I don’t spend much time on Chomsky, Pinker, and the legions of evolutionary psychologists who claim language is innate (“nativists”), but Scott Martens over at Pedantry does, and today he provides a nice long assault on their line of thinking; I’ll reproduce here the heart of the portion about language:

However, the adaptationist/nativist program has offered no insight into the neurobiology of language. It has not managed to identify genes or biochemical mechanisms that underlie language. It has not offered any useful knowledge to translators, lexicographers, language educators or people interested in natural language processing. All the progress made in those fields have come from people whose work either has no bearing on the central theses of language nativists or denies at least some nativist claims outright.
And, people outside the nativist community largely believe language is a lot like walking. Given the physiology of our senses, the structure of our mouths, throats and lungs, the properties of the human nervous system in general, and the structure of the environment in which children are immersed, language is simply the optimal solution to the problem of communication, and virtually every infant discovers it unless they are prevented from doing so by some serious physical condition. If this is the case, then there is nothing in the genome at all that specifies linguistic behaviour, or even has any sort of direct influence on it except in the trivial case where it interferes with the form of our bodies. It makes no sense to talk about innate linguistic knowledge or a language “instinct.”
This programme has—over the last decade especially—shed a lot of light on language. Every meaningful advance in natural language processing since 1980 has come from some kind of interactionist or empirical theoretical base, not from attempts to uncover innate knowledge about language. All meaningful work in lexicography, translation studies and language education has been predicated on the idea that language is firmly grounded in time and place and that it is part of the social structure of the culture in which language is spoken. No linguistic universal has ever been found other than those trivially associated with the physical restrictions of bodies and limited working memory.

Bravo, sir—I doff my headgear in your general direction.


  1. I’m agnostic on this question; haven’t made up my mind. Is walking instinctual? One assumes that the way a tiger or lion walks is instinctual, why should the way a human walks not be instinctual?

    I guess where I dissent from the Pinker line is the positing of a specific language “device” or “homunculus.” Where the hell is the lnaguage gene itself? And the fact that the hypothesis seems barren in results, doesn’t teach us about language, if what this quote says is correct about this.

  2. No linguistic universal has ever been found other than those trivially associated with the physical restrictions of bodies and limited working memory.

    That person obviously has *lots* of experience of working with children with specific language difficulties 😉

    It’s the exceptions rather than the usual successes of ‘normal’ language acquisition that give us insights.

    language is simply the optimal solution to the problem of communication, and virtually every infant discovers it unless they are prevented from doing so by some serious physical condition.

    *Physical* condition?
    Oh dear.
    I sense a certain lack of understanding…

  3. Hat, I’d really recommend reading The Language Instinct, even if you’re sure you’d disagree with it. Maybe he’s full of it, and I just didn’t catch it, but he presented what seemed to me to be fairly persuasive evidence of a kind of decision tree in languages, wherein a language with characteristic A often but not always has characteristic F, but never has characteristic G, and that sort of thing.

    I won’t try and reproduce the chapter in question and almost certainly can’t do it justice, but I do highly recommend it, and I guarantee that once you read it, it will take more than “Oh yeah, well then where’s the gene?” to refute it.

  4. Blue Witch, actually, yes I do have experience with speech therapy, as well as several family members with various histories of speech and reading trouble. Can you name a single case in which a language acquisition deficit is not accompanied by evidence of other problems? In short, do have any reason to believe there are specific language acquistion deficiencies that don’t follow from some more generalised problem, like hearing imparements, neurological problems or more generalised learning disabilities? (NB – stuttering is not an impediment to language acquisition, it is an unexplained performance problem which is most likely related to motor planning.)

    As I understand it, there is no unambiguous instance of such a condition, and the two ambiguous cases I know about are both hotly contested.

    Mike, I’m pretty sure that each instance in The Language Instinct where Pinker claims that a language with charactersitic A also has characteristic B and never characteristic C, there is at least one exception in the literature. I don’t think Pinker is still using the “parameter setting” model he advances in that book. One reason is because it does a terrible job explaining the relative freedom of word order in many languages. This is a problem for all approaches that don’t use at least partially lexically driven models of syntax.

    Furthermore, as many early critics of the parameters model claimed, the constraint of having to minimise ambiguity in language can just as easily account for these tendencies. This is assumed in all constraint satisfaction models I can think of, like optimality theory or more infomration theoric approaches.

  5. The concept of Specific Language Impairment (SLI):

    Definintion (Bishop, 1992) “SLI is diagnosed where there is a failure of normal language development that cannot be explained in terms of mental or physical handicap, hearing loss, emotional disorder or environmental deprivation.”

    Sorry Scott, I didn’t mean to be offensive before, I’ve just re-read my comment and it didn’t come out how I intended.

    I *do* have *lots* of examples of the sort you claim don’t exist.
    I work with them (assessing their specific special eduational needs) all the time.
    I have developed and set up educational provision to meet the needs of these children, and inspected and reported on provision that specialist establishements make in an attempt to meet the particular needs of this population.

    Every professional Speech and *Language* therapist, specialist paediatrician, specialist teacher would also be able to cite you multiple examples.

    The AFASIC website would be a good starting place for your further investigations.

  6. Mike, as Scott says, the problem is that there are unacknowledged exceptions to all these “rules.” That’s why transformational grammar, so simple and lovely when it was invented, got larded with more and more Ptolemaic epicycles — Chomsky kept having to take account of inconvenient exceptions that people kept bringing up just when he thought he’d got everything covered. People love Pinker because he writes well and seems so reasonable — as long as you don’t have access to the counterexamples.

  7. I have just realised the problem here: the essential element to this debate is an understnding of the difference between language *delays* and language *disorders*. The former are (generally) accompanied by other problems, the latter (generally) occur in the absence of any other contributory factors.

    Study of the language development of children on the autistic continuum is particulalry interesting.

    Hope this makes it clearer.

  8. Blue Witch, I think I see where you’re going with this and where I ended up going a different way.

    There are conditions that manifest themselves most obviously in language problems. My point, rather, was that we don’t need to evoke any sort of “language organ” to explain irregularities in language acquisition, we need merely point to the nevous system and say something is wrong with it.

    Think of this by analogy with dyslexia. Dyslexia most often manifests itself as difficulty learning to read. Yet, we don’t claim there is a mental “reading organ” that is provided by evolution. Dyslexia is widely believed to be a more generalised condition whose most immediate effect is on those brain facilities used in reading, but that doesn’t mean that those faculties are specialised for reading alone.

    I read a lot of people pushing the idea that many linguistic imparements may have to do with problems in either speech planning or in maintaining a memory buffer for sounds. My contention is that nothing here requires a specialised inborn language faculty to explain it, and invoking such a faculty has not proven very useful in understadning or treating language impairments.

    I am a little behind the literature, having moved over a few years ago to NLP rather than speech pathology or child development, however, my understanding is that the rejection of the syntax/semantics separation, rejecting the competence/performance distinction – or at least sharply weakening it – and rejecting strict modularity has provided some useful work in understadning many of these problems. I’m thinking specifically of Sue Gathercole’s work identifying SLI with a more general kind of memory problem. I think her book is called “Working Memory and Language.”

    I was thinking of the current controversy over the FOX2P gene in saying that there was no unambiguous instance of an imaprement specific to language. Clearly you weren’t.

  9. Scott – I had just come to the same conclusion myself!

    I should have read your full article before picking on bits from the truncated version LH posted.
    I seem to have fallen into the trap of those journalists/commentators that I abhor most!

    Re the dyslexia debate – as you’re obviously interested, have you come across the Jonathan Solity (Univ of Warwick) recent work?

    I won’t start baiting you on the subject of NLP though. I think that area is more of a minefield than any other I know, including “dyslexia.”

  10. This is the kind of thing that makes me glad I started a blog. I feel like a spectator at Wimbledon, watching the ball shoot back and forth and admiring the players. If I pay attention, I may actually learn something!

  11. I was under the impression that chimpanzees provided a counterexample to the arguments you’re presenting – about as intelligent as a young child in general terms, and with similar physiology but not capable of language.

  12. Blue witch – once again just to make sure we’re on the same page – by NLP I mean “natural language processing”, getting computers to deal with language, not “neuro-linguistic programming” which is something that seems to belong to the world of infomercials. I agree that natural language processing can be a minefield, but as an engineering matter rather than a part of cognitive science per se it doesn’t quite have the same sort of controversies. I work on a machine translation system that isn’t wholly useless, but I have no pretention that this system is especially like people in the way it works.

    I am somewhat behind on the literature – if for no other reason that that there is never time for everything. I remember Jonathan Solity being involved in cirriculum battles in the UK, but that’s about all I remember. I guess he goes on the reading list too now.

  13. Scott – I had no idea NLP had more than one meaning. Is there a short “idiots guide” to your “natural language processing” area that I could read?

    And yes, I agree with you about NLP as “infomercial” (what a great word!). Well, it wasn’t initially, when it came out of research and into practitioner literature (I guess about 20 years ago now), but sadly it is now a bandwagon, and of the worst kind in many cases.

  14. Yes, the two NLPs are pretty different. Neuro-linguistic programming seems to have some underlying legitimacy, despite having been largely appropriated by the dianetics/scientologist glove-puppets.

    Despite his reasonable tone of voice, I’m a little suspicious of Pinker, I must say. On the passing topic of chimpanzees, I understood that
    a] something about their mouths makes complex vocalisation difficult for them?
    b] the relative simplicity of courtship in a gang-bang-style species [sex is plentiful, and sperm from multiple partners compete inside female chimpanzee bodies, obviating the need for males and females to compete for each other in the social world with the same intricacy we do] makes the present chimpanzee social world, regardless of Franz de Waal’s claims for its sophistication, fundamentally simpler than that of our ancestors, and thus not sufficiently conducive to the emergence of early language.

    I’m getting increasingly sceptical about both the language-organ idea, and about the idea that language processing can be understood as an abstract, computational activity divorced from physical facts about everyday life.

    I think if Chomsky, Pinker, and Dennett were right [I know their positions are not identical, but I’m discussing the overlap] we’d have very high quality translation software by now. But we don’t, and it’s increasingly clear we’re not going to.

  15. Blue Witch, no, there is no “idiot’s guide” to natural language processing. The introductory texts tend to assume you know a fair bit of computer science. If you do, they’re not terribly difficult texts. James Allen’s “Natural Language Understanding” is one of the key basic texts.

    The problem is that there is no overall “theory” of NLP, just a collection of different ideas that have worked at least to some degree, for somebody, at some time, on some problem. It’s a collection of techniques rather than a real discipline.

    However, NLP is – I suspect – the major motivating force nowadays in structural linguistics. We can’t say anything with confidence anymore about how brains structure language, but we can still devise structures for computers to do so. Ivan Sag’s book on HPSG comes close to admitting exactly that.

    Mark, although Chomsky, Pinker et al haven’t made terribly grand claims for machine translation, I tend to agree with you that if things worked their way it would be a lot easier to do. Dick Hudson has written several books on why a language organ divorced from the rest of cognition and knowlege doesn’t make any sense. “Word Grammar” is the one that comes to mind first, and “Against Transformational Grammar.”

  16. Thanks, Scott. I’ll definitely look for Dick Hudson.

Speak Your Mind