LANGUAGE STRUCTURE IS NOT INNATE.

Or such is the finding of a group of researchers at the Max Planck Institute for Psycholinguistics, Michael Dunn, Simon J. Greenhill, Stephen C. Levinson, and Russell D. Gray, whose paper “Evolved structure of language shows lineage-specific trends in word-order universals” was published online by Nature a few days ago. (The abstract is here, where there is also a link to a downloadable pdf of the full paper.) Russell Gray has written a nice, clear explanation of the study and its results, and I will quote the conclusion:

These family-specific linkages suggest that language structure is not set by innate features of the cognitive language parser (as suggested by the generativists), or by some over-riding concern to “harmonize” word-order (as suggested by the statistical universalists). Instead language structure evolves by exploring alternative ways to construct coherent language systems. Languages are instead the product of cultural evolution, canalized by the systems that have evolved during diversification, so that future states lie in an evolutionary landscape with channels and basins of attraction that are specific to linguistic lineages.
One of the main implications here is that to really understand how languages have evolved, we need to understand the range of diversity in human languages. With one language on average going extinct every two weeks, the ability to understand this is rapidly being lost.

There is an ongoing discussion on Mark Liberman’s post at the Log; I hope the conclusions of the study hold up, because 1) it shows the importance of studying as many languages as possible, and 2) it’s a poke in the eye for Chomsky and his stupid theory of universals, which implies that there’s no need studying any language but your own because they’re all basically the same anyway.

Comments

  1. +1 on Chomsky.

  2. English, Italian, and Hebrew, actually.
    But to be fair, Chomskyans have long since got past that; your complaint is out of date.

  3. What John said: the ‘Universal Grammar’ is no longer, well, universal grammar, it is – and has always been – the universal human propensity for language. Add an extra ‘ugh’ for the khutzpah.
    “Languages are instead the product of cultural evolution”
    Thomason and Kaufman: “The key to our approach — and the single point on which we stand opposed to most structuralists (including generativists) who have studied these issues — is our conviction that the history of a language is a function of the history of its speakers, and not an independent phenomenon that can be thoroughly studied without reference to the social context in which it is embedded.”
    Word.

  4. I’m biased, since I’m one of the people working on the stupid theory of universals (STU, for short), but I think that Tunji C. and Olaf Koeneman, over at Language Log, are right: the study is going after a straw man. We’ve known for decades now that the Head Direction Parameter, which the study attacks, has to either be weakened to deal with languages in which head direction isn’t completely consistent, or supplemented with other principles that would alter an underlyingly homogeneous head direction. In neither case does the mature theory predict anything at all, as far as I can see, about the direction of language change; all the languages the authors study are compatible with Universal Grammar, and the authors’ conclusion amounts to a discovery that the properties of a given language at a particular time have a lot to do with the properties of that language at a previous time. No big surprise.

  5. Eman Nona says:

    *Yawn* wake me when the anti-generative dogma abates. The rest of us moved past this in the ’80s/’90s.

  6. I have no interest in trying to draw you into a debate on the merits of the STU, but I must at least protest that it most certainly does not imply “that there’s no need studying any language but your own because they’re all basically the same anyway.” I cannot say that there have never been any generativists who have acted as though it implied that, but I think that to most of us it’s perfectly obvious that if one adopts some form of Universal Grammar as a hypothesis, then it is necessary to test this hypothesis against as many languages as possible, in order to see what properties of language really are universal and what varies.
    What the STU does imply, and I hope that you will agree that this is at least different from what you said it implies, even if you don’t happen to think that it’s any better, includes the following:
    1. It can be worthwhile to spend an entire career studying a single language (even one’s) own in great depth, at least as long as there are also people studying other languages, and some people comparing the results of everyone else’s research.
    2. We might be able to learn something about, say, Icelandic by studying, say, Japanese—but we will have absolutely no idea what it is we’ve learned about Icelandic from Japanese unless and until we study Icelandic, too.

  7. Sorry, I seem to have misplaced a close-parenthesis in the preceding. I meant to say “(even one’s own),” of course, not “(even one’s) own.”

  8. I’ve been reading about these UG controversies off and on since the sixties. As a non-linguist, I now switch directly into program-notes mode when the subject crops up. “Program-notes mode” is what I call the reaction of a concert hall spectator who finds the music repetitive or unintelligible – he starts riffling through the program notes, examining the acoustic tiles on the ceiling etc., looking for something interesting in the presentation context.
    Have there been “science studies”, or history-of-ideas studies, of how the UG discussions themselves work ? There are clearly regularities in these controversies. At a very high level of abstraction I see a confusing of description and explanation, for instance. Has anyone tried to identify a universal grammar of language investigators (UGLI) ?

  9. But to be fair, Chomskyans have long since got past that
    I have no interest in being fair; those swine made me repeat a class in their gibberish before I could take my finals in grad school, and I bear the scars to this day. We hates it!
    *Yawn* wake me when the anti-generative dogma abates. The rest of us moved past this in the ’80s/’90s.
    See above. Not having continued in the field, I have had neither the opportunity nor the need to move past it. I’m a fair man, and if I see a study by non-Chomskyans supporting one of their dogmas in a convincing way I’ll grit my teeth and accept it, but I’ll always be filled with delight to see studies disproving one of their dogmas, even if they now find it embarrassing and try to bury it deep in the litterbox.
    (No offense, Norvin, you’re obviously a good sort with a sense of humor, and I’m sure you’ve got your own hobbyhorses and bugbears. Thank you for tolerating mine!)

  10. I’m going to send a combined generative-prescriptivist grammarian over as soon as I can find one, just to see what happens.

  11. I’m going to send a combined generative-prescriptivist grammarian over as soon as I can find one, just to see what happens.

  12. J.W. Brewer says:

    But you are perhaps missing the key point that the Church o’ Chomsky has refined its dogmas to the point of unfalsifiability, by (as explained in the Log thread, if I’m following it correctly) saying that the parameters interact in such complex ways that no testable predictions can be generated by assuming their existence. And I must say that the phrase “channels and basins of attraction” from the block quote from this paper has at least as much of an epicycles/phlogiston vibe as the Chomskyan “parameters” and/or the whole unverifiable just-so-story of “deep structure.” I can sort of spot the metaphor, but what’s the water, where does the gravity come from, and how do we tell which way is downhill?

  13. the Church o’ Chomsky has refined its dogmas to the point of unfalsifiability

    I thought something smelt familiar, yes.
    Well, phlogiston and epicycles were both falsifiable. And more importantly, they worked in their fields for quite a while. Much like the Julian calendar.
    What has Chomsky produced?

  14. > (No offense, Norvin, you’re obviously a good sort with a
    > sense of humor, and I’m sure you’ve got your own
    > hobbyhorses and bugbears. Thank you for tolerating
    > mine!)
    None taken! And you’re welcome. It’s your blog, after all, and it’s a blog I enjoy very much.
    I can see how the Language Log discussion could have led people to believe that Chomskyan theory about word order isn’t falsifiable, but that just isn’t true…for instance, one current proposal which I personally find attractive predicts that there can be languages in which the verb follows the object and the auxiliary precedes the verb phrase, but no languages of the opposite kind, in which the verb precedes the object and the auxiliary follows the verb phrase.
    More generally, the point (which Liberman sort of hints at) is that the annoyed noises you’re hearing now from theoretical syntacticians are not unlike the reaction you get from biologists when the media publish stories about the ‘gene for homosexuality’, or the ‘height gene’, or whatever. Word order is the result of an interaction between a variety of principles that interact in complex ways; if you consider a single principle in isolation, of course it doesn’t do anything. This doesn’t make the theory unfalsifiable, but falsifying it certainly isn’t as easy as Dunn et al seem to think.

  15. I can see how the Language Log discussion could have led people to believe that Chomskyan theory about word order isn’t falsifiable, but that just isn’t true…for instance, one current proposal which I personally find attractive predicts that there can be languages in which the verb follows the object and the auxiliary precedes the verb phrase, but no languages of the opposite kind, in which the verb precedes the object and the auxiliary follows the verb phrase.
    What do you mean by saying that “there can [not] be” languages of the second kind ? The very description (“in which the verb precedes the object and the auxiliary follows the verb phrase”) characterizes such a language, showing that it is conceivable, even though possibly non-existent at the present time. It would thus be less contentious to claim “there are no such languages”, instead of “there can be no such languages”. That at least would be falsifiable as an empirical claim, although it would say zilch about the status of the theory.
    Or are you saying after all that such a language can not, and will not, “naturally” occur ? What would an experimental set-up look like that could falsify that claim: waiting and watching ? Given that such a language can easily be constructed and used, that claim is revealed as a piece of hapless historicism.
    Or are you claiming that such a language would be unworkable / unintelligible ? Apart from the fact that that would be a claim not about structure, but about semantic or psychological sustainability: what would an experimental set-up look like that could falsify that claim ?
    I suggested above that UG controversies involve confusing description and explanation. I think this is due to the central role that mathematics is made to play there. Mathematics allows you to set up descriptions (formulas) and connect them by explanations (proofs showing how to get from one formula to another), all without leaving your armchair.
    Mathematics is still only a handmaiden of the sciences. In today’s licentious times, though, too many people spend all their time chasing the hired help.

  16. “Historicism” in the sense that Popper used the word.

  17. I meant that there are no such languages, and never have been (and there are several cases of languages which have fairly free word order but avoid that particular order). The claim would be falsified by finding a language with that word order.

  18. Norvin & Stu: There are no languages without verbs, with just affricates and diphthongs, no consonants, without monosyllabic words, without onsets in syllables, no intonation, without phrases, or sentences in which components of phrases are scattered freely within a sentence, etc., etc.

  19. KasparsM says:

    >> and there are several cases of languages which have fairly free word order but avoid that particular order
    Actually only uses them less sparingly but there is nothing in particular that makes it ungrammatical or undesirable. In fact, OVS is a proper grammar in certain cases. A classic textbook example in Russian is (1) Девочка зашла в комнату – girl (S) entered (V) into room (O) (2) В комнату зашла девочка – into room (O) entered (V) girl (S). The approximate translations are (1) The girl entered the room. and (2) A girl entered the room. In other words OVS makes for emphasis that in English is realized by the use of articles.
    According to this study the fact that SVO is still the predominant word order in Russian appears to be a historic and cultural artifact. Similarly to using base 10 numbers and base 60 for measuring time due to historic reasons only.

  20. Are there any languages with obligatory OVS morphology (it-hit-she)?

  21. marie-lucie says:

    one current proposal … predicts that there can be languages in which the verb follows the object and the auxiliary precedes the verb phrase, but no languages of the opposite kind, in which the verb precedes the object and the auxiliary follows the verb phrase.
    The point is not the basic word order (eg SVO or OVS or other possible orders) but the combination of those orders with the place of the Auxiliary (here “a”) relative to the Verb Phrase (VO or OV). So SaOV is theoretically possible in a language with SOV basic order (such as Japanese), but SVOa is not possible for an SVO language (such as English), which has to have SaVO. (I make no claim about the correctness of the theory on this point, or what it would predict for VSO).

  22. marie-lucie says:

    Are there any languages with obligatory OVS morphology (it-hit-she)?
    I am not sure, but you have to differentiate between morphology (the form of the verb and nouns) and syntax (the word order), and also whether the O and S are nouns or pronouns. For instance, in French, Italian, Spanish, etc the order for Nouns is SVO, but the order for Pronouns is SOV (which preserves the basic Latin order): Le chat mange les souris : Il les mange. Adding an auxiliary provides a test of the rule above: Le chat a mangé les souris (SaVO) : Il les a mangées (SOaV) (Note that SOaV appears to be theoretically impossible since it breaks up the Verb Phrase! the way to get around it is to say that the Object pronoun has been moved, something which contradicts actual historical evolution).

  23. marie-lucie: My question is about a morphology in which there is an object prefix-stem-subject suffix.

  24. marie-lucie says:

    GW, yes, I got a bit off-track there, but SVO, etc usually refers to syntax, with S and O being nouns.
    I don’t have an answer to your question (you would have to consult a work on typology). I know a case of object-subject-stem though (Molale).

  25. aquilluqaaq says:

    Are there any languages with obligatory OVS morphology (it-hit-she)?
    My question is about a morphology in which there is an object prefix-stem-subject suffix.
    Depending on the scope of ‘obligatory’ you have in mind here, I can certainly think of a language with obligatory OVS morphology under certain conditions, such that OVS is obligatory for bipersonal verbs with a 1st person singular object and a 2nd person plural subject: me–hit–you, i.e. ‘you (pl.) hit me’, but, by contrast, VOS is obligatory with a 1st person plural object and a 2nd person plural subject: hit–us–you, i.e. ‘you (pl.) hit us’. But perhaps that’s not what you’re asking…

  26. aquilluqaaq: Good point about ‘obligatory.’ Let me rephrase.
    Is there a language in which the morphology allows O-V-S (me-hit-she meaning ‘she hit me)? I am talking about word structure, not sentence structure.

  27. aquilluqaaq says:

    Yes, O–V–S morphology is possible, though, in the case of this language, not for the particular example you give:
    ‘She hit me’ (3rd person singular subject / 1st person singular object): /ena–ɹkəpɬə–ɣʔe/ = O–V(–ASP)
    ‘She hit him’ (3rd person singular subject / 3rd person singular object): /kəpɬə–nen/ = V–O
    ‘You (pl.) hit us’ (2nd person plural subject / 1st person plural object): /kəpɬə–tko–tək/ = V–O–S
    ‘You (pl.) hit me’ (2nd person plural subject / 1st person singular object): /ena–ɹkəpɬə–tək/ = O–V–S

  28. aquilluqaaq: Interesting. What language is this?

  29. aquilluqaaq says:
  30. Are there any languages with obligatory OVS morphology (it-hit-she)?
    Klingon.

  31. Hixkayana.
    Old Irish had this structure internal to the verb – “romthuargit” = “they beat (PST) me”. the ‘m’ is the DO marker.

  32. Sorry.
    And the “it” marks the subject.

  33. What is the unmarked sentence word order in these languages with OVS morphology?

  34. marie-lucie says:

    Complex cases like the Chukchi examples cannot be reduced simply to the general order of S and O with respect to the verb. I don’t know Chukchi but the varied order of the personal morphemes may have to do with the likelihood of agentivity of the various “persons” in different Agent/Patient combinations: “I killed him” is obviously more likely to be uttered than “He killed me”. There are also languages with unanalyzable morphemes which do not mean, for instance, “I” or “me” or “you” but “I-Subject+you-Object” or vice-versa. Such semantically complex morphemes must derive from former combinations of pronouns which became fossilized and in which the identity of the original pronouns has been obscured by phonological change.
    On the other hand, the order of nouns is more likely to be affected by pragmatic factors, such as which one is most salient in a sentence (as in the Russian examples above), and syntactic change is often the result of an originally flexible order becoming more rigid when one of two or more possible word orders becomes the default option, then the only option. This is illustrated in the transition from Latin to the Romance languages, where Latin SOV which was one of several possible orders gave way to Romance SVO as concerns nouns, while preserving SOV for the pronouns. (This description is somewhat simplified: it is valid overall, but there are a few pragmatically-based exceptions).

  35. aquilluqaaq says:

    Complex cases like the Chukchi examples cannot be reduced simply to the general order of S and O with respect to the verb.
    Absolutely. Though, while irreducibly complex, Chukchi (in this case, bipersonal) verb forms do exhibit regularities:
    [1] P(sub) – [2] T/M – [3] P(obj1) – [4] R – [5] P(obj2) – [6] A – [7] P(obj3) – [8] P/N(sub/obj)
    In the (generally) TAM-marked verb-forms, the subject, where marked, is typically initial, and final P/N(sub) is confined to the 2nd pl (and 1st sg/pl in the imperative and conjunctive); only the 1st sg appears in P(obj1) and only the 1st pl in P(obj2). All forms final P/N(sub)-marked are also object-marked P(obj1) or P(obj2).
    The predicative forms are final P/N(sub)-marked in all but the 3rd sg/pl, which are final P/N(obj)-marked and either P(obj1)- or P(obj2)-marked, or show predicative root-affixes.
    My sense, then, is that the variable positions of P/N(sub/obj) markers are dependent on sub/obj combinations, but for reasons of P/N, rather than likely frequency of agent/patient relations.

  36. marie-lucie says:

    aquilluqaaq: I suppose that P/N means ‘person/number’, but I am not sure about R.
    I hypothesized agent/patient relations as a factor because they do play a part in some pronominal systems, but I don’t know Chukchi and you do.

  37. aquilluqaaq says:

    marie-lucie: sorry, yes – P (person), T (tense), M (mood), R (root), A (aspect), N (number).

  38. marie-lucie says:

    I see! I did not realize that you were listing the order of morphemes in a potential verb form.

Speak Your Mind

*