Has Chomsky been blown out of the water? [pdf] | Hacker News Reader

Has Chomsky been blown out of the water? [pdf]

covingtoninnovations.com

26 pointsbreck36 comments

This is like asking "Has Einstein been blown out of the water?" in the context of physics.

The answer is that such a thing would be impossible, because even though his theories and models are imperfect (as all theories and models are), they were so influential that they fundamentally changed his field.

Chomsky had a similar effect on linguistics, and even if much of his work ends up being rejected, his influence will have changed the field of linguistics forever. The theories that supersede Chomsky's will be developed within the paradigm he created.

Linguist here. It really isn't. There are plenty of competing and very compelling theories in e.g. big parts of Europe. Especially Functional Grammar and its different branches. The only relation they really have to UG is that they're not it.

At my university no one in the linguistics department actually subscribed to UG or any derivatives thereof but in the English department they did. I've observed sort of the same divide in countries. In English speaking counties UG seems to be the prevailing theory while it's much less so in non-English speaking countries.

Unfortunately people are extremely religious about their preferred theory of grammar, and it's hard to have a useful discussion about it when people are unable to remove themselves from the subject and discuss it objectively.

I studied theoretical linguistics through the graduate level and it wasn't until rather late in my schooling that I even began to hear of competing frameworks such as HPSG or LFG, and much of it in passing. There was no mention of them in the coursework, and the tendency of faculty to sell modern linguistics as a hard science gave the impression that we were learning a standard approach that had a provable empirical edge and broad scientific consensus. Otherwise, why would it be called "Extended Standard Theory"?

While I know that part of the pedagogy was giving students a grounding in dominant strains of formalism (in North American circles anyway), at times I felt a bit cheated out of a broader perspective.

Of course in practice many of the competing frameworks are weakly equivalent in terms of their descriptive power and the differences come down more to things like biological plausibility, computational complexity, how strongly you interpret the claims of UG, whether you have any interest in building tractable computational implementations, and so on. Or how you feel about the proliferation of empty functional categories, which was what bothered me the most from an empirical point of view.

If you want to prove a scientist wrong, present evidence to the contrary rather than personal innuendos and hypotheses / observations without proof.

The Scientific American article [1] cited as first reference in the paper is quite interesting.

[1] http://www.scientificamerican.com/article/evidence-rebuts-ch...

This is a good criticism of a number of Chomsky's polemical papers.

"A number", as is usual with Chomsky, being a bunch of stuff you're not going to quote, because you'd likely end up eating all of it. There's gotta be a nest somewhere.

The manner in which children of diverse cultures and varied circumstances acquire language is extremely uniform. Children of accomplished academics in Cambridge pick up English at the same rate at which children of illiterate farmers in Benin pick up Gbe. They will hit the same milestones at the same age.

We also know that a stroke can seriously impair a person's ability to use language, even as their memory and powers of abstract reasoning are otherwise in tact.

The only way to make sense of that that I know of is to suppose that language engages some sort of specialized structures within the brain.

If such structures exist and are implicated in the acquisition of language by children across different cultures, then language must not be infinitely plastic. There must be certain structural properties that languages share that allow them to be acquired in such a regular, programmatic fashion.

Is this conclusion unwarranted, unreasonable, false, or dubious? Are there other ways of making sense of these phenomena?

>The manner in which children of diverse cultures and varied circumstances acquire language is extremely uniform. Children of accomplished academics in Cambridge pick up English at the same rate at which children of illiterate farmers in Benin pick up Gbe. They will hit the same milestones at the same age.

Do you actually know that? Or are you supposing it because it makes sense? Given what I know about the differences in educational outcomes between socioeconomic groups in the US, I wouldn't be surprised if the rate of language acquisition depended on the number and variety of words that a child hears. It may very well be the case that the child of the Oxbridge academics does pick up language more quickly simply because he or she is exposed to more of it.

Yes. Language acquisition follows a standard sequence with specific milestones universal to all languages. For example, the period around 6-12 months is critical for phonetic learning. Before 6 months, infants are able to discriminate between phonetic units present in any language, but after 12 months they are better at making distinctions that occur in their native language and worse at making distinctions that don't occur in their environment. For example, Japanese infants are just as good as American infants at distinguishing between r and l at birth, but lose some of this ability between 8-10 months.

(Chapter "Language", Principles of Neural Science, 5th ed.)

It depends on which milestones you look at. Of course, if you pre-select milestones which kids hit at about the same rate, then all of those taken together will seem to approximately argue in favor of UG.

If you read a bit further (pages 2 and after) it's actually not all innuendo. In fact some of it is actually laudatory about his contributions. It's basically a very brief summary of Chomsky's ideas and the impact he had on linguistics as a profession. Namely using trees to model syntactical forms and his hypothesis on universal grammar. I don't know what it is about HN that makes people refuse to read past the first little thing they object to and then bloviate about it.

> I don't know what it is about HN that makes people refuse to read past the first little thing they object to and then bloviate about it.

That's hardly unique to HN. Following Chomsky, I'd argue that it's actually a universal behavior, which merely finds (slight) variation in expression according to the medium at hand. It even has a Latin name - argumentum ad logicam - and it's closely related to cherry picking except that it's applied to parts of the argument instead of to data. In fact, I just did it to you when I picked on the inessential "about HN" part of your sentence (which reads fine without it), and if you find that annoying then perhaps you should consider that you also did it when you jumped into a response about innuendo without addressing the more substantive discussion that had already occurred.

It's funny, in the 90s at college I knew some people in the Linguistics dept and even back then, people were saying, Chomsky's deep grammar is obviously nonsense but we have to learn it anyway because it's in the exam.

Off topic, but as long as we have a Chomsky discussion - where would be a good place to start reading about his theories on the evolution of language?

More on topic, the dude is a giant, and the linked pdf not worth the electrons it's printed on.

Some of the latter work. He had a book called New Horizons in the Study of Language and Mind, which is fairly accesible and also covers some historical work, including the necessary underpinning of "where did language come from". His work itself isn't as interested in the way languages evolve. That would ironically fall a bit more under the some of the paradigm he replaced. So Bloomfields book on language was in that tradition and an example of the type of scholarly work that contributes to langauge change over time. De Saussere (I believe, been over a decade since I studied this) did work on synchrony vs diachrony. There are actual journals as well at any decent university library or (sizable) city library that would allow you to dip your toe in to working scholarship and get a sense of how historical linguistics works, if thats what you're interested in.

This paper is a good place to start lol. The paper begins with a summary of Chomsky's major contributions, and is fairly readable.

The answer is no, Chomsky will probably never be blown out of the water. His hypothesis is evolving, though.

He has one hypothesis?

It's interesting to think that connectionist theories of language are literally shipping on Google Translate and people still argue about Chomsky. Many of the cognitive scientists were poking about with neural nets too - and they were getting into huge arguments with Chomsky and like folks since the 80's.

To be fair, due to the nature of neural networks, there are a great number of domains where they can be applied in some way without that fact actually explaining anything about how the domain works.

For example, TD-Gammon plays a mean game of Backgammon, but this does not mean that Backgammon reduces to connectionism. TD-Gammon applies neural networks in a specific way with other algorithms involved, and then there are still underlying facts about what it takes to win at Backgammon which do not necessarily mandate anything that resembles connectionism.

This article shows practically no evidence of its main thesis. To substantiate that Chomsky has been "blown out of the water", he mentions 3 things:

1) "[...] greater awareness of the diversity of human languages, and especially the discovery by Daniel Everett of a Brazilian indigenous language, Pirahã, that apparently has no subordinate clauses"

That's the one and only evidence the author shows for this point. No examples on how the language works or how it stands in contrast to Universal Grammar, and no other mention of any other language that might be contradicting UG as well (even though the author implies that there are other languages that stand in contradiction to UG). Later on though the author says "It is not even undisputed that Pirahã lacks subordinate clauses" so the one and only evidence is still a maybe.

2)"The second is advances in cognitive psychology. [...] Nowadays we know that the human mind has complex, powerful, abstract capabilities in many areas, not just language, and new possibilities are opening up for explaining language from general mechanisms of thinking and learning."

This is not an argument and no evidence has been shown for this point.

3)"The third factor is the increasing use of data-driven methods in computational linguistics [...] In a data-driven approach, structure arises because sentences obviously consist of parts that can be recognized before recognizing the whole. [...] This predicts that the grammar of uncommon words will be simple and regular, but common words will have more quirks because we have more examples from which to learn them. That is exactly what we observe in language, and it’s a source of great clutter in a grammar constructed on Chomskyan principles."

As far as I know, the fact that sentences have parts that can be individually recognized before recognizing the whole, is not contradicting UG at all. To the best of my knowledge, UG only says that there is a structure, not how the structure comes about. So the only point left is the fact that common words have more varied usage and are harder to pin down as a particular specific element of the structure. I don't see how that relates to UG quite frankly. Maybe someone can enlighten me on this.

I have no particular love for UG, I just know of it on a very general sense, but if I was a teacher and a student gave me this article as a report, I would've failed him just for the lack of evidence.

  (I don’t hear good things about it from political scientists.)

Immediately closes tab. I assume the rest of this is garbage innuendos, too?

He just couldn't help himself. Just had to throw that in, even after acknowledging he is not familiar with it.

It's an appropriately weak start to a bland document.

I don't know enough about linguistics to have an opinion on that, it just seems too short and lacking sources and explanations for the layman, but to be fair, that's really just getting the "politics out of the way". I'm very sensitive to people trying to have a go at other people they can't face directly, and I was absolutely primed to be upset about that here, but I didn't get the impression at all. Maybe I'm wrong, but fwiw and all that.

Yup.

> Chomsky

Heyyy it's actually about Chomsky qua linguist instead of qua political whatever!

> It would be premature to conclude that universal grammar has been proven not to exist.

> But universal grammar has lost some credibility. Suppose the whole notion is eventually abandoned. What will be left?

> Second, psycholinguistics is going to be big business.

> In short, Chomskyanism may be on the wane,

Welp. This is some Fox News "I'm just saying" bullshit right here.

edit:

> Third, opportunities to discover more about language are continuing to open out before us.

Language attrition (and the dearth of linguists willing / able to work in the field to produce ethnographies of dying languages) are _decreasing_ our opportunities to learn about human language.

Looks like some very important people ordered Chomsky's discreditation campaign, because that guy had enough balls to reveal something about them to general public in his book "How the World Works".

http://literarylilt.blogspot.com.au/2012/02/noam-chomskys-ho...

I had that knee-jerk reaction too, but reading the thing I didn't get the impression it's some kind of stabby thing. But generally: yeah, I just know when Chomsky dies (heavens forbid), so many people will pull so many antics to sweep him under the rug. But another way to look at it is like a flare, that makes it easier to separate wheat and chaff in these murky times, so I ain't bothered. His life and mind will always be an inspiring example to those who care to really look, nothing can undo that. Kinda like people can forget and belittle but not touch Bill Hicks.

"How the world works" should be essential reading for all citizens of the world.

Maybe people will stop listening to his political opinions just because he used to be credible on some other unrelated topic.

Today I learned about Betteridge's Law [0], which seems apt here. It states that when a news headline asks a question, the answer is always "no." When it applies to a paper, I'm disinclined to take the paper seriously...

[0] -https://en.wikipedia.org/wiki/Betteridge%27s_law_of_headline...

This gets posted on EVERY submission ending in a question mark. It was old 2 years ago. Now it's just plain trite, pointless, tiresome and distracting.

I self-aggrandizingly call this tome's law:

https://news.ycombinator.com/item?id=9077549

Why not skip to the end? When enough articles (ending in a question mark or not) are discussed the probability of Betteridge's law (as well as anything else) being mentioned approaches 1. There everybody goes, forever, we are completely done with that class of stuff now. You're welcome :P

Nevertheless, the title does read a bit like clickbait.

Reply on news.ycombinator.com