Why Chomsky doesn’t count as a gifted linguist

Send to Kindle

Somebody commented on the Language Log saying “of course [...] Chomsky was a massively gifted linguist” http://j.mp/9Q98Bx and for some reason, to use a Czech idiom, the handle of the jar repeatedly used to fetch water just fell off. Meaning, I’ve had enough.

I think we should stop thinking of Chomsky as a gifted linguist. He was certainly a gifted mathematician and logician still is a gifted orator and analyst of political discourse (sometimes putting professionals in this area to shame). But I honestly cannot think of a single insight he’s had about how language works as language. His main contribution to the study of language (his only one really) was a description of how certain combinatorial properties of English syntax can be modeled using a particular formal system.  This was a valuable insight but as has been repeatedly documented (e.g. Newmeyer 1986) its runaway success was due to a particular historical context and was later fed by the political prominence of its originator. Unfortunately, everything that followed was predicated on the model being isomorphic with the thing modeled. Meaning all subsequent insights of Chomsky and his followers were confined to refining the model in response to what other people knew about language and not once that I can think of using it to elucidate an actual linguistic phenomenon. (Well, I tell lie here, James MacCawley who worked with GB – and there must have been others – was probably an exception.) Chomsky’s followers who actually continued to have real insights about language – Ross,  Langacker, Lakoff, Fillmore – simply ceased to work within that field – their frustration given voice here by Robin Tolmach Lakoff:

[Generative approaches to the 'science' of language meant] “accepting the impossibility of saying almost everything that might be interesting, anything normal people might want or need to know about language.“ (Robin Tolmach Lakoff, 2000, Language War)

So who deserves the label “gifted linguist” defined as somebody who repeatedly elucidates legitimate language phenomena in a way that is relevant across areas of inquiry? (And I don’t mean the fake relevance followers of the Universal Grammar hypothesis seem to be finding in more and more places.)

Well, I’d start with MAK Halliday who has contributed genuine insights into concepts like function, cohesion, written/spoken language, etc. Students on “linguistics for teachers” courses are always surprised when I tell them that pretty much all of the English as first or second language curriculum used in schools today was influenced by Halliday and none by Chomsky – despite valiant efforts to pretend otherwise.

But there are many others whose fingerprints are all over our thinking about language today. The two giants of 20th century linguistics who influenced probably everyone were Roman Jakobson and Charles Fillmore – neither of whom established a single-idea school (although Jakobson was part of two) but both were literal and metaphorical teachers to pretty much everybody. Then there’s William Labov who continues to help shift the “language decline” hypothesis on which much of 19th century philology was predicated. And, of course, there are countless practicing linguists who have interesting things to say about language every day – one needs to look no further than the contributors to the excellent Language Log. I don’t want to list any others of the top of my head lest I forget someone important, but here some of my favorites:

My personal favorite linguist has long been Michael Hoey whose “lexical priming” hypothesis deserves more discussion and a lot more following than it has received. I got a real chill of excitement reading William Croft’s “Radical Construction Grammar”. It is probably the most interesting and innovative view of language that has come about since de Saussure.

Most of my thinking about language has been influenced by George Lakoff (so much I translated his thickest book into Czech – http://cogling.info) and Ronald Langacker who could both be said to be ‘single-theory’ thinkers but are actually using this theory to say interesting things about language rather than using language to say interesting things about their theory.

I have said to people at one point or another, you should read one of these linguists to understand this point about language better. I have never said that about Chomsky. Not once. I have said, however, you should read this thing by Chomsky to understand Chomsky better. (Not that it always helps, I’ve come across a book called Structure of Language whose authors’ sparse reference list includes all of Chomsky’s books but who refer to his work twice and get it wrong both times.) There is no denying Chomsky’s multi-disciplinary brilliance but a particularly gifted linguist he is not. He is just the only one most people can think of.

BTW: Here’s why I think Chomsky’s wrong. But that wasn’t really the point. Whether he’s right or wrong, he’s largely irrelevant to most people interested in language, and the sooner they realize they’re wasting their time, the better.

Send to Kindle
  • Pingback: Linguistic Anthropology Roundup #12 – Society for Linguistic Anthropology

  • LBHR

    Can you clarify what your criteria would be for “elucidat[ing] an actual linguistic phenomenon”?

    Presumably the difference in acceptability between the English sentences “I did not eat” and “I not ate.” is a real linguistic phenomenon, right?

    Is your objection to Generative Grammar that it is too narrow (i.e. ignores some interesting questions)? That there’s something really deeply broken with it (i.e. it’s a pseudoscience or something)? Something else?

    • admin

      Well, actually my objection is that even the difference between “I did not eat” and “I not ate” is not a real linguistic phenomenon. It’s an artifact of the model adopted by Chomsky. Only if we think that sentences are generated by a formal algorithm do we have to be concerned not to generate ones that are not acceptable. But as Chomsky himself claims elsewhere, we mostly learn what to say and not what to say. But because there’s no evidence that we learn it in a generative manner (and lots of evidence we don’t), it’s unlikely that some internal constraints prevent us from saying “I not ate”. We say “I did not eat” because that’s the pattern we learned to say.

      Chomsky happened on some significant patterns of regularity in his description of language. But let’s remember that Erich Von Däniken also came upon some strange regularities in the relationship between the dimensions of the pyramids and the distance to the Moon.

    • LBHR

      The Generative notion of ungrammaticality is certainly an artifact of the Generative/Generativoid model, but only insofar as forces are an artifact of Newton’s model. I admit this kind of objection has made me worry about science, but it’s not unique to generative grammar.

      Would you find ambiguities in meaning to be a less controversial example of a linguistic phenomenon? For instance “I saw the man with the binoculars” (who has the binoculars) or “Every boy loves a girl” (the same girl?).

      As for the pyramids… in science, there’s always the risk that your results are going to be coincidental, and with a baby science like linguistics the risk is especially high. That doesn’t mean we should give up– on the contrary, it’s all the more reason to keep hitting our frameworks with more and more data, control for various variables, and see which fmwks are fruitful and which are not.

      [POSTSCRIPT: No generative grammarian would say that you have to specially learn what not to say. In calling a sentence ungrammatical, we're simply saying that it's not one of the things you do learn to say. I'm also puzzled by your mention of formal algorithms. Generative grammar has nothing to do with those. An algorithm is a list of instructions to accomplish some task– generative grammarians ask what the task is that's responsible for linguistic phenomena.]

    • G Bell

      Generative grammar, unlike many common and useful ways of codifying grammar rules, is so precise that for any set of generative rules, there is an a computer program (algorithm) to list all the legal sentences. Its very precision might be held against it.

  • noel

    Hmm, you do not count as someone who knows how to think.

  • Pingback: Chomsky chomsky chomsky chomsky furiously « Statistical Modeling, Causal Inference, and Social Science Statistical Modeling, Causal Inference, and Social Science

  • Ayidohi

    Labov > Chomsky. Theoretical linguistics a physicist studying an ideal gas, and claiming that the results are true for a real gas. If language can only be observed in practice, then Chomsky’s idea of “competence” cannot actually be studied, or even verified as existing. That for me is enough to question theoretical linguistics.