Definition of Consciousness: 2nd Poll

[skepticalbip]

Like Chomsky, Harris would lecture with a kind of dismissive tone towards the work of those he was criticizing--as if they were just stating trivialities that were of little interest to serious linguists--and he would not let his questioners finish their sentences without jumping in with a dismissive wave of the hand and set of remarks. Chomsky did that a lot with audience participants at the end of this lecture.

Is that common in the field? I'm in math, and I've never seen anything like it. Was uncomfortable to watch.

To me, it seems to be fairly common in all the soft sciences where "theories" are more akin to "just so stories". Since the "theories" can't be proven as in math, defense of the "theory" is limited to rhetoric and attacking detractors and their own "just so stories".

 

[untermensche]

Here is a video where 80 something Chomsky eviscerates so-called working "linguists". I'm sure they all have PhD's.

It is a great video to understand the power of mass delusion.

I've finally had time to listen to the entire lecture, and I see no reason to revise my earlier remarks that you lack the necessary background to understand most of what Chomsky's theory is about. Generative linguistic theory purports to be a psychological theory of language, because it has always been grounded in mental intuitions of well-formedness. That is the nature of the data that it attempts to explain. To say that the theory is "biological" and not "psychological" is nonsensical, because human psychology is grounded in human biology. There is no reason to make such a distinction. Sometimes Chomsky addresses biological issues, e.g. biological evolution, because he believes that our biology plays a huge role in our cognitive development.

The video is simply a guest lecture given to a room full of academics from different disciplines. There is no general criticism of "working linguists", and Chomsky focuses on a fairly narrow range of potboiler issues that have been around since his career began in the mid-1950s as a protege of Zellig Harris. Chomsky's "transformational grammar" bears a superficial similarity of Harris's "transformational grammar", but Chomsky explicitly grounded his approach to language as intuition-based rather than corpus-based. I once attended some lectures by Harris at Columbia Universtiy and was very impressed with the similarity in styles between the two men. Like Chomsky, Harris would lecture with a kind of dismissive tone towards the work of those he was criticizing--as if they were just stating trivialities that were of little interest to serious linguists--and he would not let his questioners finish their sentences without jumping in with a dismissive wave of the hand and set of remarks. Chomsky did that a lot with audience participants at the end of this lecture.

To understand the lecture, one really needs to know about Chomsky's rather long battle against statistical approaches to linguistic analysis. So most of the video was almost identical in content and tone to one that he has been giving since the 1960s. Chomsky is clearly piqued by the fact that statistical approaches suddenly blossomed in the 1980s and actually eclipsed Natural Language Processing techniques based on generative analyses afterwards. In 1988, the late computational linguist, Fred Jelinek, quipped "Anytime a linguist leaves the group the recognition rate goes up", although the popular version of that quote became "Every time I fire a linguist, the performance of our speech recognition system goes up." There are very good reasons for that, but the types of statistical methods that were initially so successful could never scale up for the the reasons that Chomsky mentioned (e.g. that important aspects of language tend to rely on structural processing rather than linear processing). So statistical methods are very good at assigning meaningful classifications to text, but they are very limited at so-called "deep understanding" applications. That is, it is possible to achieve the same results with statistical methods if you simply scramble word order within a sequence of text, effectively producing nonsensical strings of words but assigning coherent meaning to them. The problem has always been that approaches based on generative parsing systems are expensive to create and maintain. Basically, you need to achieve full "artificial intelligence" in order to make them work. So work on sophisticated syntactic parsing systems hasn't progressed much since the 1980s, and modern text analysis systems tend to combine shallow parsing techniques with very effective statistical processing. What upsets Chomsky about this is that such hybrid systems do not in any sense replicate the methods that humans actually use in processing natural language.

Human psychology presently is not well grounded in science. If we say that true science enables prediction of future behavior.

Chomsky's theories are not psychological any more than theories that look at the visual capacity are psychological.

It does not appear you have even watched the video.

What paper did Chomsky go through in detail and what were his specific criticisms of it?

You are just babbling off half-baked notions, saying exactly what you have said before, that has nothing to do with the video.

What is clear is you do not know Chomsky's work at all.

You know your acquired distortions of it. You have some unsupported prejudices. That is all.

 

[Copernicus]

Like Chomsky, Harris would lecture with a kind of dismissive tone towards the work of those he was criticizing--as if they were just stating trivialities that were of little interest to serious linguists--and he would not let his questioners finish their sentences without jumping in with a dismissive wave of the hand and set of remarks. Chomsky did that a lot with audience participants at the end of this lecture.

Is that common in the field? I'm in math, and I've never seen anything like it. Was uncomfortable to watch.

People tend to give Chomsky a lot of leeway because of his celebrity. There were a couple of folks in the room who were authors on the papers he was directly criticizing, and they treated him with far more deference. However, there are a lot of folks out there who react badly to his condescending style of rhetoric. You can see some of that come through in his political lectures, as well. Anyway, there was some tension in that lecture room during the video, and it some times erupted into nervous laughter and snickering at some of his worst polemical reactions to the subject matter and the questioners at the end of the lecture.

Perhaps some of it has to do with attitudes back in the 60s. His early followers at MIT were all very sharp in both wit and intelligence, so they tended to hold his feet to the fire in the classroom. The atmosphere was very informal. There was a very active system of underground publications via mimeographed papers, as people rushed out to get their ideas into the public discourse and couldn't wait for the slowness of formal publication. I have boxes full of old papers. Chomsky and Halle's classic  The Sound Pattern of English was first released that way in the mid-sixties and became part of the base curriculum in linguistic departments before its real publication date of 1968. I was told by one of his first generation students that Chomsky's seminars quite often turned into shouting matches. These days, you go to web discussion forums for that kind of flame war, and people tend to be much more polite to each other.

 

[untermensche]

There were a couple of folks in the room who were authors on the papers he was directly criticizing, and they treated him with far more deference.

They treated him with deference because he was destroying their arguments. He had destroyed the intellectual foundations of what they were doing.

He showed them the errors of their ways. Coming up with statistical formula's to explain the behavior of leaves blowing around can be done. But it is not what physicists do. It is not what scientists do. They look for underlying principles.

They were fighting for their lives feeling threatened by the ideas of an eighty something Chomsky. Who attacked no person and was rude to no person.

 

[Copernicus]

...

Human psychology presently is not well grounded in science. If we say that true science enables prediction of future behavior.

Chomsky's theories are not psychological any more than theories that look at the visual capacity are psychological.

It does not appear you have even watched the video.

What paper did Chomsky go through in detail and what were his specific criticisms of it?

You are just babbling off half-baked notions, saying exactly what you have said before, that has nothing to do with the video.

What is clear is you do not know Chomsky's work at all.

You know your acquired distortions of it. You have some unsupported prejudices. That is all.

TBH, unter, I really wondered whether you yourself had watched that full video as I got into it. You clearly did not understand most of the issues he was talking about, because you would not have been as familiar as I was with many of Chomsky's references. At one point he mentioned Otto Jespersen, an obscure grammarian whose heyday came in the 1930s, and I wondered whether many of his audience had ever even heard that name before. Jespersen was a hot topic of discussion back in the 60s and 70s, but a lot of Chomsky's lecture was essentially the same as the one on poverty of stimulus that he has been giving since the 1960s--as if the same literature were still considered important in graduate reading lists today. I suspect that Chomsky only brought the name up, because he was giving some of that lecture on autopilot. He was reliving the battles of yesteryear.

Did it escape your notice that Chomsky was critiquing more than one paper in that lecture? He did focus more on materials that were relevant to a few of the papers on statistical learning by colleagues who were in the room at the time, and that is where you may have gotten the impression that Chomsky was somehow attacking "working linguists" rather than a very large, successful school of computational linguists, whom he was chiding for its lack of depth in his brand of theoretical linguistics. It is possible these days to get a PhD in computational linguistics without having great familiarity with the generative theoretical literature. Generative theory is relevant to the study of natural language syntactic parsers, but such parsers are no longer generating the interest that they once did. In fact, Chomsky spent a lot of time grousing about how difficult it was for linguists interested in his largely moribund theory of "UG" to get grants from agencies like the NSF, ignoring the fact that the NSF was far more amenable to funding such research when it was run by Paul Chapin, a generative linguist. The fact is that Chomsky's "UG" crowd has largely been marginalized by its failure to produce anything recent that is of relevance to anyone who isn't a member of its own arcane academic circle.

IMO, the real problem with generative grammar is that it attempts to account for intuitions of well-formedness rather than actual behavior. So generative linguists have a psychological approach to language that refers to linguistic behavior without holding itself accountable for providing a causative model that explains that behavior. It explains judgments of grammaticality in terms of formal grammars that somehow influence "external" linguistic behavior without fully determining it. Hence, Chomsky makes a big deal about "internalization" and "externalization". So children are initially supposed to be able to make far more sophisticated judgments about grammar well before they are able to produce the structures they have knowledge about. They acquire "externalized" behavior by evaluating their internal "grammars" in terms of how well they match incoming data. IOW, children are essentially "little linguists". Unfortunately for Chomsky, there is a large literature on just why that idealization of language learning doesn't actually account for the stages children go through in acquiring language. Chomsky is aware of this literature and made some passing comments on it during the video, but he was more interested in going after statistical methods as somehow relevant to human language acquisition. His lecture would have been far more interesting with an audience of developmental psychologists rather than computational linguists.

 

[untermensche]

You clearly did not understand most of the issues he was talking about, because you would not have been as familiar as I was with many of Chomsky's references.

Knowing references and understanding ideas are two different things.

What idea of Jespersen's did Chomsky put forth and in what context?

Bragging about knowledge of references doesn't impress me.

Did it escape your notice that Chomsky was critiquing more than one paper in that lecture?

He had one paper in front of him and he was constantly referencing it. He can throw out a dozen references in five minutes but he was focusing on one paper.

What were his most salient comments about that paper?

IMO, the real problem with generative grammar is that it attempts to account for intuitions of well-formedness rather than actual behavior.

Yes. He postulates a language capacity, not a miracle.

Is vision a learned behavior? Or is it an acquired ability that arises just from exposure?

Why is language different than vision?

 

[Copernicus]

Unter, he was commenting about that paper because one of its authors was sitting in the room. Chomsky was likely invited to give his lecture precisely because of his past critiques of work by people at UCL. FTR, neither you nor I have read that paper, and I fail to see how I could possibly explain the technical issues to someone who doesn't understand binding theory and Chomsky's three principles regarding that theory. There is a reason why Chomsky went on and on about transitional probabilities, but how could you have any of the background in speech recognition to understand what he was talking about? Bear in mind that he was largely addressing people with a background in computational linguistics--a branch of linguistics that Chomsky himself had founded but that has largely left him somewhat marginalized. Nobody can deny the impressive progress that statistical and connectionist methodologies have gained us since the 1980s, especially in areas such as speech processing and analysis of large amounts of text. We now have programs that allow users to actually use them as dictation machines. Ironically, Chomsky was predicting in the 1970s that statistical methods were bankrupt and should be abandoned. He has some serious egg on his face from those claims. FWIW, I think he did score some good points. I myself have worked a lot with deep parsing methodologies, so I'm largely (but not entirely) on his side about the weaknesses of statistical methods. In principle, they cannot scale up to full language comprehension, and they do not really have much that is useful to say about how human beings process language.

 

[untermensche]

I understanding the difficulties in the determination of word boundaries. It is one place where Chomsky thinks statistical methods might have some pertinence.

That is one topic in which Chomsky is not criticizing those using statistical methods.

What Chomsky objects to is what he said.

It is absurd to say that possibly understanding something about word boundaries means all of language acquisition is about using statistical methods.

 

[Copernicus]

I understanding the difficulties in the determination of word boundaries. It is one place where Chomsky thinks statistical methods might have some pertinence.

That is one topic in which Chomsky is not criticizing those using statistical methods.

What Chomsky objects to is what he said.

It is absurd to say that possibly understanding something about word boundaries means all of language acquisition is about using statistical methods.

Not exactly. He said that you had to add other structural information, e.g. prosody, in order to get significant results, and that was the point. Structural analysis is necessary to supplement the statistics, but it still doesn't fully scale up. His claim was that you need to bring "UG" to bear in order to do real machine learning when it comes to language processing. I certainly agree with him that structural information is necessary. My problem is with his longstanding practice of treating the psychological "grammar" as somehow neutral and distinct from language production and perception strategies. He has no general theory of psychology within which to embed his psycholinguistic theory. This fact has been pointed out to him repeatedly for decades, but he wants to have his cake and eat it, too. He keeps assuming that psychologists will come up with a general theory of mind to vindicate his own framework. The problem has always been that people with a more general perspective on subjects such as biology and psychology have not found his vague ideas about some kind of special internalized "grammar" very helpful.

 

[untermensche]

I understanding the difficulties in the determination of word boundaries. It is one place where Chomsky thinks statistical methods might have some pertinence.

That is one topic in which Chomsky is not criticizing those using statistical methods.

What Chomsky objects to is what he said.

It is absurd to say that possibly understanding something about word boundaries means all of language acquisition is about using statistical methods.

Not exactly. He said that you had to add other structural information, e.g. prosody, in order to get significant results, and that was the point.

What you say is true but it is not the point. The point is the claims made about statistical methods in the paper Chomsky referenced are insane.

My problem is with his longstanding practice of treating the psychological "grammar" as somehow neutral and distinct from language production and perception strategies.

Is visual acquisition a psychological event?

Why would anybody think language acquisition had anything to do with psychology?

It is a biological phenomena like vision acquisition. An ability one acquires just by exposure and not by any effort.

Children need no training in language. They need to learn labels but that is almost instantaneous at a certain point.

But the syntax comes naturally just through exposure. Not by practice or training or effort. Or any psychological component.

 

[Subsymbolic]

What you say is true but it is not the point. The point is the claims made about statistical methods in the paper Chomsky referenced are insane.

My problem is with his longstanding practice of treating the psychological "grammar" as somehow neutral and distinct from language production and perception strategies.

Is visual acquisition a psychological event?

Why would anybody think language acquisition had anything to do with psychology?

It is a biological phenomena like vision acquisition. An ability one acquires just by exposure and not by any effort.

Children need no training in language. They need to learn labels but that is almost instantaneous at a certain point.

But the syntax comes naturally just through exposure. Not by practice or training or effort. Or any psychological component.

There are two issues with this fantasy.

First there's the issue of a neonate needing no training. There are literally hundreds of studies that show that the sort of baby talk that infants are exposed to has a clear tendency to be just a little ahead of the child's own development. This 'motherese' tends to have exaggerated intonation, simplified grammatical structure and vocabulary, combined with repetition, questions and rehearsal of key aspects of language. This just happens quite naturally, without planning and so it's easy not to notice just how much effort, practice and exposure a child is involved in. The fact is that as some poor sods discover, simply planting a child in front of a TV or radio does not allow them to develop language. It's interaction with carers.

The second issue is simply what this means. A bit like religious arguments about design, while it's the only game in town, it's easy to ask how else this could happen. This worked ok until the mid eighties and papers like this:

https://stanford.edu/~jlmcc/papers/PDP/Chapter18.pdf

Which heralded the arrival of connectionist architectures that could be trained. Chomsky's only real game in town claims died the day that the first neural net committed the classic 'ED' overgeneralization while being exposed to a training sets involving a range of regular and irregular past tenses. That this overgeneralization came about as a result of the internal and unprogrammed reorganisation of the neural weighting to allow generalisation across the trhe training set demonstrated clearly to anyone with the competence to grasp the idea that simple pattern completion across a network could achieve the same results as following rules, but without rules. That it happened regardless of learning algorithm, from backprop to Boltzmann, GA to RBF.

That was the point at which Chomsky's work on language acquisition lost whatever shreds of relevance it had left. Quite apart from anything else, the minor detail that the brain is massively parallel and not a serial rule following computational entity should have been a warning to the poor chap. Sadly, rather than see the writing on the wall, he has chosen to cover his eyes and ears and carry on with an increasingly pointless defense of an idea that has no future.

As I said earlier, my old doctoral supervisor ripped Chomsky apart a few years ago. As the original is rather long and not available on the internet, here she is responding to Chomsky's response to it.

http://www.sciencedirect.com/science/article/pii/S0004370208000970

Section IV page 1952

Quite apart from anything else, she comments on how bloody rude Chomsky is. I know how she feels.

 

[untermensche]

First there's the issue of a neonate needing no training. There are literally hundreds of studies that show that the sort of baby talk that infants are exposed to has a clear tendency to be just a little ahead of the child's own development. This 'motherese' tends to have exaggerated intonation, simplified grammatical structure and vocabulary, combined with repetition, questions and rehearsal of key aspects of language. This just happens quite naturally, without planning and so it's easy not to notice just how much effort, practice and exposure a child is involved in. The fact is that as some poor sods discover, simply planting a child in front of a TV or radio does not allow them to develop language. It's interaction with carers.

I have only three words.

Poverty of stimulus.

The acquisition of syntax is accomplished with an incredibly sparse amount and many times poor quality of data.

It is not acquired through learning. There is no possible way it could be.

That is beyond dispute.

And I do not know about television learning. I suspect it is possible. But it is unethical and cannot be tried. Humans need emotional support and human contact to develop strong controls over emotions, not for language acquisition. For language acquisition, like the acquisition of vision, all they need is exposure to acquire the ability. Even very little exposure which could not possibly explain the result.

Bad opinions about Chomsky are common. Actual understanding of his work is rare.

 

[holist]

I'll go with the definition I gave in the other thread:

Self consciousness requires the unification of two varieties of consciousness: being something that it is like something to be and being able to conceive of yourself as yourself. The former is a necessary condition for the latter, but both are necessary for self consciousness - which is what most people mean by consciousness.

But I'm still not keen of soundbites. Consciousness is difficult and trying to do difficult things quickly generally doesn't get you what you want.

Hullo everibad.

I like the soundbites.

I think consciousness is the cream on the cake.

As an inherent possibility, it was there from the very beginning of this universe.

Somewhat recently I was given another soundbite: "God is the fact that we are possible."

It is in a similar sense that consciousness to me is the great mystery. I wouldn't be surprised if it was pretty likely to arise on any sufficiently combinatorially rich substrate at the right kind of sweet-spot on an energy gradient, riding the edge of chaos into complexity.

Consciousness really makes life worth living, anyhow. I tend to think of it as upper berth.

I'll keep reading in a bit. Right now, I'm busy celebrating.

 

[Copernicus]

First there's the issue of a neonate needing no training. There are literally hundreds of studies that show that the sort of baby talk that infants are exposed to has a clear tendency to be just a little ahead of the child's own development. This 'motherese' tends to have exaggerated intonation, simplified grammatical structure and vocabulary, combined with repetition, questions and rehearsal of key aspects of language. This just happens quite naturally, without planning and so it's easy not to notice just how much effort, practice and exposure a child is involved in. The fact is that as some poor sods discover, simply planting a child in front of a TV or radio does not allow them to develop language. It's interaction with carers.

I have only three words.

Poverty of stimulus.

The acquisition of syntax is accomplished with an incredibly sparse amount and many times poor quality of data.

It is not acquired through learning. There is no possible way it could be.

That is beyond dispute.

Again, you completely misunderstand what Chomsky has been saying. If you listened to him in the video, you would have caught where he said that learning from stimulus was absolutely necessary. The thing you are missing here is that nobody who studies this topic seriously disputes the fact that language is instinctual in humans. What is in dispute is what is learned from experience and what is biologically inherited. Chomsky also jumps to a conclusion that some of us have serious doubts about--that the purpose of the linguistic "grammar" is to generate well-formedness intuitions. A more behavioral functionalist approach to language (and one that I favor) is that intuitions are a product of a general introspective capacity that predictively models sensorimotor behavior. So the evolutionary "purpose" of the psychological grammar is not to enable intuitions of grammaticality, but to produce and understand linguistic expressions. Such intuitions are simulations produced by a more general cognitive function. (One of the questioners in the video seemed to be thinking along these lines, but he didn't pose his question in quite the way I would have.)

I suspect that most of what I say bounces off your defensive shields, unter, but maybe others who bother to follow the discussion are interested. I'll just make one more remark about a flaw in Chomsky's approach to universals. He referred typological universals, but he used the somewhat dated expression "Greenberg's universals", referring to the linguist who laid some of the seminal groundwork. An example of such a universal (which Chomsky should have mentioned to help his audience understand what he was talking about) is the tendency of verb-first languages to have prefixes and prepositions, and verb-last languages to have postpositions and suffixes. Another would be that it is very common for verb-first languages to place modifiers after the words they modify and for verb-last languages to place modifiers before the words they modify. (I have greatly oversimplified this, but you get the idea I hope.) Chomsky also thinks in terms of other types of universals--e.g. his principles of binding theory. The problem is that he has no principled method for separating biological universals from accidental ones. So, if a nuclear war wiped out every linguistic community except for a community of Amharic speakers, then every feature of Amharic would become a linguistic universal overnight. However, not every feature would be biologically determined. And that is a big problem for him, because the dispute has never been whether language was to some degree innate, but what aspects of it were innate or not innate. It is not enough to merely say that certain universals are evidence of UG. One needs to be able to say which universals (if any) are evidence of UG. Otherwise, his claim becomes a "just so" story.

 

[holist]

First there's the issue of a neonate needing no training. There are literally hundreds of studies that show that the sort of baby talk that infants are exposed to has a clear tendency to be just a little ahead of the child's own development. This 'motherese' tends to have exaggerated intonation, simplified grammatical structure and vocabulary, combined with repetition, questions and rehearsal of key aspects of language. This just happens quite naturally, without planning and so it's easy not to notice just how much effort, practice and exposure a child is involved in. The fact is that as some poor sods discover, simply planting a child in front of a TV or radio does not allow them to develop language. It's interaction with carers.

I have only three words.

Poverty of stimulus.

The acquisition of syntax is accomplished with an incredibly sparse amount and many times poor quality of data.

It is not acquired through learning. There is no possible way it could be.

That is beyond dispute.

And I do not know about television learning. I suspect it is possible. But it is unethical and cannot be tried. Humans need emotional support and human contact to develop strong controls over emotions, not for language acquisition. For language acquisition, like the acquisition of vision, all they need is exposure to acquire the ability. Even very little exposure which could not possibly explain the result.

Bad opinions about Chomsky are common. Actual understanding of his work is rare.

I have only one word. But I ain't gonna say it.

In other news: a child raised without exposure to linguistic phenomena will not speak. A child whose exposure to linguistic phenomena is significantly delayed will have severe language-acquisition developments. How does that square with the "no learning possible and hence necessary" theory, Mr. Untermensche?

 

[Treedbear]

...
A more behavioral functionalist approach to language (and one that I favor) is that intuitions are a product of a general introspective capacity that predictively models sensorimotor behavior. So the evolutionary "purpose" of the psychological grammar is not to enable intuitions of grammaticality, but to produce and understand linguistic expressions. Such intuitions are simulations produced by a more general cognitive function. (One of the questioners in the video seemed to be thinking along these lines, but he didn't pose his question in quite the way I would have.)
...

I know nothing about linguistics, but it should be obvious that metaphors play a large role in the evolution of language, within a culture at least. Which leads me to believe that the brain is essentially a metaphor generating machine (the "more general cognitive function" you mentioned). That's what I mean when I suggest that the brain creates models of its environment and that these models evolve in order to adapt. How it does this I don't know, but it might be the key to understanding many higher brain functions. This would provide survival benefits to the organism. But in order to also function as a driver of brain development there needs to be an adaptive advantage for the brain itself. In that respect I've suggested that this ability must provide efficient adaptations which minimize energy utilization which generates excess heat. As with all complex information processors heat is the limiting factor. Increasing brain size and complexity therefore requires some feedback mechanism to modulate functions which generate heat. I imagine there are many. Some more or less localized in effect than others. But overall I'd categorize them as effecting the anxiety level. The release of anxiety inducing compounds might have the effect of turning off some ancillary processes in order to allow the affective ones to resolve local conflicts and inefficiencies. I'd even go so far as to say that all that we refer to as "feelings" are based on some primordial sense of anxiety which has taken on the "color" of some specific experience. That would narrow down what needs to be explained when we refer to experience, and possibly simplify how it might eventually be explained from an objective point of view. However this might be expressed psychologically, the physiological model is that the brain (and indeed brains in general) creates models and concepts that compete and evolve in the same sense as do biological organisms. I've been puzzled as to why more isn't made of this in brain research since we've become aware of the difference between brains and electronic computers in that they are so massively parallel. It should be obvious what this implies to any biologist who has studied ecosystems.

 

[Subsymbolic]

First there's the issue of a neonate needing no training. There are literally hundreds of studies that show that the sort of baby talk that infants are exposed to has a clear tendency to be just a little ahead of the child's own development. This 'motherese' tends to have exaggerated intonation, simplified grammatical structure and vocabulary, combined with repetition, questions and rehearsal of key aspects of language. This just happens quite naturally, without planning and so it's easy not to notice just how much effort, practice and exposure a child is involved in. The fact is that as some poor sods discover, simply planting a child in front of a TV or radio does not allow them to develop language. It's interaction with carers.

I have only three words.

Poverty of stimulus.

The acquisition of syntax is accomplished with an incredibly sparse amount and many times poor quality of data.

It is not acquired through learning. There is no possible way it could be.

That is beyond dispute.

And I do not know about television learning. I suspect it is possible. But it is unethical and cannot be tried. Humans need emotional support and human contact to develop strong controls over emotions, not for language acquisition. For language acquisition, like the acquisition of vision, all they need is exposure to acquire the ability. Even very little exposure which could not possibly explain the result.

Bad opinions about Chomsky are common. Actual understanding of his work is rare.

There are shorter ways of telling me you didn't read the article. The fact is that the training sets were both degenerate (in both the senses Chomsky required) didn't have explicit correction and didn't contain all the information needed. The whole point was that here was a system that when given a poor stimulus, reproduced the behaviour of human learners.

Sadly, there have been plenty of kids who were sat in front of the telly from an early age before social services got involved. While their ability to learn language is horribly impaired, their ability to see is not. There's a reason that Chomsky isn't taken terribly seriously these days.

 

[untermensche]

Where is your link to the study that looked at children only exposed to television?

 

[Subsymbolic]

Where is your link to the study that looked at children only exposed to television?

Where's your link to me using the word only?

However, you don't know something and I am supposed to do a research project?

I did you the favor and told you about something you had no idea existed.

Do a little work yourself.

 

[Copernicus]

...
A more behavioral functionalist approach to language (and one that I favor) is that intuitions are a product of a general introspective capacity that predictively models sensorimotor behavior. So the evolutionary "purpose" of the psychological grammar is not to enable intuitions of grammaticality, but to produce and understand linguistic expressions. Such intuitions are simulations produced by a more general cognitive function. (One of the questioners in the video seemed to be thinking along these lines, but he didn't pose his question in quite the way I would have.)
...

I know nothing about linguistics, but it should be obvious that metaphors play a large role in the evolution of language, within a culture at least. Which leads me to believe that the brain is essentially a metaphor generating machine (the "more general cognitive function" you mentioned). That's what I mean when I suggest that the brain creates models of its environment and that these models evolve in order to adapt. How it does this I don't know, but it might be the key to understanding many higher brain functions. This would provide survival benefits to the organism. But in order to also function as a driver of brain development there needs to be an adaptive advantage for the brain itself. In that respect I've suggested that this ability must provide efficient adaptations which minimize energy utilization which generates excess heat. As with all complex information processors heat is the limiting factor. Increasing brain size and complexity therefore requires some feedback mechanism to modulate functions which generate heat. I imagine there are many. Some more or less localized in effect than others. But overall I'd categorize them as effecting the anxiety level. The release of anxiety inducing compounds might have the effect of turning off some ancillary processes in order to allow the affective ones to resolve local conflicts and inefficiencies. I'd even go so far as to say that all that we refer to as "feelings" are based on some primordial sense of anxiety which has taken on the "color" of some specific experience. That would narrow down what needs to be explained when we refer to experience, and possibly simplify how it might eventually be explained from an objective point of view. However this might be expressed psychologically, the physiological model is that the brain (and indeed brains in general) creates models and concepts that compete and evolve in the same sense as do biological organisms. I've been puzzled as to why more isn't made of this in brain research since we've become aware of the difference between brains and electronic computers in that they are so massively parallel. It should be obvious what this implies to any biologist who has studied ecosystems.

I agree with your remarks here. Human cognition is fundamentally experiential and associative, and that has to do with the way the "connectionist" hardware works. IOW, the brain serves as a guidance mechanism for the body, so it strives to integrate information from its peripheral and central nervous systems into models that predict future changes in the environment (situational awareness) and its host body (self-awareness). Animals have evolved brains because they are organisms with brain-based guidance systems. Plants don't have brains, largely because they don't have moving bodies. Brains require a large amount of energy to develop and maintain, so plants don't really need them. Their environment is far less chaotic during their lives.

Metaphors are very complex associative structures that allow us to "understand" new experiences by associating and comparing them with existing experiences. I can recommend a very good book on the role of metaphor in language: Lakoff and Johnson's 1980 monograph Metaphors We Live By. It is a fairly short book that requires almost no technical expertise in linguistics to understand. Lakoff followed that up in 1987 with a much more detailed tour de force entitled  Women Fire and Dangerous Things. Since then, he has written numerous books and articles on the role of metaphor in language.