conical or cube-shaped…”

    You are only saying that because you haven’t met Les Shadoks:


  51. A good question for which I do not have a full answer, but certainly the system would collapse if it could be shown that there were languages into which the explications could not be adequately translated, not for lack of specific primes, but for general reasons. It’s clear that at first far too little work was done on establishing the constrained syntax of explications.

  52. It may be that NSM is like optimality theory, not really a theory at all but a framework for constructing (proper, refutable) theories. Specific optimality theories can be and have been refuted, but the general idea “from all possible outputs, pick the one that best satisfies a given list of constraints in a given order” itself cannot readily be falsified. All you can do is construct theories that do or don’t fit this framework and see if they suit your needs better.

  53. George Gibbard says

    It may be that Optimality Theory cannot be falsified but last I knew it was at least suspicious that noone had a believable account of “opacity”, or effects that are easily explained by rule ordering but not otherwise, in OT (except for stratal OT, in which one can have multiple, ordered optimizations in the derivation of a form). An example of this would be: we can account for Hungarian vowel harmony if we assume that in addition to the surface vowels, there are also at an earlier stage of the derivation the back vowels *ï *ïː *ëː; then vowel harmony works fairly exceptionlessly, but then later in the derivation these vowels merge with the front vowels i í é. As of several years ago noone had come up with a decent OT account of how Hungarian vowel harmony is optimal at the phonetic surface. Of course what is really going on is that Hungarian vowel harmony used to be phonetically natural, but through sound change it has become unnatural and abstract.

  54. George Gibbard says

    And *ë before another vowel as in l(e)ány ‘girl’.

  55. What is usually called OT is an example of a specific optimality theory in phonology, but it’s possible to have optimality theories in any area, or even outside linguistics altogether.

  56. a 21 page (!) discussion of the difference in German between Heimat and Vaterland.

    This is epistemological realism gone mad. The differences (!) between the two words are no different in kind and scope than those between “cat” and “pussy”. Ignorance of German is an adequate explanation for those 21 pages.

  57. A 21 page discussion of the difference in English between “cat” and “pussy” – composed by a Chinese scholar.

  58. David Marjanović says

    Heimat: “home” in a geographic sense.
    Heimatland: Heimat at specifically the country level (current or desired/irredentist).
    Vaterland: patriarchal, militaristic version of Heimatland.

    There. 21 pages.

    ‘New mathematical methods’ in linguistics constitute the greatest intellectual fraud in the discipline since Chomsky, by Roger Blench

    I’ll have a lot more to say about this later. In short, half of it is good, half is bad for a convoluted heap of reasons.

    But “fraud”? WTF? What happened to Hanlon’s Razor!?!

  59. Hanlon’s razor: “Never attribute to malice that which is adequately explained by stupidity.” Thanks, I’d seen the statement but didn’t know its name.

  60. David Marjanović says

    (I prefer versions with “incompetence”. Never assume malice where incompetence would suffice for example.)

  61. Some “cultural scripts” (which are like explications but at a higher level of discourse) from this paper (edited for continuity):

    Script A [below] is arguably a master script of Anglo culture, expressing a cultural preference for something like personal autonomy . It is associated with the Anglo cultural key word freedom.

    A: an Anglo cultural script connected with “personal autonomy”

    (people think like this:)
    when a person does something, it is good if this person can think like this:
    “I am doing this because I want to do it”

    Many important Anglo speech practices flow from this script and others allied to it—above all, the avoidance of direct or “bare” imperative and the existence of a range of alternative strategies such as the prolific interrogative imperatives, so characteristic of English, and common “suggestive” formulas, such as You might like to …, Perhaps you could …, and I would suggest …. Other related phenomena include relative avoidance of the performative use of ask (cf. *I ask you…) and of pleading, begging, and other modes of “insistent asking”, and the existence of common disclaimer formulas, such as It’s up to you, You don’t have to, Only if you want, and so on. Having regard to the strongly negative impression that direct imperatives can create in Anglo culture, Wierzbicka goes so far as to say:

    The avoidance of the imperative in modern English and the development of an extended class of interrogative directives (so-called ‘wh-imperatives’’, e.g. ‘could you/would you do X’) is a linguistic phenomenon whose cultural and linguistic significance can hardly be overestimated. It is a phenomenon which should be the subject of the first lesson in acculturation taught to every immigrant to an English-speaking country ….


    Script B is arguably a master script of Russian culture, expressing a cultural endorsement of, roughly speaking, an “expressive” stance in speech and action, and linked with the Russian key word iskrennost’ (roughly, ‘sincerity’).

    B: a Russian cultural script connected with “expressiveness”

    (people think like this:)
    it is good if a person wants other people to know what this person thinks
    it is good if a person wants other people to know what this person feels

    Script B is one of a family of Russian cultural scripts which help explain and motivate the characteristically Russian preference for and appreciation of frankness of expression, uninhibited by concerns as to whether the thoughts and feelings expressed are confrontational, negative or socially unacceptable. Linguistic manifestations include various particles, interjections and response expressions (such as Nepravda! ‘Untruth!’, Ty ne prav ‘You are wrong’, Da net ’emphatically no!’), and the high frequency in Russian of words expressing extreme moral evaluation, such as podlec, negodjaj, merzavec ‘scoundrel, base person’ and blagorodnyj ‘noble, lofty’.

    Societies are of course heterogeneous, and not every member of Anglo and Russian cultures would accept or endorse scripts A and B, respectively. However, the claim is that even those who do not personally identify with the content of such scripts are nonetheless familiar with them, i.e. that they form part of the interpretative backdrop to discourse and social behaviour in their own particular cultural contexts.

    Just as these two cultural scripts can be portrayed in English NSM, so they can equally well be portrayed in Russian NSM, as in A’ and B’ below. The scripts will therefore “work” equally well on both sides of the cultural fence: to help English speakers articulate their own Anglo cultural attitudes and to recognise how these differ from characteristic Russian attitudes; and equally, to help Russians articulate Russian cultural attitudes and to recognise how they differ from Anglo attitudes. This can be done in language which is culturally neutral.

    A’: an Anglo cultural script connected with “personal autonomy”

    (ljudi dumajut tak:)
    kogda čelovek čto-to delaet, xorošo, esli ėtot čelovek možet dumat’ tak:
    “ja delaju ėto potomu, čto ja xoču ėto delat’”

    B’: a Russian cultural script connected with “expressiveness”

    (ljudi dumajut tak:)
    xorošo, esli čelovek xočet, čtoby drugie ljudi znali, čto ėtot čelovek dumaet
    xorošo, esli čelovek xočet, čtoby drugie ljudi znali, čto ėtot čelovek čuvstvuet


    Reading these scripts from other cultures for the first time, speakers of English and Russian often experience a sense of strangeness or unfamiliarity, because the ideas and attitudes being expressed are culturally strange. Nonetheless the scripts themselves are clearly intelligible and highly explicit. They articulate the culturally strange in terms of the linguistically familiar, i.e. in terms of simple common words whose meanings are shared between the languages concerned. This is in stark contrast to culture-specific words such as freedom [and] iskrennost’ […]. Needless to say, neither do cultural scripts have recourse to any value-laden language-specific terms of the sort which are commonly used in everyday talk about the impressions conveyed by foreign ways of speaking; for example, that Russians are often seen as intense (from an Anglo perspective), and that Anglos can be seen as xolodnyj (roughly) ‘cold’ and bezdučnyj (roughly) ‘soulless, heartless’ from a Russian perspective […].

  62. @John Cowan

    Do these concepts correlate well with bureaucratic language or public signs? E.g., “It is prohibited to…” or “No loitering”. Unlike everyday speech, these form a relatively closed set that could be objectively sampled.

    Japanese is also highly indirect in the way that it expresses requests. It goes even further than English, however, in that Japanese also tends to shy away from making direct assertions. Hence the preference for expressions like ‘it is thought that’ (と思われる).

    A very long time ago I wrote a piece on Mind the Gap, in which I noted differences between English and Japanese requirements on how to express the instruction to ‘mind the gap’. Writing instructions in Japanese requires more than a little thought in order to get it right. I don’t know how well this kind of phenomenon fits into NSM — theoretically it should but it looks to me like a bit of a stretch.

  63. That “cultural scripts” passage is interesting and seems convincing; I note a minor error: “bezdučnyj” should read “bezdušnyj” (without a duša ‘soul’; there is no word or morpheme “duč“). If it had been JC’s error, I would have silently corrected it, but my copyeditor’s oath forbids me to change a quote.

  64. The gap is extraordinarily narrow (two inches or so) in almost all NYC subway stations, so no warning is needed or given, as even a toddler’s foot will easily span it. At 14th St. / Union Square on the 4, 5, and 6 lines, however, the downtown platforms are sharply curved and the trains (which consist of eight cars) necessarily gape away from them. In most places along the platforms, there are gap fillers that move out to permit walking right up to the train. But because these do not handle all doors, an announcement is also made both on the train and on the platform: “Please be careful of the gap between the platform and the train”.

    Those who have read Bathrobe’s excellent article will see that this is very nearly Japanese in its explicitness: we New Yorkers have some regard for Allegheny bumpkins, as collectively they constitute our second-largest industry. However, it is not given in the most polite possible form, because please plus the imperative is not typically polite in American English. Indeed, it may be less polite than the plain imperative; I suspect this has to do with our positive-politeness and solidarity-based culture.

  65. my copyeditor’s oath forbids me to change a quote.

    Hah! I once made up a quote (in English) for Mikhail Gorbachev; it caught the drift of what I was told he said (in Russian) at a certain event. The quotation was placed in a news release we were convinced would not be picked up by any important media outlet, as he was no longer in power. The release was not widely distributed, and to the best of my knowledge was indeed not picked up.

    In any case, when writing an article for publication based on an actual interview, one pretty much must polish whatever the interviewee said, if only because almost nobody speaks in “written text.” Back when, once a month for a number of years I interviewed a different executive in a major downtown Toronto office complex. I made handwritten notes only. Interviewees always reviewed and approved the interview prior to publication in the complex’s ‘newspaper.’ I observed that the higher the snack bracket of the interviewee, the less he or she tended to change my text. Crowning moment: A former premier of Ontario changed three words in a 1,900-word interview.

  66. The one time I was interviewed (for a company newsletter) I was represented as speaking such gibberish (bits and fragments of a whole talk that the person I was speaking to had caught very imperfectly) that I offered to rewrite it wholesale for clarity and sense, an offer the reporter/editor was glad to accept.

  67. In any case, when writing an article for publication based on an actual interview, one pretty much must polish whatever the interviewee said

    Well, by “one pretty much must” you mean “it is the longstanding practice of the profession to.” This is a vexed topic which has been the subject of much (largely hostile) commentary at Language Log; this post has good discussion as well as links to a couple of other posts. My take: there is some justification for some journalistic distortion, but the profession tends to run with that and use it as wholesale whitewash material. (Cf. cops and violence.) Ink-stained wretches could do much better at this aspect of their job.

  68. J. W. Brewer says

    http://www.newyorker.com/humor/daily-shouts/stand-clear-of-the-closing-doors is an attempt at riffing (including more detailed explanation for the benefit of tourists or other clueless riders) of THE stock imperative phrase one hears riding the NYC subway. The humor (if one thinks the piece succeeds as humor, which is not clear to me . . .) is precisely the incongruity that comes from doing variations on a uniform stock phrase that in the real world is delivered invariantly. (In the not-too-old days one could at least appreciate the variousness of accent, intonation, etc. by the varied MTA employees delivering the stock phrase, but now it’s generally an automated robot voice, so you don’t even get that pleasure.)

  69. J. W. Brewer says

    One datapoint on Br-v.-Am usage of “please.” Some years ago I would occasionally go to services at a particular church where the priest was English and had not attempted to Americanize his language use (if anything perhaps the other way in order to take advantage of Anglophiliac congregants?) despite considerable years living in the U.S. At points in the service where a default/median American clergyman would say to the congregation e.g. “please be seated” (and omitting the please in that particular context would be odd to my AmEng ear) he would say “will you please sit down,” which to my AmEng ear sounded a lot bossier/impolite, but I (politely) ascribed not to actual bossiness but to him not talking proper what with being foreign and all. I’m not sure how much of the effect was the active “sit down” v. passive “be seated” versus the superfluous “will you.”

  70. Paul, “snack bracket” is new to me. I can see what you mean from the context (something like “pecking order”, or “pay grade”?), but where does it come from? The internet is letting me down. I’m getting an Urban Dictionary definition to do with acceptable age-ranges for dating and a bunch of graphics apparently playing with tournament knock-out diagrams (or whatever they’re called). Needless to say, I’ve limited my research to the first page of hits, and maybe not all of that.

  71. I’ve used the term ‘snack bracket’ for many years. Yes, it’s like ‘pecking order.’ I don’t know its origin or history, and I don’t recall hearing or seeing it in a sports or dating context.

  72. My take: there is some justification for some journalistic distortion, but the profession tends to run with that and use it as wholesale whitewash material. (Cf. cops and violence.) Ink-stained wretches could do much better at this aspect of their job.

    For sure, for sure.

  73. The Chicago el train announcements cue up the automated “doors closing” soundbite as soon as the doors open, (This is Monroe. Doors closing.) which a) frightens tourists and b) means that the drivers have to verbally announce when the doors actually are going to close. They don’t stick quite as strictly any script, but it’s pretty close. (I assume there’s a reason for this, because it never happens any other way, but what that reason is…)

  74. This use of snack bracket clearly shows that it means ‘level of income, tax bracket’.

  75. David Marjanović says

    On please: see also the American why don’t you, which is an order.

  76. Rather, it is an order if it comes from a legitimate giver of orders. Otherwise, it is a recommendation: “Why don’t you leave your husband?” is not an order even from the most thuggish of boyfriends.

    Last summer I had a long conversation with a friend of a friend, who said that her greatest personal struggle was over the issue of obedience to her husband. She wasn’t a Third World immigrant or a survivor of the 18C, it just felt like it. My (cautious, tactful) efforts to persuade her that her self-condemnation resulted from being sold a crock of shit were utterly unavailing.

  77. Nohara Shigeru says

    My recent experience in Japan was full of signs on the formula “Let’s make/do/go …”. I thought this might be school English for some sort of polite Japanese imperative, but it could also be a way to avoid English terseness in English.

  78. A rather late rejoinder:

    how is that an improvement over just figuring out how to explain it (kuyashii) in terms an English-speaker can understand such as “go through a failure or shame, and can’t give up or forget it”

    Because that totally fails to capture the feeling of the word kuyashii. It’s typically used to express something like “Damn, that’s so maddening! I/we could have done it but we missed it! Damn, damn, damn!” (Possibly an over-narrow characterisation but you get the picture.) That’s not what “go through a failure or shame, and can’t give up or forget it” tells you at all.

  79. Thanks, that’s helpful.

  80. Revisiting this thread:

    Chomsky’s claim that Roger Bacon, Port Royal etc. are his intellectual forebears is self-serving nonsense. He is basically reacting against structuralist linguistics, which attempted to shake free of the narrow linguistic base of traditional grammar and clear the decks by considering “just the facts, ma’am”, justifying its analyses in terms of the objective distribution of grammatical forms (morphemes) rather than an appeal to abstruse philosophising. Yes, the adoption of behaviourist assumptions became a self-imposed straitjacket, but that does not make their analyses “wrong”. Limited, maybe, because limiting themselves to objective, palpable patterns was their whole point, but not “wrong”.

    Chomsky’s appeal to an innate human ability to learn language underlain by “universal grammar” is simply a cover for his rampant adoption of formalisms like X-bar theory (since abandoned, like so much else) or “Merge”. I am doubtful that Chomsky’s crazy formalisms that supposedly describe what goes on “in the mind” are really a good basis for any kind of “universal grammar”. Chomksy is engaged in bold bullshitting when he tries to characterise his opponents as dealing in trivialities (“butterfly collecting”) in comparison with his own rich and deep analyses.

    This is pretty much the conclusion and summation of my own view of Chomsky.


  1. […] exactly is Universal Grammar, and has anyone seen […]

Speak Your Mind