#generativism

I used to think structures like this existed only in the wildest fantasies of generativists. And yet, here is a real-life example of triple center embedding from Plato’s prose:

We know that human beings evolved from ancestors without language. This means our language faculty must have evolved and it must therefore be capable of evolving. The innate grammar module the generativists propose is not "evolvable" and poses other problems besides, and cannot therefore be part of our language faculty. On the other hand, there probably are language instincts linked to babbling, selective hearing and the learning of words. Language is also based on a range of social instincts, but these are not language-specific.

~ Sverker Johansson, The dawn of language, p. 197

Want to hear all the drama in linguistics? Check out “The linguistics wars”, which documents one of the most vicious intellectual battles between what have become the two dominant camps in linguistics today—generativism and functionalism.

The level of vitriol in linguistic controversies never ceases to amaze me. Some people really, really don’t like Chomsky’s theories.

I think this paper is grossly unfair to Chomsky. It makes some reasonable criticisms of his ideas, which is fine, but then talks as if those criticisms definitively, objectively prove Chomsky’s ideas to be wrong (which I don’t agree with, and which in any case is not conducive to fruitful discussion), then goes on to accuse Chomsky of intellectual dishonesty because these criticisms are supposedly so devastating that he surely can’t have kept believing in his ideas after the criticisms were first published, yet he still kept acting as if he believed in them!

It’s also a great illustration of the pitfalls of assuming a strict binary distinction between grammatical and ungrammatical sentences. On pages 3--4, Chomsky is excoriated for his assertion in Syntactic Structures that every simple active sentence in English can be put in a passive form; the authors cite the following counterexamples:

(1) The kids want ice cream. / *Ice cream is wanted by the kids.

(2) That movie starred Julia. / *Julia was starred by that movie.

(3) The ocean liner neared the iceberg. / *The iceberg was neared by the ocean liner.

(4) Karen’s remarks betrayed contempt for socialism. / *Contempt for socialism was betrayed by Karen’s remarks.

The existence of these counterexamples, the authors say, demonstrates that Chomsky’s assertion contained a “gross factual error”. But their real complaint---since, they admit, “even honest researchers” make mistakes now and then---is that Chomsky had already given examples of ungrammatical passives in an earlier publication from 1955 (*three pounds were weighed by this; *his punishment was gotten by him). So, they say, Chomsky was lying in Syntactic Structures---he knew the claim he was making was false.

Yet if your intuitions are anything like mine, you might notice something funny about these counterexamples: some of them are not actually counterexamples, according to your own intuition. For example, I think (4) is completely fine, whether it’s in the active form or thr passive form. I also don’t object too much to the passives in (1) and (3), though they do sound a bit funny; (2) is the only one I would definitively say is ungrammatical to me.

So it’s really a lot more complicated than either side in this dispute is making it out to be. Even if you’re committed to not concluding that grammaticality isn’t a binary, you have to at least admit that intuitions about grammaticality are not entirely reliable. They vary from speaker to speaker, or even at different times from the same speaker.

My charitable interpretation of Chomsky’s actions would be that between 1955 and 1957 he simply changed his mind about whether “three pounds were weighed by this” and “his punishment was gotten by him” were straightforwardly ungrammatical. Perhaps he realized that passive sentences were one of these domains in which grammaticality judgements seem to vary a lot between different speakers, and therefore concluded that his intuition on which passive sentences were grammatical or not wasn’t very reliable. Perhaps he talked to a colleague who claimed that “three pounds were weighed by this” was grammatical---I wouldn’t be at all surprised if such speakers exist. One can fault Chomsky for not addressing these conceptual issues with his theory, but that’s a different matter from accusing him of lying. Besides, from the small parts of it I’ve read, Syntactic Structures seems like a rather short, concise, elegant expository sort of work---perhaps he just didn’t want to get bogged down in these issues too much in the text.

There are similar problems with Levine & Postal’s discussion of the A-over-A principle, where they cite counterexamples (this time grammatical counterexamples, which should be ungrammatical according to the A-over-A principle) which make use of topicalization. I find it very hard to judge the grammaticality of sentences making use of topicalization because topicalization always sounds somewhat wrong in English. I’m not sure which of “Robin, I love to hear stories about.” and “Stories about Robin, I love to hear” is better, but I can imagine somebody thinking the latter; but Levine & Postal make out the former to be straightforwardly better. In any case, the existence of edge-case exceptions to a principle doesn’t imply it’s not useful as a generalization or as an initial hypothesis which can be refined by subsequent work, whether those exceptions take the form of ungrammatical sentences or grammatical sentences (Levine & Postal seem to think grammatical exceptions are a more serious matter than ungrammatical exceptions).

9/15/16

Thing-A-Day #867 - Generativists

In broadest terms, generative linguistics is one way to study and model language structure. It is therefore a part of the broader structuralist program (see for instance F Newmeyer).

In particular, generativism shares these structuralist commitments:

Language as an interconnected hierarchical semiotic system

Distinction between competence and performance (languge/parole)

Meaning and function expressed through distinctive features

Language susceptible to syntagmatic and paradigmatic analysis

Specific concepts such as phoneme, morpheme, lexeme

Language units as symbolic pairings of meaning and form

Linearity of language

Etc.

While many of the generativist approaches grew out of a critique of the structuralist orthodoxy of the time, they still stayed firmly within the broader structuralist framework. In some way, generative linguistics could be seen as a contribution to a structuralist theory of syntax.