Chomsky's Linguistic Theory

[dash_two]

There's that pesky social interaction again.

Indeed, but out of the vast range of grammars which those children could construct, the ones they do conform to UG, i.e. a subset of all possible grammars.

There was a theory going around about ten or twelve years ago, inspired by the revitalisation of associationist learning models, that the rules of UG were in fact better characterised as regularities, and these arose in children's minds due to various attentional biases, informational bottlenecks, and neural noise.

It was a kind of soft nativism, in that complex mental structures were held to develop thanks to relatively simple and non-specific initial constraints. Quite an interesting take IMO, but dunno how it was developed later on. Unlike the social constructionist crap, it did attempt to produce and test hypotheses on simple model grammars, sometimes done with the aid of simulated neural networks.

 

[kyser_soze]

Chomsky speculates somewhere about whether a child could learn language just from watching TV. Such questions are fascinating, but only a Dr. Mengele would conduct the experiments necessary to answer them.

Hmm, judging by the linguistic levels of some kids today and the lack of actual parenting done, this is already happening I suspect...

The actual physical neural pathways we need for speech are there from when we're born, probably from when the brain first starts forming, but it needs to be 'activated' - i.e. you need to talk and be talked too (which includes learning what shapes to make with you mouth).

I don't know enough about universal grammar to comment, but are languages like Basque and Magyar included in that? I know that they are seen as really hard to learn because they don't have the greco-latinate base of other Euro languages, and are radically different from most other, non-Euro languages...

 

[littlebabyjesus]

Indeed, but out of the vast range of grammars which those children could construct, the ones they do conform to UG, i.e. a subset of all possible grammars.

There was a theory going around about ten or twelve years ago, inspired by the revitalisation of associationist learning models, that the rules of UG were in fact better characterised as regularities, and these arose in children's minds due to various attentional biases, informational bottlenecks, and neural noise.

It was a kind of soft nativism, in that complex mental structures were held to develop thanks to relatively simple and non-specific initial constraints. Quite an interesting take IMO, but dunno how it was developed later on. Unlike the social constructionist crap, it did attempt to produce and test hypotheses on simple model grammars, sometimes done with the aid of simulated neural networks.

Yes, this sounds like a better way of thinking about the problem.

This is what I meant about Chomsky - not that his ideas were right or wrong, but that they constituted the wrong way of thinking about the problem.

 

[dash_two]

Because they had a nativist aspect to them? Lots of people don't like that kind of thing, usually it's for political reasons. By the same token, lots of people don't like environmental explanations, again for political reasons.

I can't think offhand of any instance where a political, moral or religious prediction about how the world ought to be has been strongly confirmed by a later scientific discovery of how it is. (Would be interested to learn of any instances contradicting this.)

 

[goldenecitrone]

I think there must be some kind of 'universal grammar' as children very quickly start producing complex utterances that adhere to rules they have never been taught. Pinker describes it best in 'the Language Instinct'. Though, what exactly this 'universal grammar' is and how it comes from the connections in the brain is still a pretty grey area, if you'll pardon the pun.

 

[Knotted]

Sorry but what is that controvosy related to universal grammar. Language would have a darwinian explanation in any scientific theory except intelligent design or or something pretty out the mainstream.

Chomsky maintains that there is no adaptionist explanation for language. I think he rejects any scientific explanation (or indeed non-scientific explanation). Its all beyond our ken.

 

[Knotted]

Because they had a nativist aspect to them? Lots of people don't like that kind of thing, usually it's for political reasons. By the same token, lots of people don't like environmental explanations, again for political reasons.

I can't think offhand of any instance where a political, moral or religious prediction about how the world ought to be has been strongly confirmed by a later scientific discovery of how it is. (Would be interested to learn of any instances contradicting this.)

I share littlebabyjesus's attitude but not for political reasons. I'm not particularly well informed about Chomsky's theories but I find the supposedly ex cathedra nature of universal grammars a bit alarming.

The theories you mention above seem to be attempts at understanding the question by breaking it up into smaller questions. This is surely the correct approach.

The nature versus nurture debate will undoubtably go on and on. But regardless the way forward is bottom up rather than top down. Surely?

 

[dash_two]

Agree that the decomposition of a big question into smaller ones is sensible, but not aware of any linguistic theory which doesn't do this in some way. Re. 'bottom up', I guess that you are referring to theories which address the 'hardware' level of explanation, i.e. what the brain structures enabling language are like and how they grow etc. That seems sensible too, but it doesn't necessarily rule out explanations at different levels.

A readable and reasonable one page summary of the subject here:

http://thebrain.mcgill.ca/flash/capsules/outil_rouge06.html

 

[littlebabyjesus]

I'm slightly confused by the introduction of politics here. When I say that one way of thinking about things seems 'better' than another, I do so because I think it will produce better insights. That is all.

Re: Nature vs. Nurture, this is a non-debate for me. Nurture acts on Nature - or, you might say, nurture is nature changing through time. I don't see any need for debate about this as it seems fairly self-evident.

Some points brought up in this thread are very interesting. I'll come back.

 

[laptop]

I'm reading Henry Plotkin's The imagined world made real at the moment.

I don't know precisely what it is that's "innate" - or even precisely what "innate" means - but he's re-convinced me that there must be something that is.

Even if that something is more abstract than grammar by as as much as grammar is more abstract than the sound "aaargh!".

Meanwhile, this letter from Steven Pinker alerts to an anti-Chomshyan who may bear looking at:

Philip Lieberman has been stewing in his bile over Noam Chomsky for so long that he sees signs of a Chomskyan conspiracy everywhere, including in my book The Stuff of Thought (6 October, p 57). I make a few mentions of Chomsky, half of them critical.

http://www.newscientist.com/article/mg19626300.400-innate-bile.html

 

[nosos]

I don't know precisely what it is that's "innate" - or even precisely what "innate" means - but he's re-convinced me that there must be something that is.

The idea that nothing is innate, that it's all learnt and culturally constructed, is completely fucking absurd. I don't understand why people who would recoil at the idea when Locke expresses it, are happy to accept it when they read it in Foucault or Judith Butler. Of course the epistemological question of how we can know those innate properties - indeed what it would even mean to say we know them - is of a different order entirely.

 

[gurrier]

Funny thread. I thought that the concept of there being some sort of inate universal grammar was essentially proven decades ago and that this was as close to universally accepted as you can get in an academic field that includes non-scientists.

I was also suffering under the impression that the latest proposed grammars can successfully describe over 99% of all human language and have been applied, with similar levels of success, to languages which the designers were entirely ignorant of.

It's been a few years since I actually constructed a Finite State Machine based NLP parser or worked directly on the problem, but I would be very surprised if I'd missed the debunking of the entire field.

 

[dash_two]

Read a while ago about a people in the Amazon whose language, it was claimed, violated UG by not showing any signs of recursion. There was a big row over this, some claimed the anthropologist who'd met them had got it wrong - can't remember how it was resolved. Other than that, some 5,000-6,000 languages have been found to comply with UG so far, iirc.

Edited to add: here's a recent article from Prospect about this:

http://www.prospect-magazine.co.uk/printarticle.php?id=9606

 

[littlebabyjesus]

Funny thread. I thought that the concept of there being some sort of inate universal grammar was essentially proven decades ago and that this was as close to universally accepted as you can get in an academic field that includes non-scientists.

I was also suffering under the impression that the latest proposed grammars can successfully describe over 99% of all human language and have been applied, with similar levels of success, to languages which the designers were entirely ignorant of.

It's been a few years since I actually constructed a Finite State Machine based NLP parser or worked directly on the problem, but I would be very surprised if I'd missed the debunking of the entire field.

One of the problems, as I understand it, is that rapid language extinction is making verification of UGs very difficult, and those languages which differ from 'ours' most are precisely those that are dying out. There is a danger when studying UG that you will find what you are looking for precisely because you are looking for it - ie you're looking for similarities, which means that you are liable to ignore the significance of the differences. Also, the more similar a society's view of the world is to yours, the easier it is to study that society's language, and the more likely its language is to resemble yours.

 

[maomao]

One of the problems, as I understand it, is that rapid language extinction is making verification of UGs very difficult, and those languages which differ from 'ours' most are precisely those that are dying out. There is a danger when studying UG that you will find what you are looking for precisely because you are looking for it - ie you're looking for similarities, which means that you are liable to ignore the significance of the differences. Also, the more similar a society's view of the world is to yours, the easier it is to study that society's language, and the more likely its language is to resemble yours.

What actual, demonstrable alternative do you have to UG?

 

[dash_two]

Also, the more similar a society's view of the world is to yours, the easier it is to study that society's language, and the more likely its language is to resemble yours.

Why should similarity of another society's view of the world with that of one's own society reliably predict similarity of syntactic features? Not saying it necessarily doesn't, but I am curious as to your reasoning.

 

[littlebabyjesus]

Why should similarity of another society's view of the world with that of one's own society reliably predict similarity of syntactic features? Not saying it necessarily doesn't, but I am curious as to your reasoning.

We're into tricky territory here - does language constrict thought, or is the form language takes determined by thought. A bit of both, I would have thought. Certainly, when I talk and think in Spanish, I can feel my personality change.

A particular world view will require certain concepts to describe it, and words will need to be combined in certain ways to deal with those concepts.

For instance, I could imagine a society which considers reality to be a series of events rather than anything concrete (in a similar way to some modern physicists' conceptions), which would describe everything in terms of action and may not have a category that is exactly equivalent to our noun.

Concepts of time differ considerably across cultures. The Incas saw time as circular, not linear. This will have repercussions on categories of verb tense, but there is nobody left alive who has grown up only thinking in terms of the Inca concept of time, so we cannot study this.

 

[kyser_soze]

What does Chomsky's universal grammar have to say about this:

lol nah id hit and and shes tight and down will anal.

 

[Knotted]

Agree that the decomposition of a big question into smaller ones is sensible, but not aware of any linguistic theory which doesn't do this in some way. Re. 'bottom up', I guess that you are referring to theories which address the 'hardware' level of explanation, i.e. what the brain structures enabling language are like and how they grow etc. That seems sensible too, but it doesn't necessarily rule out explanations at different levels.

A readable and reasonable one page summary of the subject here:

http://thebrain.mcgill.ca/flash/capsules/outil_rouge06.html

I'm not being particularly coherent I suppose.

Nature versus nurture is usually a natural history explanation versus a social history explanation. That is, at the end of the day, we appeal to the history of environmental intereractions of either the individual or the species. With Chomsky it is neither. I don't know whether this is just Chomsky's personal beliefs or whether this is a logical consequence of his theories. I know that Pinker says that its just Chomsky, but I suspect that Lieberman would say that its Chomsky's theory.

 

[littlebabyjesus]

Nature versus nurture is usually a natural history explanation versus a social history explanation.

This is the problem. The explanation must consider both.

 

[dash_two]

Concepts of time differ considerably across cultures. The Incas saw time as circular, not linear. This will have repercussions on categories of verb tense, but there is nobody left alive who has grown up only thinking in terms of the Inca concept of time, so we cannot study this.

I dunno much about the Incas at all. Maybe time was circular in the cosmological thinking of the priesthood, but in everyday life? Seems hard to imagine how any people can get by without concepts like 'before', 'after', 'later', 'now', 'yesterday', 'today' etc. The Inca language was Quecha, which is still spoken by around 8 million people (it says here). It has past, present and future tenses eg rimarani 'I spoke'; rimachkani 'I am speaking'.

If the linear concepts of time were imported from the Spanish (or imposed by them), then we might expect present-day Quecha to inflect verbs, for example, in the way Spanish does. But this is not the case.

 

[Fruitloop]

What does Chomsky's universal grammar have to say about this:

Dat expresiv complexitee iz feetur of langauge itself, uver stuff just social innit.

 

[laptop]

What does Chomsky's universal grammar have to say about this:

lol nah id hit and and shes tight and down will anal.

a) It clearly has a grammar; and

b) the grammar it has conforms to the UG.

In fact, its grammar is very, very close to English.

Question. (It's been a long time.)

Is it right to think of the UG as a grammar-for-constructing-grammars?

(Is it right now, if not necessarily in Chomsky 1955?)

 

[dash_two]

Is it right to think of the UG as a grammar-for-constructing-grammars?

hmm er, when first heard Principles and Parameters explained it made me think of the way hox genes work. (It's easy to come up with ideas like that when you know next to FA about either grammars or hox genes.)

A grammar-for-constructing-grammars is snappy and memorable, but I would personally play safe and describe the initial-state-of-the-language-faculty-they-call-UG as a 'schema'.

 

[goldenecitrone]

What does Chomsky's universal grammar have to say about this:

I think you're confusing that with Chomsky's theory of universal dickheadery.

 

[littlebabyjesus]

I dunno much about the Incas at all. Maybe time was circular in the cosmological thinking of the priesthood, but in everyday life? Seems hard to imagine how any people can get by without concepts like 'before', 'after', 'later', 'now', 'yesterday', 'today' etc. The Inca language was Quecha, which is still spoken by around 8 million people (it says here). It has past, present and future tenses eg rimarani 'I spoke'; rimachkani 'I am speaking'.

Another problem when talking about tenses is that there isn't an exact one-to-one mapping from one language to another. An English speaker learning Russian has to think in a different way to use the perfective and imperfective correctly. Similarly the preterite and imperfect in Spanish. French speakers struggle with the present continuous in English. Each language has its own set of categories and concepts that are specific to it... And my examples all developed from a common ancestor, Indo-European.

 

[laptop]

play safe and describe the initial-state-of-the-language-faculty-they-call-UG as a 'schema'.

So I shouldn't use the grammatical construct "meta-"

 

[laptop]

the way hox genes work

Hm... that could be quite deep.


* ruminates on "gradients" in the sense-perception-universe *

 

[goldenecitrone]

Another problem when talking about tenses is that there isn't an exact one-to-one mapping from one language to another. An English speaker learning Russian has to think in a different way to use the perfective and imperfective correctly. Similarly the preterite and imperfect in Spanish. French speakers struggle with the present continuous in English. Each language has its own set of categories and concepts that are specific to it... And my examples all developed from a common ancestor, Indo-European.

However fancy the tenses get though, they are still only talking about four basic concepts, ie past, present, future and hypothetical. I've learnt three languages so far, admittedly all indo-european, but I wouldn't say I've noticed any differences in how I think about things. Maybe I should give Mandarin a go.

 

[littlebabyjesus]

However fancy the tenses get though, they are still only talking about four basic concepts, ie past, present, future and hypothetical. I've learnt three languages so far, admittedly all indo-european, but I wouldn't say I've noticed any differences in how I think about things. Maybe I should give Mandarin a go.

Is that the only way of categorising them though? What about this:

There are only two tenses: 'Now' and 'Not Now', following on from the insight that, from an idealistic perspective, 'now' really exists while past and future are concepts I hold onto in my mind.