The 50th Anniversary of Noam Chomsky's Aspects of the Theory of Syntax

http://chronicle.com/blogs/linguafranca/2015/06/22/revolutionary-methodological-preliminaries/

by tintinnabula • 11 years ago

50 26 11 years ago

snitzr• 11 years ago

Recommended: Is The Man Who is Tall Happy? on Netflix. A nice introduction to Chomsky and his linguistic ideas.

nsajko• 11 years ago

Does somebody know any alternative (legal) sources? Netflix is not available to me without proxying through another country.

formerlinguist• 11 years ago

This was actually not Chomsky's first book on this topic. That was Syntactic Structures (1957).

simonebrunozzi• 11 years ago

Correct. That book, over time, changed a lot of things.

bdchauvette• 11 years ago

Neither the title here nor the article itself claim it to be his first book, only an incredibly influential one.

The article even agrees with you:

§3 sets out a radically revised structure for transformational grammars, quietly abandoning the notion of “kernel sentence” from Chomsky’s first book, Syntactic Structures, and introducing the term “deep structure,” which in a sense replaces it.

samch5• 11 years ago

"Chomsky's theories carried an atomistic, bottom-p approach to language to it's natural conclusion. It certainly did inspire a generation of research, most of it misguided. But its real flaw was one it inherited from its structuralist predecessors. It treated utterances as decomposable structures rather than as wholes, which is how the brain treats them. From the very first the Chomskyites had to admit that no research supported any connection whatsoever between the structures they posited and what went on in the brain. They assumed that further brain research would be needed to uncover those connections. But all the brain research since then has shown that transformational grammar (and even its rivals, which I studied) bears no resemblance to what the brain does with language. So what Chomsky gave us was the foundation for the ultimate mathematical elaboration of a structuralist theory that bears no relationship to anything in nature."

-My dad (Ph.D. Linguistics, Yale - way back in the day)

raverbashing• 11 years ago

was the foundation for the ultimate mathematical elaboration of a structuralist theory that bears no relationship to anything in nature

So, like most theories?

Or do you really think of electrons as 'tiny things'?

You do the math and you get the results.

But yeah, Chomsky structures seems to be more useful when applying it to computer languages than to real ones.

samch5• 11 years ago

I think I've missed the point of your response (electrons?). Just to clarify, Chomsky's theories setup a framework for artificial reduction of language which did nothing to help elucidate the true nature of language (which is obviously a product of the evolution of the brain). The problem seems to be that Chomsky's influence sent a lot of grad students down a path that didn't really advance linguistics in terms of the human element. I don't even know if his theories really helped with computer analysis of language.

lnanek2• 11 years ago

Very accurate. Honestly, reading Chomsky in a lot of places he says things like there's no way a baby could learn this, it must be hard coded in their brain! Then in baby studies done later we actually find babies do have a natural understanding of things like probability and can do the things he says are impossible.

There are many other flaws in his work. I tend to think of him as just an old man who wrote his stuff before we knew as much as we do about genetics, so was unfortunately really off base. It's nice that people who don't know the science were excited by him, I guess.

discreteinfinit7• 11 years ago

Could you link some of those 'baby studies' you are referring to ?

discreteinfinit7• 11 years ago

Thing is, Chomsky is (at least partly) responsible for the so-called Cognitive Revolution. Look up B.F Skinner and notice what happened to his career after Chomsky reviewed one of his papers. Skinner's brand of behaviorism was surely something we didn't need around anymore. He also changed the way everyone thinks about language. Before Chomsky, there really was no real theory of syntax that was even close to being scientific and formal.

Although probabilistic machine learning methods replaced the Chomskyan brand of formal theoretic methods in NLP, a lot is owed to is work. Obviously there is the Chomsky Hierarchy but even things like the Penn Treebank POS tags are taken straight out of work in the Generative tradition from circa 1970 (not sure why no one has tried the updated and more accurate tags currently being used in theoretical linguistics). Anyways I could go on and on but Im wondering what your Dad has to say about this ?

P.S There is an ongoing joke in the field of Linguistics about people who either think they understand Chomsky or just straight up don't understand him. Its difficult stuff.

samch5• 11 years ago

"No question that a lot of the tools developed under generative grammar have proved useful, and one can certainly cite the Huxleyan principle that truth comes sooner from error than from a vacuum. I think I would say that the American structuralist tradition was headed in a wrong direction from at least 1933 (Bloomfield's Language) and that Chomsky (despite the claims of his followers that he made a revolutionary break from structuralism) simply gave structuralism a mathematical rigor that allowed it to be taken to its logical conclusion. The tools that were developed in the process have allowed the field (or the more brain-oriented parts of it) to move in a rather decisively different direction.

Full disclosure: I got my PhD at Yale under Syd Lamb, so naturally I would not be classed as a Chomskyan. It took me some years to realize that Lamb's StratGram itself suffered from the same reductionist flaws. The only theoretical school I know of that had a chance of moving in the right direction (until very recently) was Firth's. Unfortunately, neither he nor his followers developed the theoretical tools that were needed. Certainly Halliday's Systemics became as reductionist as any of the others. Hudson's word grammar made some good moves but never really developed well. So yeah, Chomsky set something in motion that produced some useful theoretical tools that could be used to help push off in new directions. But the TGG tradition, being a well developed extension of the Bloomfieldian tradition, is still basically a dead end."

-My dad

panglott4• 11 years ago

Perhaps. It's kind of a trope these days to deride pre-Chomsky linguists as a bunch of wrong-headed behaviorists. But Skinner was advocating a theory of psychology. Most linguists prior to Chomsky were not really committed to behaviorist psychology, just happy to borrow it for use against the prevailing scientific racism.

Chomsky is a theoretical linguist, but not really a scientific one. Generative linguistics does not have a good tradition of basing conclusions about language on empirical study of language use, with the implication that there's hardly any point in studying the structure of diverse languages (since they're all the same deep structure). Take the speech establishing the Principles & Parameters theory, which likened the linguistic feature of pro-drop to a series of switches the brain could flip on or off to establish a language as pro-drop...when subsequent research immediately showed pro-drop to be more of a continuum than a binary feature. There is all the wasted effort, like generative grammars of endangered languages that relied on morphology rather than syntax to communicate the same information.

Then there are the bad assumptions, like the "poverty of the stimulus" argument, that had no real empirical basis when he made it. This supported ideas like linguistic nativism that got exported out of linguistics since Chomskyan linguistics was a dominant theory at the time. Whereas in truth linguistics has always been a multi-paradigmatic field.

syllogism• 11 years ago

But meaning is compositional! At least in large part.

Most sentences we encounter are novel, and yet they're understood. So, somehow, we break up the meaning of the novel sentence, relating it to sentences we've seen.

And actually the process there isn't that mysterious. The basic facts of English syntax are pretty obvious. Any English speaker will agree that "The blahdies merikaled the beemers" implies "The beemers were merikaled by the blahdies", without knowing what these words mean.

I see from your Dad's comment below that he's fond of the systemicist school, following Halliday etc. I've done substantial work with this theory, as it's what I learned in my undergrad. The rest of my career has been spent working with others.[1]

If I had to identify a central objection to SFL, I might say this: SFL is very fond of proposing non-linear effects. The theory goes that it's all about interactions, and if you study anything in isolation, you won't find anything out --- because everything's non-linear.

This assertion is basically unsupported. It just sort of sits there as an assumption. But we should assume simplicity, not complexity. The simplest assumption is that there is no relationship between X and Y, but the next simplest is that there's a linear relationship. If you want to tell me there's a non linear relationship, you'd better come with evidence. And SFL never does.

[1] https://scholar.google.com/citations?user=FXwlnmAAAAAJ&hl=en

samch5• 11 years ago

"Two comments: First, I'm not fond of the systemicists. I think Halliday took a wrong turning.

Second, you say that "somehow, we break up the meaning of the novel sentence, relating it to sentences we've seen." Quite correct. But I think the essential point of the Firthian approach is that there is not necessarily a single decomposition of the string. There can be multiple decompositions, each yielding components that can be related to other utterances. All of the standard approaches start by assuming single decompositions (or decomposition trees) on each level. That's where I think the whole structurallist (and systemicist) project went off the tracks. I did manage to crank out a few papers in this direction many years ago, but left it before I had gotten my act together, and I succumbed to the lure of Milton's "that bad eminence" (administration)."

-My dad

pc2g4d• 11 years ago

It is rather surprising that more has not been done this year (thus far, anyway) to commemorate a significant semicentenary: the 50th anniversary of what could reasonably be called the most influential linguistics book of the 20th century.

Linguistics as a field is in a post-Chomsky hangover and none-too-happy to celebrate the man that put them there. Chomsky's many years as the dominant voice in the field were traumatic for linguistics. Once a decade or so he would promulgate The New Way of doing syntax, handed down from on high from this very influential man, which meant incredible disruption not only for syntacticians but also for those working in semantics and morphology/phonology.

Much of Chomsky's influence came not from his intellect (though that was surely prodigious) but from his forceful personality. His will to publicly ridicule not only opposing ideas but opposing personalities was a big part of his rise and continuing dominance. Linguistics was a nasty place to work for a long time. (See "The Linguistics Wars").

Now that he's not actively working to maintain it, Chomsky's academic reputation has suffered substantially and the field is largely trying to move on. Large-scale corpora and computational capabilities enable more empiricist rather than rationalist approaches to human language, which has made the argumentation less personal and more productive.

At least, that's my take.

The 50th Anniversary of Noam Chomsky's Aspects of the Theory of Syntax

Related Stories