Here’s a short post linking to what I think is a very
accessible summary of Chomsky’s current views about language and its biological
basis. The discussion is not extended and the cognoscenti will not likely learn
anything new (though I did). Here are
some points he touches on:
1. Distinction
between generalizations about language and UG, which is “the genetic basis for
language.
2. That
there are no real group differences (vs individual differences) between humans
as regards FL/UG, implying that there has been no significant change in FL for
a very long time.
3. Early
theories of UG allowed for a huge amount of variation between languages. Over
the last 40 years, theory has narrowed the range of this difference.
4. A
simple statement of the goals of a useful theory: “A plausible theory has to
account for the variety of languages and the detail that you see in the surface
study of languages – and, at the same time, be simple enough to explain how
language could have emerged quickly, through some small mutation of the brain,
or something like that.”
5. Analogy
between UG and Jacob’s Universal Genome hypothesis.
6. Argument
for species specific native capacity for language starts from the simple
observation that humans alone can “pick out anything that is relevant to
language” from the “great blooming, buzzing confusion” that is the stimulus
input.
7. Fast
mapping of words to meanings indicating taht Quine’s “museum myth” is in fact
reality.
8. Pattern
recognition insufficient for language acquisition.
9. Theory
of Mind orthogonal to language acquisition problem.
10. Linguistic
interests are different from Languistic ones.
11. Is
the idea that culture influences language meaningful?
The distinction in (1) is particularly important to
reiterate as there has been rampant confusion on this point. Chomsky’s views
are not Greenberg’s and a lot of the criticism of Universals has come from
running the two together. I also liked
the observations concerning the implications of research on autism for what
look like Tomasello-like views about language development. The remarks are
blunt, but raise a relevant point.
I also found the observations concerning how little UG has
apparently changed over the last 30,000 years important to stress (2). Recall the
Papuans (and the Piraha?) who were very isolated until recently are all capable
of learning the same languages in the same way as anyone else. In fact, I know of no group of people whose kids suitably located cannot learn any language, on contrast to any other animal. Why? because humans have essentially the same UG and other animal's don't have one. This is sufficient to raise the generative research question: what do we have that they don't and how did it get there?
Similarly culture changes have left UG pretty much intact (11),
so far as we can tell. Older varieties of English, Icelandic, Japanese look
from a UG vantage point pretty much like their contemporary counterparts, indicating
that the same UG operated then as does now. Chomsky develops these points more elaborately elsewhere, but if you are like me and people ask what you do then this is
something short and readable to give them, if, of course, these are the questions you are
interested in.
One weird thing about this interview is when Chomsky says:
ReplyDelete"In biology it was plausible quite recently to claim that organisms can vary virtually without limit and that each one has to be studied on its own. Nowadays that has changed so radically that serious biologists propose that there’s basically one multicellular animal—the “universal genome”—and that the genomes of all the multicellular animals that have developed since the Cambrian explosion half a billion years ago are just modifications of a single pattern. This thesis hasn’t been proven, but it is taken seriously."
The only reference I could find to the notion of a 'universal genome' is from a paper by a biochemist called Michael Shermer (http://www.ncbi.nlm.nih.gov/pubmed/17660714). The idea doesn't seem to be taken seriously at all, if this is indeed the hypothesis Chomsky has in mind. In fact it seems like a bit of a crank paper (http://deepthoughtsandsilliness.blogspot.ca/2007/09/re-cell-cycling-front-loading-pt-i.html).
Am I missing something here? Is there a serious literature on the notion, or was the analogy too good to pass up and nobody did their homework?
I think that Chomsky is thinking of the evo-devo stuff. It is not a universal genome but there is a hugely conserved system with different switches responsible for the huge variation attested. The fact that eye genes from flies can function when transplanted into mammals is very suggestive. I think that this is what he is thinking of. At least he often refers to this work.
ReplyDeleteI think it is pointless to comment on Chomsky's amateur biology. Nic M. is right that the work Chomsky cites is not taken seriously, and biologists are able to tell you why.
ReplyDeleteAn answer by Chomsky that reveals that he does not take his audience seriously representative for many, is:
Q: Clearly, culture influences and shapes language, even if it doesn’t determine it.
A: That’s a common comment, but it’s almost meaningless. What’s culture? Culture is just a general term for everything that goes on. Yes, sure, everything that goes on influences language.
What do we learn from this answer? At best that Chomsky has a disdain for anyone suggesting culture has an important influence on language [and anyone who has read Chomsky's work already knows this]. Certainly a man as brilliant as Chomsky cannot fail to notice that his answer is at least as meaningless as the claim he attacks.
"Culture is just a general term for everything that goes on" - this is either a platitude or plainly false.
"Everything that goes on influences language" - this is either a platitude or plainly false.
Language is a fascinating subject of inquiry, Chomsky has a brilliant mind - it is really sad to read answers like the above...
Culture definitely shapes languages (to some extent) but not linguage ;).
DeleteThere is no evidence that I know of that Culture has an effect on I-language; the properties of the generative procedures that humans acquire. So, UG is not culture sensitive, so far as I know. It's not like democracies tolerate Gs that violate islands and that cultures that use garlic extensively allow string linear transformations. Nope. None of that. So far as we can tell regardless of the I-langauge acquired, they meed the same conditions and look the same. Of course, culture affects vocabulary. But I have never seen anything indicating that it affects the fundamental principles.
DeleteAgain: let me be clear: Culture can affect which particular G gets developed (after all, the Norman conquest did change "English"). this is expected given the program. And it can clearly affect how you use this G in expressing your the thoughts. But it has had no discernable effect on UG or FL. Chomsky has defined his area of interest very carefully. Many say that to do what Chomsky is interested in doing we must pay attention to culture. So far as I can tell, Chomsky is right to think that so far as 'culture' has any meaning there is no evidence for this at all.
Umm, this looks like a different concept of i-language than, say, the one in Knowledge of Language, p22 top, where the i-language is what is acquired by the speaker that allows him to frame new sentences, which certainly is a part of 'culture' (which I would think of as everything that an animal learns by participation in a community of animals, including dogs teaching each other how to knock over garbage bins, but not stuff that bears figure out for themselves but never transmit).
DeleteShifting the meaning of 'i-language' in this way is bound to cause a lot of confusion.
It seems neither one of you has paid any attention to Chomsky's definition of culture:
ReplyDelete"What’s culture? Culture is just a general term for everything that goes on"
Now if that is true, biology is part of culture because biology is part of "everything that goes on". And hence culture as defined by Chomsky shapes language [even I-language].
This was my point: Chomsky's definition is either a platitude or plain false. Both of you have a different definition of culture in mind but without knowing what it is i cannot tell whether or not you're right. Is culture a product of our biology or not? That seems a fair question given that we all are biological creatures. So if language is shaped by our biology why is culture not - or is it?
When Chomsky wrote that the claim "Clearly, culture influences and shapes language" is almost meaningless I had hoped he would provide a meaningful definition for culture and then show why culture so defined can not have any of the impacts Norbert describes. But no such luck, just a platitude. He might have said at least: "Culture is just a general term for everything that has no discernable effect on UG or FL"
So, Christina, since you think the given answer was pointless, it would be interesting to know what your answer would have been to the question/statement "Q: Clearly, culture influences and shapes language, even if it doesn’t determine it."?
DeleteAlso, not that I want to dissect every word that Chomsky utters (given that I think what he means and argues for is sufficiently clear to me), but as I understand that answer, he seems to be attributing that definition of culture to the general/common understanding of the term. Why would he want to define it precisely if he didn't think it was meaningful in the first place.
If you assume incoherence in his answer, then of course, your interpretation stands up. However, if you assume coherence in his answer, then other reasonable interpretations crop up.
Thanks for your questions. I do not have a favourite definition of culture. If I had I probably would have just done what Norbert did and inserted mine in the big empty space generated by Chomsky's remark. There are just two problems with that strategy:
Delete1. Norbert and I could have very different definitions of culture. Yet, if we never spell out what we mean by this term, we could end up in an endless pointless battle about whether or not culture has an impact on language. Since, hopefully, Norbert and I are not the only people who read Chomsky's remarks this problem multiplies for possible additional definitions of culture. There is a long history of Chomsky complaining that others have misinterpreted what he said. For this reason i prefer to go just by what he said and not to pretend I know what he meant.
2. Chomsky is famous for challenging definitions that 'everyone agrees on'. May I remind you that until he came along most people thought they knew what 'language' is. Chomsky has claimed that the 'common sense definition of language' refers to an uninteresting epiphenomenon: E-language. He gives some reasons for why he thinks it is misleading to talk about languages like German or Dutch. Instead we ought to talk about I-language as defined by him. So if this is the case for language what makes you think it is not the case for culture? Maybe the general/common understanding of the term 'culture' is an equally uninformative definition for E-culture and we really ought to talk about I-culture? Maybe we have not only an innate language faculty but also an innate culture faculty and the two do not interact? But this would be speculation on my part so for this reason I have to stick to what Chomsky says.
Now what answer would I have given to the question/statement "Q: Clearly, culture influences and shapes language, even if it doesn’t determine it."?
Presumably something like: "There are several definition of 'culture'. Can you please define the term so our audience knows what my answer refers to." What I would have expected from an expert like Chomsky is providing the definition he goes on to criticizing. "X claims culture is A,B,C,D. Now here are 4 reasons why A,B,C,D can have no discernable effect on UG or FL..." This would have been a service to his audience and allowed them to agree [or disagree] in a meaningful way with him. As it is our agreement or disagreement is based on prior commitments and has nothing to do with the vacuous remark Chomsky made. If you want detailed reasons for why I am not convinced by Chomsky's recent published views you may find this interesting: http://ling.auf.net/lingbuzz/001592
I am afraid, the writing you refer to at the end presupposes equally uncharitably that he is incoherent. Those who think he is coherent seem, at least to the extent that I can asecrtain, to have no trouble interpreting any of the material you discuss in your paper. So, again, in some sense you are essentially arguing for something that you have pre-supposed.
DeleteI wonder rather genuinely, if the stuff that you regularly disagree with is more a reflection of our (yours and others) biases, and far less a reflection of the content of the discussion.
As for your answer to the particular question, I agree with you - yours would have been a clearer way of dealing with it.
As for this claim "There is a long history of Chomsky complaining that others have misinterpreted what he said" - do you have any references to it? One of the regular complaints against Chomsky is that he doesn't engage his critics at all, to the point of ignoring what they have said.
I wish it would not be so easy to provide the references you asked for but below is a small sample of people who, according to Chomsky, have failed to understand his arguments. If these experts were not able to understand Chomsky's claims it would be quite foolish of me to insert what I think he may have meant with the above cited remarks instead of taking them at face value.
Delete“Searle has also missed the central point...” (Chomsky, 1975, p. 215). “Lakoff presents a very confused picture of the issues that have been under discussion” (Chomsky, 1973). “[Cooper] has, however, ... seriously confused the issues” (Chomsky & Katz, 1975, p. 71). “McCawley has missed the point of my remarks” (Chomsky, 1980, p. 47). “The post-1979 shift that Pullum perceives is imaginary... Pullum's conclusions are
based on serious misunderstanding throughout” (Chomsky, 1990, p. 147). “Godfrey- Smith cites only the last phrase quoted, misreading it as ... exactly the opposite of what the passage unambiguously states” (Chomsky, 1995, p. 2). “Putnam and others ... reject...’the innateness hypothesis’ [which they attribute to me] ...I have never defended it” (Chomsky, 2000a, p. 66). “...there is a huge literature arguing against the innateness of language; there’s nothing defending the thesis” (Chomsky, 2000b, p. 50). “[In Boden’s account] every reference to me ... is fanciful, sometimes even bringing to mind Pauli’s famous observation ‘not even wrong’” (Chomsky, 2007, p. 4).
As for the incoherence charge, someone way more linguistically competent than me {Paul Postal] has argued repeatedly that Chomsky's position is incoherent. You will note that there is no reference to Postal in the above quotes, so maybe we can assume he does not misunderstand Chomsky. A refutation has not been issued by Chomsky. This suggests to me that he is unable to refute Postal's arguments. Whatever one can say about or against Chomsky, one can certainly not accuse him of treating intellectual opponents with velvet gloves....
Re C: Whatever the length of the history of Chomsky's complaints, they are not nearly as long as they ought to be. Many have and many will continue to misinterpret him in the future. The relevant question is not if he has been misinterpreted but whether there is a reasonable interpretation of his views. In this case there is. Chomsky's point is that insofar as the term 'culture' has content there is no reason to think that it has any serious effect on I-langauge. To the degree that it's interpreted broadly, there is nothing to discuss as no real claim is being made. That's Chomsky's point. Is he right?
DeleteLike I mentioned before, to me it looks like he is. There is no good evidence I know of that shows that cultural input shapes UG. There is reason to think that it shapes the lexicon (as Chomsky notes in the piece) and one might claim that it shapes the Phonology of particular Gs. So, if we take I-language to be FL/UG there is no evidence that culture matters.
Of course if one's interests are in how particular Gs get used, then culture matters a lot. Why some particular Gs have particular kinds of agreement may also be influenced by culture (think honorificaton). However, UG? No evidence whatsoever. So, it looks like once again, when his position is not mangled beyond recognition, Chomsky is right.
If one were to be unreasonable, one could claim that all you have shown is that he has stated that others have misunderstood what's been said. I don't see any "complaining", which surely involves more than just stating.
DeleteHow is the above style of discussion anything more than just trolling for the sake of it. At least to me, it misses the forest for the trees.
And the argument from authority (Re: Paul Postal) is not very convincing. There are many more at least as knowledgeable as Paul Postal, who have argued to the contrary.
You are right that if it were correct that these others actually have misunderstood him, then he would merely be stating facts. But these others dispute that they have misunderstood him. They claim he distorts their view. And in some cases they appear to be right. But this is not really relevant to my point. There is a long list of distinguished people who either have misunderstood Chomsky or have been accused of misunderstanding him. I do not want to be in either group - so I stick to what he says.
DeletePlease provide reference to published work in which the "many more at least as knowledgeable as Paul Postal" have refuted Postal's arguments concerning the incoherence of Chomsky's biolinguistics view.
http://facultyoflanguage.blogspot.ca/2013/03/going-postal.html
Deletehttp://facultyoflanguage.blogspot.ca/2013/03/guest-post-jeffrey-watumull-on-postals.html
""What’s culture? Culture is just a general term for everything that goes on"
ReplyDeleteNow if that is true, biology is part of culture because biology is part of "everything that goes on". And hence culture as defined by Chomsky shapes language [even I-language]."
What Chomsky seems to have omitted in that sentence (but probably even birds understand by now what he meant) is that everything that is EXTERNAL to biology is essentially a cultural construct, involving groups of individuals; whereas I-language is part of biology, and therefore unaffected by culture. In the context of I-language biology determines culture, whereas culture cannot determine biology.
What you have got here is nothing but potpourri that doesn't even smell nice, because you took out one line out of a decades-long very clear general line of argument and misinterpreted it.
"...that Chomsky has a disdain for anyone suggesting culture has an important influence on language [and anyone who has read Chomsky's work already knows this]."
Saying that culture can determine how concepts and events are related in our heads is like saying that hardware of a computer can be shaped by its software. It's a fallacy. The software must conform to the way in which hardware operates, otherwise one would be able to install windows8 on a PC made in the 90s, or apes could learn human languages. So, what you call "disdain" I see as strong argument-based criticism. A very warranted criticism. That it sounds like disdain to some is directly related to how ludicrous it is to even suggest that culture determines the computational component of grammars.
Lets leave the insults in the coat roam, shall we? They are especially unbecoming when one has one's foot as far in the mouth as you do.
DeleteYou claim: "What Chomsky seems to have omitted in that sentence (but probably even birds understand by now what he meant) is that everything that is EXTERNAL to biology is essentially a cultural construct, involving groups of individuals; whereas I-language is part of biology, and therefore unaffected by culture. In the context of I-language biology determines culture, whereas culture cannot determine biology."
I am sure Chomsky is not as dumb as this suggests. To take just one example that refutes your categorical claim: Domesticating animals that provide us with milk is probably uncontroversially part of culture not part of our biology. Yet in the process some humans lost lactose intolerance [which clearly is part of our biology or hardware in your terminology]. If you are interested in other ways in which our biology can be and has been affected by things that are EXTERNAL to biology I recommend
Current Anthropology Volume 48, Number 1, February 2007 39
The Evolution of Human Speech Its Anatomical and Neural Bases
by Philip Lieberman
Further, until you can actually spell out what the computational component of grammars is [something Norbert claimed a few blogs back no one can currently do] it is at least premature to accuse those who suggest culture may have some influence on language as ludicrous. If you can provide arguments that are backed by credible evidence I am very interested in learning about them. But if you just want to insult those who disagree with you I suggest you find someone else to interact with.
You're omitting key parts of the text. You have done it for the second time now in your responses to this post alone. Read carefully what I said "In the context of I-language biology determines culture, whereas culture cannot determine biology." Key words: IN THE CONTEXT OF I-LANGUAGE. In what universe can a specific claim about I-language be taken to incorporate lactose intolerance? How did you manage to infer from my comment that I made such a far-reaching claim is beyond me...
DeleteYou were claiming culture can not determine biology. I was giving you an example where it does. Unlike you I do not believe in I-language, at least not as long as no one can tell me exactly what it is. Each time I ask for clarification I am told my questions are unreasonable. The work Chomsky has done on language has nothing to do with biology. The examples Norbert uses to show that 'connectionists' do not have a satisfactory account of crucial phenomena have nothing to do with biology. Yes we know that something in our brains underwrites language but we do not have the tools yet to discern whether it is domain specific LAD/FL or domain general *intelligence'. So essentially I am asked to take it on faith that there is I-language. That would be fine if it would be treated as one hypothesis among others. But that is not the case. Chomsky acts as if anyone who does not accept his view is a complete moron. Norbert tells me I have to be charitable interpreting Chomsky's work. Fine but that should be a two way street. Where is Chomsky's charitability in this passage:
Delete"There are some very strange ideas out there. For instance, a lot of quite fashionable work claims that children acquire language because humans have the capacity to understand the perspective of another person, according to what’s called theory of mind. The capacity to tell that another person is intending to do something develops in normal children at roughly age three or four. But, in fact, if you look at the autism spectrum, one of the classic syndromes is failure to develop theory of mind. That’s why autistic kids, or adults for that matter, don’t seem to understand what other people’s intentions are. Nevertheless, their language can be absolutely perfect. Furthermore, this capacity to understand the intention of others develops long after the child has mastered almost all the basic character of the language, maybe all of it. So that can’t be the explanation.
There are other proposals which also just can’t be true, but are still pursued very actively. You read about them in the press, just as you read things about other organ- isms having language capacity. There’s a lot of mythology about language, which is very popular. I really don’t want to sound too dismissive, but I feel dismissive. I think these ideas can’t be considered seriously."
Chomsky does not give a single reference to 'very strange ideas', 'fashionable work'. If I had to take a guess i would think he refers to Tomasello's work here but i am not sure because it would be 'mangled beyond recognition'. The view that ToM alone explains language might be absurd but no one holds such a view [just as no one holds the bizarre views about evolution Chomsky attacks in Science of Language'.] Here the barrage of accusations continues: "There are other proposals which also just can’t be true, but are still pursued very actively. You read about them in the press, just as you read things about other organisms having language capacity. There’s a lot of mythology about language, which is very popular"
Not a word WHAT the other proposals are or WHO pursues them actively. Now you take offense when i do not take everything you write into account in my reply. How do you think researchers who have spent decades working on language might feel when Chomsky dismisses their work as 'mythology' without discussing any of it?
I do not want to accuse Chomsky lightly of double standards so maybe you can explain to me why it is okay that Chomsky attacks without providing any reference to actual work that his audience could check? And why is it not okay when his critics provide detailed reference to passages he actually wrote?
This comment has been removed by the author.
ReplyDeleteI have also to protest for the unnecessary harsh words on Christina Behme. Not that she needs anybody to defend her, since, as it´s obvious, she is very much capable to defend herself, but in general there is no need to get virulent in any discussion, less of all, when she has been polite and restrained in her comments here.
ReplyDeleteI have two things to comment on what she said, though.
First, the issue of Chomsky´s work not being relevant to biology.
Well, if by that you mean that, in general, biologists would not advocate for a course on formal grammar on their core curricula, well, you may be right. But that´s not the level on which the claim is made. In order to include the study of language as a part of biology, Chomsky has to redefine biology. And in fact, that is what he has done. It is pretty clear that language is not an organ in the same sense than the stomach (you cannot put it in a formaldehyde jar, let´s say). For language to be an organ, you have to redefine organ. And that´s precisely what Chomsky has done. Chomsky claims that language has sufficient functional and structural "autonomy" to be considered an organ, even if we do not yet know its particular morphology (which body tissues correspond to it, if any). In that sense, saying that "we do not have the tools yet to discern whether it is domain specific LAD/FL or domain general 'intelligence' " is not really an argument against (or even about) Chomsky. An argument against Chomsky would be showing that language is not autonomous, that ALL of their functions are accomplished by other cognitive components (or, worse for Chomsky, by elements external to the cognition).
Notice that, otherwise, we could use the same argument against the very existence of "general intelligence" since it is true that "we do not have the tools yet to discern" if it exists---but of course people do take the general behavior of human beings as evidence that there is something that we could call “general intelligence”. Precisely, Chomsky´s point is that we DO have the tools to know if language is different from "general intelligence" or not, just that the tools are different from the ones needed to take, for instance, the temperature of your liver. We can study language structural properties and make correct inferences from that (more or less in the same way we knew that life was not possible in Mars before sending the rovers). If that´s true, then Chomsky is extending the realm of biology, even its methods. Not bad for an "amateur".
It is indeed very much possible to disagree with this extension of biology. But, as mentioned by other commentators, Chomsky points out to certain research programs inside biology that have a good chance to connect with his own project (the evo-devo program). And there are indeed many people working on making connections between the two (like Cedric Boeckx, just to drop a name).
Of course, you may dispute the autonomy of (I-)language, and many people try to falsify it. I hope they do---if that ever happens, I may have something to blame for the balky behaviour of Spanish clitics, but for the time being, I’ll continue attempting a Merge-Interfaces account.
---My second observation goes in the next post (not enough space here)
First, thanks for your kind comment, it certainly is appreciated.
DeleteSecond, let me be very clear that the criticism of Chomsky I issue is in virtually all cases specific to the work [or interview] I cite and should not be understood as implying I think Chomsky has never made any worthwhile contribution. Au contraire, he certainly has.
Now let me try to address some of the points you make above:
1. I have said Chomsky himself has not worked in biology, not that his work has not inspired any work by biologists. When attempting to define UG Chomsky has often invoked work BY biologists but he has not himself done such work nor do I think he has anywhere claimed he did. The discovery that not all organs can, as you say, be 'put it in a formaldehyde jar' has been made by physiologists quite independently of Chomsky. So it is misleading to say Chomsky redefined biology. What he has done is attempting to incorporate some of these new insights in his own work. This is no criticism of his contribution.
2. It is uncontroversial that humans have both: 'generai intelligence' and 'linguistic abilities'. What is controversial is how these two are related to each other. But for anyone who adopts a full fledged naturalism both need to be eventually accounted for in neurophysiological terms. I do not think even Chomsky believes that language is completely 'autonomous' [that you could have a human who has zero 'general intelligence' but 100% language].
3. If language is strongly autonomous [so we justifiably can call it an organ] from the biologist's perspective we would expect to find some indication of this organ: specific cells, enzymes, genes etc. found no where else in the animal kingdom. Remember Chomsky has always stressed that language is a species specific property. For this reason, according to Chomsky, we cannot do any comparative research on animals as we can with the visual or the digestive or the immune systems. But what we can do is searching for structures in the brain that are specific to humans and, if strong autonomy is right specific to language. Such a research program could be carried out to confirm or disconfirm Chomsky's hypothesis. My point was that at the moment we do not have enough evidence to decide one way or the other.
4. I am aware of the work of some biolinguists who attempt to apply findings from say evo-devo to language. I am not very convinced by that work and can supply you with reference to work by others who draw very different conclusions. But this is a digression since it has no relevance to the claims of Chomsky I was discussing.
5. If you encounter difficulties to account for 'the balky behaviour of Spanish clitics' within the biolinguistic framework it might be a good idea to look at work done by linguists who work in another framework. There is a rich body of work by now suggesting that "We can study language structural properties and make correct inferences from that" is problematic in so far that not all the predictions made in the past have been panned out. For some linguists that was motivation to look for other frameworks. Depending on how far you are willing to venture from Minimalism you may want to have a look at Culicover/Jackendoff's "Simpler Syntax" [OUP, 2005] and more recent work by these two. You could take also a serious look at Geoff Pullum's work or at Paul Postal's "Edge-Based Clausal Syntax" [MIT Press, 2011]
I am glad you are familiar with Postal's work. I did not read the MIT-published 2011 version, only the 2007 manuscript. As always, he does a great job on putting together very interesting facts, and makes a serious effort to relate then through a very formal algorithm (I particularly like the first chapter of Postal 2004, and his book on extraction, 1998). As for clitics, as Postal himself duly reports, he relies on (well-known) generalizations established by others---which in turns are based on analyses related to the dative alternation.
DeleteThe same can be said about Culicover & Jackendoff 2005, even if they said a lot less about clitics than Postal, merely plugin in Perlmutter's 1971 on an linearization-based account. Both moves are possible in a biolinguistic framework (in fact, I would say that at least Jackendoff is a biolinguist).
And since there is a much, much richer GB-Minimalist tradition dealing with Spanish clitics, both empirically and theoretically, I'll stick with that line for now.
If we cannot have a human being with zero "general intelligence" and 100% "language", it will be of not consequence, even if we assume a narrow, fully autonomous, notion of language. Most certainly, nobody who considers the bones an organ will expect a human being with only bones and no flesh.
I am not aware of any physiologist redefining the notion of organ to incorporate language independently of Chomsky---I know that Chomsky 1980 (and some previous works by him) uses Richard Gregory's works on vision to draw a parallel with language, but, as far I know, it is Chomsky's work what was dubbed "the new organology” (not always in a flattering way). Of course, I would appreciate reference showing something different
I agree that we do not have sufficient evidence to accurately describe how the brain morphology with respect to language is. But I don't think anybody disputes that. Chomsky's point is to claim that we can speak of language as an organ, based on what we know about how language works. That does not deny that plausibility of other kind of work, but you cannot disregard biolinguistics based on the fact that our knowledge about the brain is incomplete.
To answer you last question first: many people who reject biolinguistics do so based on what we know about language, not based on the fact that our knowledge about brains is incomplete. You will have noticed that no one replied to my request to provide a refutation of Postal's ontological arguments that show the internal incoherence of biolinguistics. If these arguments cannot be refuted it does not matter what we learn about the brain - biolinguistics is impossible. That Platonism is very unpopular at the moment does not prove it is false. Heliocentrism was very unpopular at one time. Also note that there are many people who reject both: Postal's platonism and biolinguistics. Again, many of them do so because of what we know about 'how language works'. Their arguments may not convince you. But there are certainly alternatives to the false dilemma you construct.
DeleteI am not sure what your example with the bones is supposed to show? No one disputes that bones and flesh are different 'organs'. We know that because skeletons are 100% bones, 0 % flesh. That they are not living organisms is another matter but they surely exist. Many people dispute that language is an independent organ. And since no one thinks language is the kind of organ that bones are [at least I hope no one does] I have no idea what you were trying to establish?
As you say Chomsky has introduced the notion that language is an organ long time ago. According to Science of Language he was doing biology from the very beginning [though some people who worked with/under him in the 1950s have no recollection of such early biolinguistics]. But even if we assume biolinguistics started in the mid 70s by now there should be bioligists outside Chomsky's framework who have incorporated his findings in their work just as biologists incorporated the findings of Watson and Crick. Except now Norbert tells us that Chomsky believes linguistics is at a level genetics was before Watson and Crick. So apart from the theory [which by now has been downgraded to a program] that language is a biological organ - what is the actual work in biology that backs this theory? Norbert tells us he has no training in biology. That has not prevented him from becoming a successful linguist with an impressive publication list. Can you name anyone who works as successfully on the visual system but has no training in physiology?
The second issue has to do with the relation between Chomsky and his critics. It is unfair (to say the least) to suggest that Chomsky does not engage his critics properly. First, long and detailed debates with Chomsky and his critics are well-known: with Quine, with Piaget, with Foucault, just to mention three where there are extensive records and discussions (and all of them are relevant for the internal/external debate). There is even a book called "Chomsky and his critics", where he writes individual responses; and of course, there are hundreds of long and short responses on Chomsky´s work, discussing with Wittgenstein, Searle, Kripke...(without mentioning the LEGIONS of his own students who disagree with Chomsky in many particular details of grammatical theory). Of course, I do not think this is news for Christina Behme. I am sure she knows all of this already. Precisely that is why I said that it is not fair to say that Chomsky does not engage his critics properly. He does.
ReplyDeleteOf course, maybe Christina means that Chomsky does not answer each and every critical comment in every interview, not failing to accurately indicate pages and full references. Shame on him, indeed. Maybe he catches up in his next reincarnations.
On the other hand, sometimes it seems that Chomsky may have some acceptable excuses for not spending too much time in at least some critical comments. Let´s take, for instance, this criticism:
"one finds misattributions and distortions: "a very good English philosopher wrote a paper about it. [it = Everett’s work on Piraha, CB]. It’s embarrassingly bad. He argues that this shows that it undermines Universal Grammar because it shows that language isn’t based on recursion. Well if Everett were right, it would show that Piraha doesn’t use the resources that Universal Grammar makes available” (Chomsky, p. 30). The very good English philosopher informed me that he had not written an academic paper but an 800-word book review for The Independent (Papineau, p.c.). This review (Papineau, 2008) does not contain anything substantiating Chomsky’s charge" (Behme 2013: 4).
It is a horrible sin, indeed, to confuse a paper with a review. Shame on Chomsky, he should pay closer attention to what he says in the middle of an interview. But the critical remarks in Behme´s review (or paper?) do something that it may be considered, well …worse. It cuts half a paragraph, and gives to Chomsky´s answer a complete different meaning. I recover the complete paragraph here:
“There is a lot of discussion these days of Dan Everett´s work with a Brazilian language, Pira---it´s described in the New Yorker, among other places. David Pesetsky has a long paper on it with a couple of other linguist, and according to them, it's just like other languages. It´s gotten into the philosophical literature too. Some smart people--a very good English philosopher wrote a paper about it. It’s embarrassingly bad. He argues that this shows that it undermines Universal Grammar because it shows that language isn’t based on recursion. Well if Everett were right, it would show that Piraha doesn’t use the resources that Universal Grammar makes available”
As it should be obvious, there is NO qualification whatsoever with respect to the English philosopher´s review. There is no “charge” against him. “It” in “It’s embarrassingly bad” refers to Everett´s book. “He” in “He argues that…”does not refer to Papineau but to Everett. Chomsky is saying nothing about Papineau, except that he wrote a “paper” (which was really a review: ts, ts, bad Noam!) on Everett´s book.
So my question here is the following.
What should Chomsky do to confront this kind of criticism? Calling the author to explain what is the right reference of his pronouns? Well, I guess it’s hard to be Chomsky.
just an update that by now i have added a section to my review that deals with the ambiguity issue you first drew attention to: http://ling.auf.net/lingbuzz/001592
DeleteThank you for this comments. let me address them in reverse order.
ReplyDelete1. You are the first one telling me that you believe 'it' refers to Everett's book not Papineau's review. Papineau said in his review indeed that Everett's work may undermine UG and he showed me e-mails he received after the review had been published, accusing him of making a bad argument. So it was not clear to me [or to anyone whom I showed the review and asked for feedback] that Chomsky was talking about 'Dan Everett's work'. We were all mislead to believe that 'it' referes to the paper he had just mentioned and 'he' to the English philosopher who so argued. Given that both interpretations are legitimate I would think it is the obligation of the author [or his editor] to make sure that no confusion arises. It would not have put a huge burden on either to use 'Everett's work' instead of 'it' to remove all doubt - if that was indeed what Chomsky intended to say. But I will make a footnote in my review to this effect. So thank you again.
2. I have no where claimed that Chomsky has never responded in a satisfactory manner to any of his critics. So the citation of interchanges with Quine, Piaget or Foucault is not relevant to my claim that in the cases i cite he has not engaged with critics in a satisfactory manner. [note though, that even in the cases you mention there is room for doubt. Quine is on record saying "Chomsky's remarks leave me with feelings at once of reassurance and frustration. What I find reassuring is that he nowhere clearly disagrees with my position. What I find frustrating is that he expresses much disagreement with what he thinks to be my position [Quine, 1969, p. 302)]
3. In the cases I cite my criticism was that Chomsky fails to identify clearly who his opponent is, what their position is and why this position is wrong.
4. Whether Chomsky has in some cases distorted the view of a critic needs to be analyzed case by case. i am prepared to give specific examples where he has done so. And as one example for his failure to engage with explicit and detailed criticism lets take
"Chomsky's Foundational Admission" by Paul Postal, posted on LingBuzz in July 2012 [http://ling.auf.net/lingbuzz/001569 ]
Recently Norbert posted a comment suggesting the passage containing the admission has been misinterpreted. But note that Postal provides also a "brief discussion of the incoherence independent of the admission, documenting various contradictions in Chomsky's foundational writings".
So first, even if Norbert is right and there was a misinterpretation only part of the criticism has been addressed. Second, it is not really Norbert's task to address a criticism directed at Chomsky. Especially not a criticism that has been formulated in detail decades ago and repeated numerous times. Avoiding to deal with it directly also seems to contradict what Chomsky says here:
"I was told...I should stick to my own work and leave other people alone. But that struck me as anti-intellectual counsel" Noam Chomsky
[quoted by Harris, 1993, p. 51]
While the context was different it seems very relevant here; not responding to a serious criticism to once's position strikes me as anti-intellectual behaviour.
an afterthought on the Everett/Papineau issue. I have re-read the passage several times and think my interpretation was correct. In fact I hope it was correct because if you are right, then it would have implications for Chomsky that are worse, not better than if I am right.
ReplyDeleteIn either case the criticism remains, that Chomsky should have named Papineau and said the work by him was a book review not a paper. I admit that his editor shares blame here, he could and should have supplied these details if Chomsky forgot to mention them.
Now if 'it's embarrassingly bad' refers to Everett's work, then what is the evidence that Chomsky supplies to support this very harsh verdict? He refers to the 2007 paper by Pesetsky et al. but fails to mention that Everett has replied to this paper: The Shrinking Chomskyan Corner: A Final Reply to Nevins, Pesetsky, Rodrigues [2010] http://ling.auf.net/lingbuzz/000994
Chomsky can of course disagree with the arguments Everett provides there but then he ought to specify what he disagrees with. Instead he simply neglects to mention that this defence of Everett's work exists.
Chomsky further claims: "what Everett claims is probably not true" but again he gives no data supporting this claim. it is an unsupported accusation denigrating the work of another linguist. Further Chomsky provides an analogy to walking vs. crawling people which is irrelevant because it presupposes that UG as defined by Chomsky exists. But that is exactly what Everett questions. So Chomsky needs to provide an argument that establishes UG exists, he can not simply assume it exists.
So if you are right and 'it is embarrassingly bad' refers to Everett's work, it is a very harsh criticism that is not supported by sufficient evidence. It may well leave some [many?] readers with the impression that Chomsky attacks Everett because his own theory is threatened.
Now anticipating that i will be called uncharitable here, how charitable do we have to be? Chomsky is addressing a general audience that has little or no linguistic background knowledge. He can not assume that his readers are familiar with Everett's work and the reasons for why this work has been considered by some as a serious threat to Chomsky's theory. So at a minimum Chomsky should have explained what Everett's findings were, what Everett claimed to follow for UG and then he should have gone on to present counter evidence. Note that it is not relevant that someone like Norbert may know these details and would be able to fill in all the blanks Chomsky left here. Norbert does not need 'Science of Language' to learn what Chomsky thinks about Everett's work...
Chomsky is entirely correct. Everett's work is embarrassing. This has been discussed again and again. I discussed it in the first post that launched this blog and extensively in the comment sections of the Chronicle article. And my conclusion is quite independent of the Pesetsky et al discussion of the data. The problem is that even if all his data is correct it is entirely irrelevant to Chomsky's claims about UG. Everett and Chomsky are not using 'universal' in the same sense and thus all of Everett's discussion is a simple equivocation. Chomsky has repeatedly distinguished between his sense of 'universal' and Greenberg's (again in the little thing I recently linked to). If Everett or anybody else does not understand the difference the fault is not Chomsky's. So yes, Chomsky is right to dismiss this work out of hand as it is based on a pun, and though amusing once or twice it gets old very fast, especially when those who know better (or should) keep reviving this 'Zombie Idea'. So, to end, he cannot present counter-evidence as there is no debate!! None, zero, nada.
ReplyDeleteThis may all be correct [I disagree but lets leave this aside], but it misses the point of my criticism. In the passage in 'Science of Language" Chomsky has not provided enough information that would allow an audience that has little or no background knowledge in linguistics to evaluate whether or not his criticism is justified. I know it is very difficult to read the passage and pretend you do not know all the details you know. But that is what you have to do when you evaluate whether there is enough information for the intended audience. You cannot assume Chomsky's audience has read your or David Pesetsky's comments in the Chronicle debate. Some people may have read the name Everett for the first time in Chomsky's book. Would they know what the criticism is about? You have provided in your short paragraph more information than Chomsky did in his published work. He had an editor and CUP prides itself being a scholarly press. They should have been able to come up with something better than what was printed. That is my point and it has nothing to do with whether or not Everett's work is bad.
DeleteThis book is intended for those who want to know what Chomsky thinks. His brusque evaluation suffices for these purposes. Is it as didactic as it could be? No. Are there other places where Chomsky is more didactic. Yes. Starting with 'Language and Mind,' Chomsky has written accessible didactic expositions of his work. Those interested in not only what Chomsky thinks, but why he thinks it have many resources to reach for. Nonetheless, perhaps your view is no matter the venue, nor the intent of the piece, Chomsky must satisfy the most stringent conditions? Nope. This is a set of interviews. I found them very useful for many "inside baseball" issues. But this is not a book for the neophyte. It is a book for those who know something and are interested in fleshing out areas of ignorance informally. Not every book must be written for the same audience and for the same purpose. And to repeat, Everett's work was really bad and given the fact that it was widely discussed and disseminated entitles Chomsky to a peremptory dismissal.
DeleteI think you are asking from the editor something that you yourself failed to do. You also failed to indicate that the passage was ambiguous. Even worse, you cut the part that provided the relevant interpretation! (and that's the reason why everybody who reads only your review agrees with you). Of course, assuming good faith, you actually failed to notice it. Furthermore, there is a far cry between accusing a book of being ill-edited, and accusing its author of "misattributions and distortions" as you said in your review with respect to this very case.
DeleteAnd there is more. You accuse Chomsky and his editor of not having read Lassiter 2010 and express surprise for them saying that they don´t know if Lohndal & Narita 2009 will be published. Perhaps you did not read page 3 of the Introduction, where the editor says that the interviews on which the book is based were conducted in 2004 with the last one on January 2009!!!
So how charitable do you have to be? Well, what about as charitable as you are with yourself? No...a second thought...I think Chomsky will be fine even if you are a bit less charitable than that. Just don't overdo it.
I am afraid I have to plead not guilty here. As you say I have interpreted the passage different than you. We do not know who is right. But it is irrelevant because [i] this was one example from many and [ii] the misattribution in this case was calling a book review an academic paper.
ReplyDeleteAs for the accusation that I was uncharitable about the Lassiter case I am afraid you are not entirely correct either. May I draw your attention to the bibliography, p. 309. Here you will find under McGilvray, James reference to a 2010 publication. On p. 310 is a reference to a 2011 publication by John Mikhail. So McGilvray updated the bibliography even after 2009, he just ignored the Lassiter paper which would have been quite relevant to the attack on Lassiter. Further, on p. 309 is also reference to the Lohndahl & Narita paper - not in press or forthcoming but a reference with exact page numbers - so McGilvray was quite aware that the paper had been published. The publication date of Science of Language was 2012 and there is good evidence that McGilvray continued to work on the reference section well after the last interview. Would it really have been such a difficult task to add a footnote that Lassiter had published a reply? Or to even MENTION Lassiter's name? Why is there no reference even to the original paper by Lassiter which must have been already published at the time of the interview (Chomsky referred to it)?
No, it is not just one example. We have already two.
ReplyDeleteBut I am glad you recognize that the only misattribution in the case you call irrelevant is to call "paper" something that it is a "review". I guess that means that your claim that "This review (Papineau,2008) does not contain anything substantiating Chomsky’s charge" is misleading since you did not actually attributed Chomsky any charge against Papineau, or at least you did that on the basis of a misleading quotation.
I am also glad you now shift the blame to the editor. Most certainly, if you publish an interview whose materials are 9 to 3 years old, you should update the references more carefully. But this is not the same than saying, as you do in your review/paper talking about Chomsky (no the editor): "He harshly attacked an author whose paper he knowingly distorted. This would be a reprehensible act no matter who commits it. But given the status and exalted influence Chomsky enjoys, it is outrageous that he would resort to such unprofessional behaviour to demean someone who disagrees with him". Rather, it seems that it is you who is harshly attacking an author whose opinion you knowingly distorted.
And it is not hard to find more examples. I’ll add one case in the next comment, since it would take more space than the one permitted here.
I must say you certainly know how to play the distortion game. Sticking to facts on the other hand seems not to be your strongest suit.
DeleteI wrote whether I interpreted Chomsky right in the Papinaeu case has no impact on the fact that it is misleading to call a book review a paper. IF I got Chomsky right it is also correct that Papineau's review does not contain anything substantiating the charge that it was embarrassingly bad. But since, I can imagine to be wrong I am willing to withdraw this statement. It does not change anything on the fact that in many cases Chomsky distorts what others say. Let me ask you: are you at least willing to entertain the possibility that it is your interpretation that is wrong?
The paper Chomsky distorted was Lassiter [2008], the one in 'Mind and Language' that he MENTIONS. I included reference to it, maybe you should read it before you hurl accusations of distortion my way? The way Chomsky dealt with THAT paper was reprehensible. Unlike the case with Everett where some people might be willing to accept Norbert's excuse [that the case is so well known that Chomsky can assume that some, maybe most of his readers know about it and may feel justified to offer a very brusque dismissal] I doubt that even among linguists too many people are familiar with Lassiter 2008. If you believe that it is acceptable to deal with Lassiter's 2008 paper in the way Chomsky did, then you and i have very different opinions about acceptable academic conduct.
The fact that before publication of Science of Language additional material had become available makes one wonder why Chomsky did not feel compelled to change his earlier statement. He did not drop dead but is extremely active even now as you easily can confirm on for example his facebook fan-page. Does he really care so little that his comments have the potential to be very damaging for someone who did not even deserve them in the first place?
Since you like to quote how about being less selective: "Professor Chomsky reviewed the edited and commented text and made many suggestions" [Science of Language, p. 3]. So Chomsky certainly had an opportunity to re-read the passage in question at a time when [i] McGilvray had added the reference to the Lohndal/Narita paper and [ii] the second Lassiter paper was available. He could have quite easily suggested to McGilvray that the passage needed some reworking given that McGilvray did not seem to think so. McGilvray includes of course the customary statement "For mistakes that remain, i am solely responsible" [ibid]. But i think we can safely rule out that Chomsky proposed to provide a more accurate account of Lassiter's work and McGilvray refused to include that.
Now since you seem to see no fault whatsoever with Chomsky's behaviour here, maybe you can provide some evidence from Lassiter 2008 that justifies the remark I quote:
"Some of the stuff coming out in the literature is just mind-boggling...The last issue [of Mind and Language] has an article - I never thought I would see this - you know this crazy theory of Michael Dummett’s that people don’t know their language? This guy is defending it." (Chomsky, p. 57)
Where does Lassiter [2008] say anything that is mind-boggling or crazy? Where is he even defending Dummett? Please provide page references for all three.
I am not sure what Lassiter said as I have never read the paper. But I am pretty sure that you misrepresented what I said just one comment ago. First, I did not provide an "excuse" for Chomsky. I simply noted that different books aim to do different things and that every time you mention someone you are not compelled to make sure that what you say is accessible to every kind of reader. Second, I did not say
Delete*that the case is so well known that Chomsky can assume that some, maybe most of his readers know about it and may feel justified to offer a very brusque dismissal*
or anything like it. I said that there are people, (e.gh. me) who are interested in what Chomsky thinks about some topic without requiring that he tell me right then and there why he thinks this. A book of interviews is just that, interviews. You learn some things and not others.
At any rate, I don't know what Lassiter said, but given how you interpret me, I would not be all that confident that you got him or Chomsky right.
I am sorry that you feel i misrepresent what you said. I had made an accusation [that Chomsky should have provided more detail for his Everett critique]. You have in response given several reasons for why you think what Chomsky that was quite sufficient.
Delete1. after saying how widely Everett's work has been discussed you say: "So yes, Chomsky is right to dismiss this work out of hand as it is based on a pun, ....
2. This book is intended for those who want to know what Chomsky thinks. His brusque evaluation suffices for these purposes.
3. But this is not a book for the neophyte. It is a book for those who know something and are interested in fleshing out areas of ignorance informally. Not every book must be written for the same audience and for the same purpose. And to repeat, Everett's work was really bad and given the fact that it was widely discussed and disseminated entitles Chomsky to a peremptory dismissal.
These are all reasons for why you think Chomsky was not required to say more than he did. I call this you provided an excuse for him. Maybe you think i mean this in a demeaning way? I don't and if you prefer you can read my comment 'Norbert provided reasons for why he thought Chomsky did not have to say more about Everett's work than he did"
I disagree with you about the value of Everett's work but I have no issue whatsoever with you stating that from your perspective Chomsky provided enough information. In fact it would not surprise me if you think Chomsky should not have mentioned that work at all and talked about something of interest to you instead.
Regarding Lassiter's paper I would suggest you read it and then judge if you agree with what Chomsky said about it. It is of course possible that i do not understand the paper [that's why i contacted him and asked if my interpretation was correct. he confirmed it was but feel free to do the same]. But as philosopher you know that even if it would be the case that i am wrong about it, this does not prove that Chomsky was right about it. Remember he made only 3 claims; it was [1] mind-boggling, [2] crazy and that [3] Lassiter defended Dummett's theory. I am only saying that these claims are false, not whether or not i agree with the content of Lassiter's paper.
Let's take your comments on the "Norman Conquest". This is what you quote from Chomsky:
ReplyDelete"Take the Norman Conquest. The Norman Conquest had a huge effect on what became English. But it clearly had nothing to do with the evolution of language - which was finished long before the Norman Conquest. So if you want to study distinctive properties of language - what really makes it different from the digestive system ... you’re going to abstract away from the Norman Conquest. But that means abstracting away from the whole mass of data that interests the linguist who wants to work on a particular language. There is no contradiction in this; it’s just a sane approach to trying to answer certain kinds of far-reaching questions about that nature of language"
And this is your characterization:
"If, when studying L1, one should abstract away from the whole mass of data of interest to the linguist about L1, the same logic would hold for L2....Ln. So one would have to abstract away from everything of linguistic interest about all languages to uncover the nature of language and explain how it differs from digestion"
There is a problem, though, of the same kind discussed before. There is an ellipsis in the quote you used, three dots with no brackets like this: ... It looks almost like it belongs to the original text. But in fact you took a phrase out of it. Here is the quote without the omitted phrase in capital letters (the bracket is in the original):
"Take the Norman Conquest. The Norman Conquest had a huge effect on what became English. But it clearly had nothing to do with the evolution of language - which was finished long before the Norman Conquest. So if you want to study distinctive properties of language -what really make it different from the digestive system- AND SOME DAY MAYBE [STUDY] THE EVOLUTION OF THOSE PROPERTIES- you’re going to abstract away from the Norman Conquest. But that means abstracting away from the whole mass of data that interests the linguist who wants to work on a particular language. There is no contradiction in this; it’s just a sane approach to trying to answer certain kinds of far-reaching questions about that nature of language"
First, it is crystal clear that Chomsky is NOT saying that “when studying L1, one should abstract away from the whole mass of data” about L1. He is saying that, when studying THE EVOLUTION OF THE LANGUAGE FACULTY, the historical data (which very much “interests the linguist”) per se is not very informative. You need to move beyond that, to some other level of abstraction. I do not see how this could be controversial: how do you study the evolution of language taking into account the Norman Conquest? Furthermore, as Chomsky clearly explains in the sentences previous to the ones you quoted, “abstracting away” is just to create the right conditions to make an experiment. It is quite obvious that if your theory does not work in ideal conditions, then for sure it will not work in real, messy conditions. That is NOT to deny that we need to also explain all the messy conditions in the real world. It’s just to adopt the methodological principle that it is better to isolate one property to study it more productively.
Of course, you could disagree with this stance, and have a different view of the scientific method. But then say so. Why do you have to resort to such a silly distortion of Chomsky´s view? It is not only that you crossed out a whole sentence just because it made his actual point clearer, but even in the parts of the quote you kept still says “it clearly had nothing to do with the EVOLUTION OF LANGUAGE” (I guess you forgot to take it out).
And all of this, in the very paper where you are accusing Chomsky of “misattributions and distortions”, and generously providing plenty of moralizing remarks. Well, let me use your own words: “this would be a reprehensible act no matter who commits it […] it is outrageous that [you] would resort to such unprofessional behaviour to demean someone who disagrees with [you]”.
I am rather amazed with how much certainty you accuse others of malice. Does it really not occur to you that possibly you could be wrong?
DeleteIn the case of the Papineau interpretation I was willing to give you the benefit of the doubt even though none of the roughly 100 people who sent me feedback on my review read the passage the way you do [most of them own the book]. I still find it not a very natural reading but it is not impossible to interpret it the way you do. Still it would never occur to me to imply you intentionally use a rather odd reading of this passage to make a point against me.
Here the situation is different. Chomsky mentions 2 research projects [1] and [2] below;
So if you want to [1] study distinctive properties of language -what really make it different from the digestive system- AND [2] SOME DAY MAYBE [STUDY] THE EVOLUTION OF THOSE PROPERTIES- you’re going to abstract away from the Norman Conquest.
Having a word limit for my review i dropped [2] because it is not relevant to my discussion. Why not? As everyone who is familiar with Chomsky's recent work knows, he believes that the study of language evolution is not a worthwhile task - at least not currently but quite possibly always beyond our cognitive grasp. There are countless passages to that effect in Science of Language and other publications. One of the best known arguments is that we do not even know how to begin to study the evolution of bee-communication which is by magnitudes easier. I hope you are familiar with it so we do not need to discuss it here.
What Chomsky wants to study NOW is "the distinctive properties of language" and for that study he suggests we need to abstract away from the Norman conquest. No where in this volume or elsewhere does Chomsky suggest we should study the evolution of the language faculty. In fact he provides many arguments supporting his conclusion that such a study would be misguided [at least currently] and he provides a speculation for how in his opinion the language faculty came into being: by one mutation in a single individual. It is not really relevant here what i think about these speculations but If it is not cristal clear to you that Chomsky has no interest in studying the evolution of the language faculty, I recommend you read the volume with more care.
You misinterpret the passage you cite. There are no TWO research projects (in this particular paragraph). There is one [1].
DeleteI repeat:
"Take the Norman Conquest. The Norman Conquest had a huge effect on what became English. But it clearly had nothing to do with the evolution of language [1] - which was finished long before the Norman Conquest. So if you want to study distinctive properties of language - what really makes it different from the digestive system -and [1] maybe some day [study] the evolution of those properties- you’re going to abstract away from the Norman Conquest. But that means abstracting away from the whole mass of data that interests the linguist who wants to work on a particular language. There is no contradiction in this; it’s just a sane approach to trying to answer certain kinds of far-reaching questions about that nature of language"
Although it is correct that Chomsky has been skeptical about research on language evolution, here he is trying to explain what goals we need to meet first in order to address the evolutionary issue. Obviously, for him, to have a relatively good understanding of the formal properties of language is one of those necessary goals. But even more, we need to frame the explanation in a way that will make it possible to compare the faculty of language with other organs. And that means abstracting away from a lot of historical data, which is, otherwise, very, very important.
Notice that the same can be said about the stomach. Historical/cultural events can indeed affect its evolution, as it is the case with lactase persistence--which is actually caused by an enzymatic change in the duodenum, but let's keep rolling (after all, if a review is a paper...). But you cannot come to this conclusion before knowing the kind of enzymes that are involved in the digestive process, or the role that enzyme lactase plays in the digestion of dairy products. But we do not need to even know any of the historical and cultural facts related to lactase persistence, even if those very facts where its direct causation. Even worse, we may not have ever known of such a neat interplay between history and biology, if we would have not do the biological job first..."abstracting away from the whole mass of data that interests [somebody] who wants to work on a particular [something]".
That's Chomsky's point. So, if this is not crystal clear to you, I would suggest that you follow your own advice and "read the volume with more care".
I am not accusing you of malice. Not, at least, in the sense that I believe you are cheating. I don’t believe so. What I am accusing you is of bias. You have already decided that Chomsky is wrong no matter what, so you look for every word, every sentence, and every half paragraph, which could be used against him. Like a prosecutor who is committed to hang the defendant no matter what, assembling a potpourri of misattributions and distortions.
I am stunned that you would insist to insert a meaning into this passage that would make no sense whatsoever in the context of Chomsky's discussion. Are you accusing him of ignoring context and McGilvray's question here? Section 13 is about 'Simplicity and its role in Chomsky's work' and McGilvray had prompted the reply we talk about by remarking "It's a complicated task disentangling all the various contributing factors, dealing with a child's course of development..." [p. 84]
DeleteWhat would a research program on language EVOLUTION have to do with THAT?? It would be pretty rude if Chomsky would ignore McGilvray's question and talk about what you suggest he does. Are you saying Chomsky is so rude? I am assuming he is reasonable and giving an answer that is relevant, namely one concerning the study of the distinctive properties of language. You have to give me very good reasons to believe that Chomsky is NOT engaging in a serious conversation with McGilvray here but instead talking about something that is of virtually no interest to him in general and has nothing to do with this particular context.
To repeat: on Chomsky's view the distinctive properties of language had evolved in a single mutation event some 50.000-100.000 years ago and they have not changed since. For that reason they were already in place at the time of the Norman Conquest and for that reason the Norman Conquest is not relevant to the study of THOSE properties. Your reference to the evolution of the digestive system is a nice diversion but you ignore that for Chomsky language did not evolve in the same manner as the digestive system did.
Further Chomsky clearly says; "If you want to study X AND some day maybe the evolution of X you're going to have to abstract away...." he further says "It's just a sane approach to trying to answer certain kinds of far reaching questions about THE NATURE OF LANGUAGE" [p. 84]
For your interpretation to hold Chomsky would need to go out of his way to confuse his audience. I would never accuse him of doing that and neither should you.
As for your accusation: There are countless passages in Science of Language that I could have used in addition to those I did. I had thought I had said enough but i certainly can add more if you want me to. I also note that you did not bother to answer the question I asked: "Where does Lassiter [2008] say anything that is mind-boggling or crazy? Where is he even defending Dummett? Please provide page references for all three." - I take this as indication that you are not able to establish that my specific claim about the Lassiter paper is a distortion. I think there is really no point in continuing this conversation.
"I am not accusing you of malice. Not, at least, in the sense that I believe you are cheating. I don’t believe so. What I am accusing you is of bias. You have already decided that Chomsky is wrong no matter what, so you look for every word, every sentence, and every half paragraph, which could be used against him. Like a prosecutor who is committed to hang the defendant no matter what, assembling a potpourri of misattributions and distortions."
DeleteHaving read all posts and comments in this blog in the last month or so I have come to the same conclusion. Hence my somewhat harsh comment above, which came after I have read many of Christina's posts. My apologies to Christina if she felt insulted. That was certainly not my intention. In any case, Chomsky deserves better critics. I recommend reading recent papers by Jan Koster ("Recursion and the Lexicon" is a good one) which contain good criticism of Chomsky's views with respect to any effects of culture on language. Maybe Christina can use those to develop reasonable argument-based criticism of Chomsky’s very clearly defined views -- in the interest of science and not to bring down Chomsky because he is Chomsky. As far as I’m concerned (and many other generativists I’m sure), such criticism would be welcomed.
Yes, sorry for not answering your question yet. It was in the previous trend, though, and it was made less than 24 hours ago.
DeleteI completely agree that Lassiter has been mistreated, and that Lassiter 2008 should have been properly cited. This has never been an issue here, though. My discussion (as you can check in the previous trend) was not about your comments on what the interview says on that paper, but about your comments with respect to Lohndal & Narita 2009 and Lassiter 2010.
Your questions about Lassiter 2008 are, to quote you, "a nice diversion," but it bears no relation with the specific comment I made. So, where is he even defending Dummett? Nowhere, but that was not under discussion.
Your other question is very informative about your rhetorical strategy. "Where does Lassiter [2008] say anything that is mind-boggling or crazy?" I have to admit, I love this one. First, "crazy" is not used in the original text as referring to something on Lassiter 2008, but it's a qualification on Dummett's idea. I repeat the relevant part of the quote:
"Some of the stuff coming out in the literature is just mind-boggling...The last issue [of Mind and Language] has an article - I never thought I would see this - you know this crazy theory of Michael Dummett’s that people don’t know their language?"
Second, "mind-boggling" or "crazy" are such a subjective predicates that we cannot falsify any proposition using them. If John says "the idea X is mind-boggling"...well, the only think I can say against it is "I don´t think so". Accusing John of distortion in this case would be...mind-boggling.
There are more things to be said, but since the discussion is starting to be about itself, it may be a wise idea to stop it, as you suggested.
Just a final comment.
Interviews are a cheap target for a full scale criticism against anybody. It will always be something missing, something ambiguous, some exaggeration, a self-indulgent comment, and so on. Interviews are for showing off. There is nothing wrong with trying to ruin the show they play, of course. There are not sacred cows (or at least, there should not be). But trying to ruin the show is just another show.
I am not trying to mount a personal attack on you---if it seemed like I was, I apologize. But I honestly think that the mistakes you find are editorial or characteristic of an interview. They are still mistakes, of course. But they do not justify the frontal attack you mount (or seem to mount) on Chomsky.
Let me make just one final comment since this discussion has gone on already way too long. The potpourri was a criticism of ONE BOOK by Chomsky not of his life work. i have stated that but maybe it needs to be said even more explicitly. Unlike Norbert I have taken the endorsements on the back of the volume seriously. Here we are told "Noam Chomsky has long been recognized as a founding father of Modern linguistics. These compelling and CAREFULLY organized interviews ILLUSTRATE WHY...The book is TRULY EXCEPTIONAL in affording an accessible INTRODUCTION to Chomsky's broad-based and CUTTING-EDGE theorizing. A must read" Robert Stainton
DeleteStainton is an expert on Chomsky's work. Having him say what he did makes the potential reader expect that 'Science of language' is as good as it gets regarding Chomsky's work, that he can't do any better. And that is simply not true! Chomsky can and has done much better than this. Implying otherwise is an insult on Chomsky! I had several people writing me that they bought the book based on the advertisements and were disgusted by the content [a lot more than me because unlike me they did not seem to know that Chomsky has done some amazing work]. If, as Norbert says above "this is not a book for the neophyte. It is a book for those who know something and are interested in fleshing out areas of ignorance informally" why was it not advertised as such?
I am not interested in the "many "inside baseball" issues" but had expected a book that is focussed on the SCIENCE of language and as such it is a disappointment. At the end of Chomsky's answer about what he takes to be his most important contributions I was almost in tears because that answer was so bad. I could not understand why he had said NOTHING about the amazing contributions he is famous for. Does he truly think, as some have suggested, none of them hold up? Why did McGilvray, who is intimately familiar with Chomsky's work, accept such an answer and did not say: Okay Noam, lets try this again, your readers want to know what your truly important contributions are? If interviews are, as Miguel suggests for 'showing off' do we have to assume there is even less of lasting value in Chomsky's work? I hope not!!
I have written the harsh review to make it cristal clear that Science of Language is NOT the place to look for Chomsky at his best. If it was intended as a show for Chomsky it is a very bad show for those who are outside the target audience that Norbert identifies. I leave it for you to decide what is better: to call a bad 'show' just that or to pretend it was a great show and risk that people who are not familiar with Chomsky's work judge his work by that show.
Interesting to see this discusion here, Christina and Miguel. Here are my $0.02: I doubt that there was ever any intention by Chomsky and McGilvray to misrepresent anyone, and yet that book is full of rather uncivil attacks predicated on misrepresentations and (in a few cases) obvious falsehoods. What's going on here? In my case, it looks like Chomsky read my abstract, saw that I was arguing against him, figured I must be arguing for salient opponents, and didn't bother reading the paper to find out that I was arguing against them too. The rest is typical confirmation bias: Lohndahl & Narita's paper got noticed because it confirmed the belief that I was one of the anti-Chomsky bad guys, but my response -- where I pointed out that Lohndahl & Narita had completely misread my position, in part due to a general lack of understanding of the background literature -- was ignored because it failed to confirm this belief. This is not very surprising: even great scholars like Chomsky are still people, and people are susceptible to these kinds of biases.
ReplyDeleteThe only thing that I do find surprising is that no one went and actually read my paper before publishing a one-liner which was both factually inaccurate (I think Dummett's position is totally hopeless, and spend around 1/3 of the article in question arguing this point) and unscholarly ("crazy", "mind-boggling", etc.).
This is just a really small point in a sizable book, of course, as is the memory lapse that led Chomsky to claim that David Papineau had written a journal article defending Dan Everett, etc. [Sorry, Miguel, but your clever re-reading of the passage relies on a pretty implausible resolution of the pronoun.] But, along with other inaccuracies and misrepresentations there, it makes one wonder about the continued value of doing interviews with Chomsky, or more generally of the methodology of doing linguistics or cognitive science by trying to find out what Chomsky currently thinks about various topics. Personally, I derive much more value from re-reading wonderful books like Syntactic Structures, Aspects, and Knowledge of Language.
Yes, there is no question that you have been mistreated in the book, your paper should have been properly cited, and your conjecture about how you ended up in that line is very plausible.
DeleteI also agree that interviews are not a very reliable source to find out the details on the contributions made by any author. Interviews are just interesting from a sociological or historical perspective.
By the same token, interviews are a cheap target for a full scale criticism against anybody. It will always be something missing, something ambiguous, some exaggeration, a self-indulgent comment. Interviews are for showing off. There is nothing wrong with trying to ruin the show, of course. There are not sacred cows (or at least, there should not be). But trying to ruin the show is just another show.
Don't worry, Dan. Even negative publicity is publicity. Before this show, I didn't know of you. Now two of your papers are waiting on my pile of to-be-read papers.
DeleteFWIW, the interpretation of the pronoun that Miguel indicated is the only one that I'd ever considered. It hadn't really occurred to me that the pronoun could refer to the philosopher rather than to Everett's work on Piraha.
DeletePoint taken. But since you never noticed an ambiguity you may hopefully grant that neither did I until Miguel mentioned it. I did not intentionally interpret a sentence I knew to be ambiguous in a way that fit my purpose [as Miguel implied]. I am already working on changing this passage in the review and am grateful that the ambiguity was pointed out to me.
DeleteThe above promised passage has been now added to the review:
Deletehttp://ling.auf.net/lingbuzz/001592
I agree that rereading these great books is always a good idea. I personally don't know anyone whose research strategy consists in reading Chomsky interviews, though I can think of many worse ideas. The one point I find refreshing is L's observation that Chomsky is like others and has foibles like the rest of us. What is surprising is that given the huge volume of his productive work just how high the general standard is. Given the obsessive level of scrutiny everything he says or does gets, it's amazing to me how well most of his views have held up. I know that most of what I have said or written would not hold up nearly so well.
ReplyDeleteLast point: the best parts of the M interviews for me were the historical reconstructions, e.g. Why we moved from eval theories to P&P accounts, how earlier discussions with biologists shaped C's thinking, etc. I don't know if I agree that the shifts were always improvements or if the analogies go through smoothly, but I do know that hearing C discuss what moved him and why has proven very stimulating. I even like this when I am on the receiving end of one of his pets points of view. Nothing has gotten to me to think harder about the movement view of control than knowing that C thinks there is something fundamentally flawed with it.
The core Postal’s (2009) argument seems to be that on p. 140 about “illegitimate equivocations on the concepts *idealization*, *recursive*, *rule*, and *generate*.” (my emphasis) Let’s consider *idealization*. Why not to consider infinite sets of sentences? If it helps, of course. Postal compares it to a solar system with an infinity of planets in a finite space. He rightly calls it a silly idea and so I wonder why he didn’t think of Achilles and the Tortoise instead. The occurrence of some utterances is more probable than of others and some (in fact far most) of them are extremely improbable. Provided the probability decreases faster than 1/rank, we can say that in a certain sense they can all be produced within a finite time. If I made you laugh read the next paragraph.
ReplyDeleteHe mentions only one example of idealization, again quite an irrelevant one, viz frictionless movement. But consider the canonical ensemble in statistical mechanics, where the system in question is supposed to be in contact with an infinitely big thermal bath. In the grand canonical ensemble, the ensemble average (of an open system) is assumed to pass through all its possible states far most are extremely improbable (eg. those with no or very little molecules or those with the numbers of molecules approaching infinity).
I can tell about the type-token problem, but I'd prefer that Christina first kindly prove the existence of the empire of ideas.
First, i am no Platonist so not really the right person to ask for any proof. Paul Postal has convinced me that Chomsky's biolinguistics faces a very serious foundational problem that needs to be addressed because it threatens the framework itself, not just some minor detail [that we could leave aside and hope to get fixed later]. Regarding Platonism he has only convinced me to keep an open mind.
DeleteNow i assume you wanted proof that abstract objects are real? [my apologies I do not like the term 'empire of ideas' because it seems to suggest that there is somewhere a place where these ideas hang out]. If by proof you mean what philosophers mean then i am not aware of one [again i am no Platonist]. But I recall a comment Jerry Katz made decades ago: "Unlike 'naturalism' Platonism is assumed guilty until proven innocent" What he referred to is the fact that we also do not have any proof for any of our naturalist scientific theories - yet we do not require them to provide proof before we accept them, we are content with convincing evidence.
Take for example biolinguistics. i would be very surprised if Norbert would tell me he had proof for it. What he has is evidence that convinces him it is the correct framework. [When i disagree with him it is not because he has no proof but because i draw different conclusions from the evidence]. You will have probably read by now that Chomsky talks about many 'promissory notes' and that Norbert said there are many details about how language is encoded in brains that we do not know yet. This is no reason to question biolinguistics [though there might be others]. That is how science works and as long as everyone is willing to look at all the evidence that is available there is nothing wrong with that.
So if you're content with evidence i can provide some for why i think we should not rule out Platonism but for evidence for Platonsim you really have to ask a Platonist. Now for your comments I find them entertaining but they do not convince me, so lets see what you have on offer for type/token.
I agree that demanding a proof was not fair from me. However, what I had in mind was this:
ReplyDelete(a) You need not believe to the biolinguistics enterprise to agree that there exists some innate bias, either domain specific, or domain general, that makes us able to learn language. And you have to admit that biolinguistics is a falsifiable hypothesis.
(b) We have problem with type-token issue and so we introduce an eternal empire of ideas (sorry) whose existence can be neither proved nor falsified. Moreover, I don't think this problem is that serious for we have at our disposal the empire of ideas as a cognitive construct after all.
I certainly hope that biolinguistics is seen as a falsifiable hypothesis by those who propose/defend it. Chomsky made recently comments that make me wonder. In 'Of Minds and language' he said: "You just see that some ideas simply look right, and then you sort of put aside the data that refute them and think, somebody else will take care of it’ (Chomsky, 2009, p. 36), this does not sound to me as if he is at least considering the possibility the ideas that simply look right could be wrong. No matter how brilliant Chomsky is, he is a fallible human being and as scientist he needs to be at least open to the possibility that his ideas could be wrong. So he needs to specify what would falsify them. I have asked at a conference a few years back what a biolinguist would consider a finding that falsifies his hypothesis. i got no answer. That worries me...
DeleteRegarding the confirmability of abstract objects, again i rely on Katz here. He said that it is a mistake to demand that the existence of abstract objects can be confirmed with the same methods as the existence of concrete objects. This makes sense to me. We may not have methods right now to confirm that Platonic objects exist. But remember, throughout history people have accepted the existence of objects that were unconfirmable at the time. Of course in many cases they were wrong but then many of the once 'accepted' findings of science turned out to be wrong as well. Platonists claim they have indirect ways to confirm the existence of abstract objects. Postal cites some of the properties of natural language. You cite some other properties of language that convince you UG exists even though at the moment we have no method to confirm UG in a human brain. Let me repeat, i am no Platonist but i see no reason to reject it off hand. To cite Einstein: "Condemnation without Investigation is the height of Ignorance" I think he had a point...
Is Platonism falsifiable? Is materialism falsifiable? Is monism (opposite of dualism) falsifiable? Could you specify what would falsify them? The bio-linguistics program qua program cannot be falsified, at least if dualism is not an admissible option, for it amounts to trying to discover the biological bases for language. Does anyone seriously believe there are none? Are there any dualists around? What is open for contention are the contents of a particular version of the program, e.g. Are there specified locality restrictions as part of UG, but the program itself cannot seriously be up for grabs.
DeleteThere are all too many ways for specific proposals to be falsified. Programs, not so much. They run out of steam. I get the impression that C would like those busily advancing the program to give up. I suspect this will be I possible until it becomes obvious to practitioners that they are inning their wheels. I wouldn't hold my breath, but of course that's up to you.
Why these hostile mind-reading attempts? I would be absolutely delighted if biolinguists would discover some brain tissue that is implicated in language say. At the moment i do not hold my breath on a Chomskyan UG because i do no longer know what that is. But if someone defines it in an unambiguous way and even better confirms it empirically that would be great. At the moment what you call 'empiricist work' seems more promising to me but they surely have not all the answers. So if Chomsky would provide a reasoned criticism of a specific proposal he certainly would have my full attention. For me science is not like a sports competition where I only want my 'home team' to win, it is about getting to the truth and i do not care who gets there first.
DeleteYou know as well as I that your first questions have negatie answers. we either accept or reject materialism or Platonism or dualism or... just as we either accept or reject the existence of God. Does anyone seriously doubt there are biological bases for language. Yes Postal certainly does because he defines language very differently from you. But he does not doubt that there are biological bases for a child's ability to acquire knowledge of a language. Do you believe there is a biological basis for Cantor's theorem? If not does that make you a dualist?
I employed “biolinguistics” as a label for the hypothesis that there is some non-empty FLN. Of course, the biolinguistic program does not depend on whether it's empty or not.
DeleteOkay. So the only way to disconfirm this hypothesis would be if we can show there is NO non-empty FLN. Demonstrating the non existence of something is difficult at the best of times, so it is important to have a detailed definition of what FLN is. But at least in principle you have a hypothesis that could be disconfirmed.
DeleteNow you also say the biolinguistic program does not depend on whether FLN is empty or not. This sets my alarm bells off because i can no longer see how, even in principle, I could disconfirm the existence of an empty FLN. I also have no idea how i would confirm the existence of an empty FLN. So again i have to ask, if you accept such a possibility [as you seem to], then why do you reject Platonism based on the claim that it is not confirmable or disconfirmable?
If Postal is a platonist then he believes that minds can see forms. If minds are brains, he is not a dualist, then he thinks brains are built to see forms. The question is how. What a Platonism rejects is that languages are merely biological objects, that their ontological status is merely neurological. A non platonist can accept the view that languages have no status beyond their biology. So, yes, everyone admits to a bioling program, even a Platonist (see Soames on this).
DeleteWhat is hard to interpret is what it means to ask if this is falsifiable. What would it mean to falsify the claim that human brains are built to acquire language. Given how self evident this is, almost nothing could. What would falsify the claim that dropped objects fall? Though this is not a logical truth, it is close to being self evident. So to with the idea that brains are built for language.
There are important differences between Postal and Soames' view but it is probably best you ask Paul whether he believes in minds seeing forms, I do not want to provide any misinformation...
DeleteYou are right that virtually no one would dispute the virtual tautology you describe. But you now have no longer a program that distinguishes you from say Tomasello's or even from Everett's. Given how much ink you* [pl] have spilled to argue against each other what is of real interest is how your program is different from that of say Tomasello or Everett. So you need to say a bit more than "human brains are built to acquire language".
* please do not take this the wrong way again, I do not have the time to check if you personally have ever argued against Tomasello but Chomsky certainly has so this is an inclusive 'you' not a personal one
Ok, we agree on the tautological nature of the bioling program. The disagreements? Over what the fine structure of this FL is. Over that there is lots of disagreement. I cannot imagine, however, that you think that these claims are not falsifiable, e.g. There are locality relations like subjacency, minimalist, binding domains etc. these are substantive claims. There are minimalist versions of these as well. So the details are up for grabs and this is where the disagreement lives, and there is a lot. But here we get into details. Chomsky has defended his details, others have argued against them, in other words normal scientific practice.
DeleteAfter reading Everett's latest book [a review of which is here: http://ling.auf.net/lingbuzz/001696] I would have thought the disagreement between him and Chomsky runs much deeper than disagreement about specific claims about locality relations like subjacency or binding domains. Are you saying this is not the case?
DeleteAre we really going down this road again. Everett's work is irrelevant to the bioling project as Chomsky envisaged it. I've discussed why elsewhere as I've noted before (see for example the first post on this blog. I would call this a disagreement if there was not an equivocation at its heart. You cannot disagree over a pun. So, I see confusion but little else. BTW, this is my last reply on this. Having run full circle indicates to me that there is little left to explore.
DeleteActually I was confused by your answer to my question how your and Chomsky's position differs from that of Everett or that of Tomasello. In your answer you provided a few examples about which Chomsky disagrees with other generativists. I deny neither that such disagreements exist nor that work on them is interesting. But they were not relevant to the question I asked.
DeleteIf you do not want to talk about Everett we do not have to. But maybe you can briefly outline the core differences between your framework and Tomasello's and also tell me what empirical findings could confirm these frameworks. In other words I am looking for an answer of the form; If X occurs Tomasello is right and we are wrong; If Y occurs we are right and Tomasello is wrong.
Chiming in briefly, if you can identify any linguistic phenomena for which "Tomasello's framework" has an account that competes with an account in "Chomsky's framework", there could be a discussion about "disagreements" of the sort you request. I have made some attempt to uncover examples of this sort, and have failed. Perhaps I haven't tried hard enough. As far as I can tell, however, no such examples exist, because Tomasello puts nothing specific about language structure on the table. Give us Tomasello's discussion of control vs. raising, Verb-Second, cross-linguistic variation in quantifier scope, the status of "need" vs. "have" across languages, syntactic locality restrictions of the sort that Norbert has been mentioning, or children's acquisition of Principle B -- i.e. any serious topic of empirical debate among linguists -- and perhaps we can start talking about "empirical findings" that can adjudicate among "frameworks". Do you know of any place where we can find such discussions?
DeleteThere are some attempts to take Tomasello's claims seriously at a higher level of abstraction. A recent exchange between David Adger and Adele Goldberg posted to LingBuzz comes to mind. Adger in turn cites several papers by Charles Yang, which I recommend to you.
David put the point well. And T's "disagreements" are not, sadly, all that uncommon. In one corner, there are concrete proposals on the table concerning what FL/UG would look like and why. In the other are claims that none of this is needed but there are no accounts of how to explain the relevant data. Other data is sometimes cited, but these are not incompatible with the position from corner 1 so even if T were right about these (he most likely isn't but that's another story) his claims are either irrelevant or unargued. This is the general state of play and I understand that to mean that contrary to appearances there is nothing to discuss.
DeleteC, why not offer an actual argument rather than more references. Put one together and we can discuss it. That would make moving ahead quite a bit easier.
Always tongue in cheek Norbert, it really starts growing on me. Clearly your last paragraph is a joke. You both tell me that Tomasello, who has been working on language for decades huge lab which guys that can do brain work and all the nine yards at his disposal, has after all this time put "nothing specific about language structure on the table" and provided "no accounts of how to explain the relevant data". Surely you cannot seriously expect a philosopher can, on a rainy afternoon, write up an account that would knock your socks off?
DeleteNow joking aside, I am surprised about the reluctance to answer what I thought were fairly basic questions. David has in the 2013 LSA plenary noted that editors of leading journals seem ignorant about the work you guys are doing. He has talked about a better world in which things would be different. Why not see this as an opportunity to work toward such a better world? I certainly have been very patient trying to gather information here. Why make me feel the whole time as if i am pulling teeth? Why not, as Miguel said so nicely, 'put on a show' and dazzle me?
What you have given me so far as reason to prefer your framework over Tomasello's seems like the Lassiter dismissal: "Here are a few lines of the abstract, doesn't look interesting to us so don't bother". There may be good reasons to reject Tomasello's work but you have not given me any. Judging by what I read he accounts for tons of data. Are these accounts flawed? If so why? Why should he account for the data you mention above? Will it be impossible without them to learn anything interesting about language? It would seem, given David's account of the situation in important journals. their editors believe work like Tomasello's is important while either they think yours is not or they do not even know about yours. Do you think the kind of evasions you offer me would convince any of these editors that they should publish more generative work? If you really think Tomasello and you play in completely different ballparks maybe you can answer my questions for Role and Reference Grammar (like say Van Valin & LaPolla 1997) - what are the core differences, what would be evidence for yours and against theirs?
C I cannot dazzle you because I don't think you know much. But here's a case:
DeleteControl has certain properties: it is licensed in certain positions (non-finite subjects, requires local cc antecedents, gets de se readings, only sloppy readings under ellipsis. Why? Here's a proposal: Control structures live on A-chains and hence is a movement dependency. Prediction:PRO sits in non-finite subject positions, must have a local c-commanding antecedent etc. That control lives on A-chains, let's say, is part of UG. From this I derive the relevant properties e.g. explain why 'John persuaded Mary to leave' does not mean 'John persuaded Mary that he should leave' but does mean 'John persuaded Mary that she should leave.'
So a specific data set, with specific properties and a specific proposal to explain them. Now let's go to T. What's his account of these data? He has none. So, I cannot debate with him about whose is better, can I? ow he does have some data that he thinks relevant: so he thinks that being able to suss out communicative intent is important for language. Ok, what specific property of linguistic structure does this explain? So far as I know, he does not say. Does locality follow from this? Does structural conditions on antecedence? Not so far as I can tell. Are there other accounts. Sure: Kayne has a different theory, as does Landau, as does Jackendoff. But, and this is important, though the details differ they all require quite a bit of native language specific machinery. In that sense, they are "friendly" alternatives to the standard generative account. Unlike T they believe roughly what I believe save for the details. T (and you) seem to think that this is nuts. Ok, why? Now, please do not tell me that some people think X and others Y etc. I am asking YOU to present YOUR reasons for thinking that I should take this seriously. Provide an argument. If you cannot, then it is unlikely that I can (or should try) to convince you as you don't know enough to evaluate the arguments. In my day, Philosophers could give arguments for positions. Is this no linger something philosophers do? So, take any of the analyses of binding, movement, control, and reanalyze them in terms that don't invoke quite a bit of ling specific capacity. When you've done this, we can debate how well it has been done. I have tried to show that this is rarely if ever done. Until it is, there is nothing to debate.
So, here's a simple recipe for arguing against Chomsky's/my/David's conclusions: take a well studied phenomenon that people have explained by showing that it seems to require rich parochial assumptions about linguistic structure. Reanalyze in terms not requiring these but based on more general learning procedures/rules etc. Do this for a reasonable range of the relevant cases and this will show that there is little reason to assume that there is a FL. I'll make it easier: do this just for GB as outlined in Haegeman. Reanalyze it in non ling specific terms. Do this and you kick out the struts behind the standard program.
Note, that this is what some minimalists are trying to do. It seems pretty clear to me that their results have not yet been that convincing. However, I understand this program. I don't understand T's because it does not deal with these data, the data used to argue for the interesting conclusion. Were it able to do this, it would go a long way to convincing me that my case was at best weak. THe lack of such persuades me that so far the case is strong. Your turn. Show me your cards.
[written before Norbert's reply appeared above]
DeleteChristina, you keep repeating that you're not a linguist. It's not trivial to become a linguist, but because of the youth of the field it's not the hardest thing either. If you want to argue linguistics with linguists, it might pay to pick up some expertise, so we can have a proper conversation.
My advice: temporarily stop reading Chomsky, and stop endlessly picking apart this one book of interviews. Instead, read a few standard syntax textbooks. I usually recommend a combination of Andrew Carnie's textbook and David Adger's when people ask me for self-study advice in the area of syntax, and that seems to be good combination. Given your specifically Chomskyan concerns, you should add to it the more advanced textbook by Hornstein, Nunes and Grohmann. Honestly, though this is a bit of work, it's not like studying Kant or String Theory. Given your background in philosophy, you will probably find it particularly rewarding to supplement this reading with one or more of the standard syntax-friendly textbooks in formal semantics, e.g. the famous textbook by Heim and Kratzer, or if you prefer, the rather different approach taken by Larson and Siegel.
Then, if you are still interested in the questions you have asked here, you will be in a position to look at the "tons of data" that Tomasello and his colleagues have allegedly accounted for and ask yourself if any of it is relevant to what you've learned from these textbooks. I think the answer will be a resounding "no", but at the very least, we will then be in a position to have an informed discussion about the topic.
Thank you for the suggestions, for the first time something concrete, I appreciate that.
DeleteNow to, as Norbert requests, put my cards on the table: It is true that i am not calling myself a linguist. But this certainly does not mean I have not read more than that one book of interviews by Chomsky. When I worked on my dissertation I have read EVERY book Chomsky has ever published [on linguistics I really do not care for his politics and skipped a few of those] + a good deal of his articles [I can send you the bibliography of my dissertation if you want proof]. How else would I know that many of the problems in Science of Language are not isolated occurrences but representative for quite a bit of his recent [though not his early] work?
If you believe that after this [which BTW was not the only linguistic literature I read, you can add quite a bit Postal, a lot of Pullum, some Jackendoff&Culicover, Partee, Boeckx, Uriagereka, Bever, Katz [though you may not count him as linguist?] and believe it or not even some Hornstein and Pesetsky to name a few] I am in no position to evaluate what you tell me you must think either [1] I am really stupid or [2] the life-works of Chomsky contain nothing that allows one to evaluate your proposals. I hope you think neither. In addition I exchange regularly e-mails with some linguists and should i really not understand something they have so far always been able to set me straight. I do not call myself a linguist because i do not think that reading no matter how many books by linguist from a philosopher's perspective makes me a linguist any more than reading the books by psychologists i read make me a psychologist or reading books about airplanes would make me a pilot. I know this is different from Chomsky who believes reading books about biology or by biologists makes him a biologist [at least he has called himself such numerous times].
What you [pl] mention above is for the most part internal to your theories [or program] so it is not terribly surprising that others focus on different phenomena. But this does not mean there is no way to compare what people who work in different frameworks are doing. If you send your papers to one of the top journals you mention in your talk and the editor sends you a note saying: "Look this stuff is very different from what we usually publish and seems to contradict what X, Y, or Z say, why should I publish this?" would you reply as Norbert did above; "I don't think you know much"? or would you give reasons for why your paper is every bit as good as the stuff that is usually published in the journal and find a way to explain the things I asked you to explain, most notably why your work is good even though it seems to contradict [or actually contradicts] their X,Y, and Z? I would really hate to end this conversation believing neither you nor Norbert know enough about Tomasello's work to answer the questions I asked
Christina, take into account that Tomasello has a diversified portfolio of interests in human cognition and behind (such as chimp’s theory of mind, social skills in dogs). He simply can't manage such details in grammar as these guys do.
DeleteIn his comment on Everett 2005, p. 640, he claims, as far as I can understand, that first comes a linguistic community with thematic structures in their heads and only then, in the course of communication, comes grammar through the grammaticalization of content words. If you wonder why, the answer is because “human cognition and communication work the way that they do”. (ibid)
The development of language sketched by him is roughly that found in children. But what does it mean that, in his view, linguistic community precedes grammar in language evolution? Basically that, in his view, language evolution followed the same trajectory as the Nicaraguan sign language did. One would be inclined to expect that UG must have preceded the Day One of the linguistic community. But it doesn’t seem to be his idea.
"If you believe that after this [which BTW was not the only linguistic literature I read, you can add quite a bit Postal, a lot of Pullum, some Jackendoff&Culicover, Partee, Boeckx, Uriagereka, Bever, Katz [though you may not count him as linguist?] and believe it or not even some Hornstein and Pesetsky to name a few]"
DeleteYou may as well copy/paste here the entire linguistic community, if you don't understand or can't discuss control constructions; it's pointless...
"I do not call myself a linguist because i do not think that reading no matter how many books by linguist from a philosopher's perspective makes me a linguist any more than reading the books by psychologists i read make me a psychologist or reading books about airplanes would make me a pilot. I know this is different from Chomsky who believes reading books about biology or by biologists makes him a biologist [at least he has called himself such numerous times]. "
That's quite a dogma you're holding onto there. You can label yourself whatever you want, who cares? I for one care only about good arguments, whatever theory you happen to profess...
"What you [pl] mention above is for the most part internal to your theories [or program] so it is not terribly surprising that others focus on different phenomena."
Those are the phenomena found all over the place. A good theory of language should account for them. The fact that T and others don't concern themselves doesn't make these phenomena any less valid/interesting.
"But this does not mean there is no way to compare what people who work in different frameworks are doing."
Yes it does. Sure, you can (uselessly) philosophize about the differences all day, but what it comes down to is that you can't have proper scientific comparison – period.
"I would really hate to end this conversation believing neither you nor Norbert know enough about Tomasello's work to answer the questions I asked"
It sure ends with me believing that you, C, know very little.
"What you [pl] mention above is for the most part internal to your theories [or program] so it is not terribly surprising that others focus on different phenomena."
DeleteThat's really not the logic of the situation. We say "phenomena X teach us that the human language faculty contains Y", and Tomasello says "there is no Y". We say, "then how do you account for X?", and there's no response. In the specific case of Tomasello, I've had exactly this kind of exchange with him in person (including literally the absence of response, on two occasions).
There is nothing very "theory-internal" about the phenomena we bring up in these conversations. On the contrary, they are the stuff that language is made of: morphemes, words, how they do and do not combine to form sentences of the world's languages, what the results sound like, what the results mean, how children gain this knowledge, and how people put it to use.
And in general, there's nothing that linguists like better than arguing about what these phenomena teach us. If Tomasello and company actually participated meaningfully in such arguments, the world would be a fine place. But they don't and it isn't.
We are back at insults, nice. Exactly what makes you think I don't understand or can't discuss control constructions? But since you apparently know so much more than i possibly could have learned from reading about control constructions [among other things], can you please tell me what the BIOLOGICAL bases of control constructions are? We are all doing biolinguistics here, aren't we? So don't be shy what IS the biology of control constructions? If I do not understand your answer I make sure to ask for more detail until we all know exactly how and where in the brain they are generated. Or maybe they are not generated in the brain given that the details about this biological organ remain a tad sketchy. No matter I am confident you have all the answers.
DeleteMy last comment was directed at Seid Tvica not at David whose post appeared only after wards.
DeleteThis reply is to David:
DeleteYou say "That's really not the logic of the situation. We say "phenomena X teach us that the human language faculty contains Y", and Tomasello says "there is no Y". We say, "then how do you account for X?", and there's no response. In the specific case of Tomasello, I've had exactly this kind of exchange with him in person (including literally the absence of response, on two occasions)."
Thank you for this, now we are finally getting somewhere. This was exactly the kind of stuff I was asking for because it gives me a reason to prefer your framework over Tomasello's [at least for X]. Now all we need is the phenomenon X stands for - and since you say Tomasello had no answer in 2 cases i assume we have not just X but also Z? And what was the Y the existence of which Tomasello denies?
This comment has been removed by the author.
DeleteThe denied "Y" was what Tomasello calls "Universal Grammar", by which he and his colleagues seem to mean "Chomsky's proposals about universal grammar" -- but we can leave that terminological muddle alone.
DeleteX arose at an MIT talk to the Brain & Cog Sci department some years ago (so my memory of the talk is not perfect), Tomasello gave what must be a standard speech in which some experiments showed children extending constructions to novel verbs fairly slowly, and concluded that the way language acquisition proceeds is by learning word collocations and combining them with each other like beads on a string, rather than by learning structural rules of the sort that generative linguists posit. Then came the expected "there is no UG, Chomsky's as dead as a dodo" conclusion. In the question period, I asked him how his proposal could scale up to *explain* any complex piece of syntax, and especially those that have been strongly argued to illuminate universal properties of the language faculty.
As an example, I mentioned verb-second, since several of the reported experiments involved German-speaking children. Verb-second involves the verb dislocating to a specific structural position, arguably the complementizer position in German -- and is a productive process that applies to every verb. It is also known that German-speaking children acquire verb-second quite early. (Basically it's in place as soon as they produce multiword utterances.) It is also an easy-to-present and easy-to-understand example of parametric variation, since (1) it shows up in languages on every continent, yet (2) not every language is a verb-second language, and (3) no language is obligatorily verb-third or verb-fourth -- verb-second is all we find. In other words, the very picture of random variation constrained by UG and acquired early -- and for this reason, a standard example in introductory courses and elementary presentations.
Tomasello seemed to have no idea what I was talking about. He asked me to repeat the point, still didn't get it -- and as far as I remember, juyst said that he had lots of colleagues working on German, but wasn't German himself, and therefore couldn't answr me -- or something like that. End of conversation. I believe Alec Marantz, then at MIT, asked a similar question, with similar results, but I don't remember the details.
As for Z: a few years later, I was the discussant for a day of "construction-grammar meets generative syntax" talks at the LSA Institute in 2005, a workshop organized by Charles Yang. Tomasello was one of the speakers. I don't remember the details of Tomasello's talk, but I have my slides and my memory of their bizarre aftermath. I gave several examples of syntactic phenomena demonstrating the law-governed nature of linguistic structure (a gap in verbal selection possibilities and the cross-linguistic nature of the that-trace effect), and tied the logic of these phenomena to specific experimental work on the acquisition of Russian by the late Maria Babyonyshev. (No I'm not going to give the details at this point -- it would be far too long.) A slide was devoted to the logic by which these phenomena were counterexamples to specific quoted claims in Tomasello's paper. I had provided instances of rules that children appear to know with no specific evidence, and rules that children appear to not acquire despite abundant evidence, as well as an argument for maturation. There was ample time for Tomasello (or any of the others) to disagree, respond, or object in the question period, but he just sat there, blank-faced, as I recall. I remember Adele Goldberg asking a question, and no other response from the assembled constructionists. Considering the fact that I'd taken their claims seriously, and done my best to present a provocative and to-the-point response, I expected lots of opposition and objection -- but in fact got essentially no response at all.
I'm not going to be able to participate further in this discussion, unfortunately. Today was a public holiday in the US, so I had a bit of time, but tomorrow is not a holiday any more.
DeleteThank you very much for taking the time - i appreciate it and am sure should i have further questions Norbert will be able to answer them now that we got a ball rolling. Reading week here - so I got lots of time at the moment...
DeleteI don't obsess about the biological component. I's a very basic assumption that it's brought down to a bare minimum, which is hardly crazy to assume in light of what all non-human creatures can do, in fact it's reasonable. People still disagree on it so it's hardly a complete story. For me the assumption that Merge is a basic domain-specific computation is just fine. You want to know what Merge is? Read Chomsky's latest paper "Problems with projection." If you can’t handle it, start with the books that David mentioned. Most BA students do quite fine with Carnie and Adger, so I’m sure you can do too, and you can even call yourself linguist afterwards if you want, and I promise not to call you “amateur linguist” --- like you’ve been calling Chomsky “amateur biologist”, irrationally I should say --- as if that could somehow discredit your arguments. :)
DeleteIn any case, like David said, there is nothing theory-internal about control constructions. I don't have to assume that Merge or else are biological to recognize that control constructions need explaining.
For me what's great about generative grammar as opposed to other theories (functionalist, usage-based) is that any account of the phenomena like the control constructions is purely argument-based. No introspecting into your own thoughts or resorting to common-sense --- nobody is ever buying it, and nor should they. In contrast explanations in other frameworks tend to be very general and very vague. In CG, RRG, FDG just about all accounts tend be of the sort:
- the children use analogies, they learn constructions, they abstract. Grammar can be anything… (CG, Usage-based)
- people have syntactic inventories of constructions in different languages, they just insert syntactic elements into the slots. No limits as to how many you can have, no restrictions apparently (RRG).
- negative concord in creoles was first used due to communicative inefficiency (people couldn’t understand each other) and so it became grammaticalized and systematic. Something always gets grammaticalized after use. How? Why? Is it systematic? Nobody seems to ask… (FDG)
This is not the whole list but it covers a lot. In any case, if you carefully examine these kinds of explanations, in my view they’re assumptions at best, hardly telling us anything new/interesting. One tends to think they seem logical but there’s no solid reasoning to explain it without resorting to some kind of “common sense.” And I hate common sense explanations in sciences (not that that’s an argument or anything of the sort :) ). Common-sense assumptions I can handle, but not explanations.
In contrast to these types of accounts, many publications by Chomsky, and I dare say it by some of the people posting in this blog, are masterpieces, addressing every detail, re-evaluating every assumption at every step, searching for this simplest explanation, breaking down concepts to a bare minimum to figure out what they’re made off. It’s just centuries ahead.
This comment has been removed by the author.
Delete“If the cartoons you give about alternatives to GG are all you know about these frameworks I am not surprised you reject them, I would too.”
DeleteI stand by my characterization of those frameworks. Sure they put them in nice terminology and everything, but in the end they do not go in their accounts beyond what I said there, certainly not very far. Those are in fact the exact accounts, albeit summarized, I heard in the lectures/presentations by some of the theorists working in those fields.
Yes you made it clear many times "Not everyone agrees that there is a biological component." I disagree with those who disagree. Let's leave it at that.
I don't know if that’s Chomsky’s latest, it says on my copy: "received in revised form 30 November 2012; accepted 3 December 2012.” Seems recent, not that it matters. I’m focused on what he argues and not on tracking down when and where he said X. That's your job evidently.
Chomsky centuries ahead of anything? I didn't say that. Ahead of RRG, CG and FDG? A simple answer is Yes. The term centuries is clearly a figure of speech, but you can grab onto it as firmly as you want, and maybe one day you can write a potpourri story on me. Who knows…
"Worrying a bit more about how Merge might be biologically realizable sounds like a worthwhile project."
Like I said, Merge is an assumption. Given that ALL humans concatenate syntactic objects into larger ones, it's a reasonable one. It’s not my job to do anything with it, certainly not to satisfy your curiosity. If it’s not to your liking then you figure out a way to prove that mind isn’t modular, Merge can't exist, and that humans unlike a very long list of other animals, do not have domain-specific capacities. The burden of proof is as much on you as it is on me; and, as Adger pointed out recently in his criticism of CG, it's more on you since you’re following the CG argumentation line that language is different from other capacities. When you've successfully done that I'll gladly drop my fantasy that Merge is biological, I'll rise from my chair and applaud to Christina Behme.
I'm very enthusiastic about GenG and what it aims to accomplish. Keep calling it "juvenile boasting" if you wish. That's fine with me.
This my last post on this.
This is a reply to vkodytekFebruary 18, 2013 at 2:43 PM
DeleteApologies that I overlooked you interesting comment last night. you say:
"take into account that Tomasello has a diversified portfolio of interests in human cognition and behind (such as chimp’s theory of mind, social skills in dogs)".
You are absolutely right. It is also true that a lot of the work he does is of no interest to the work David refers to. I was not trying to suggest David [or anyone here ] should pay attention to all of Tomasello's work but asking if there is any overlap. And if so if in this area there is good reason to believe Tomasello is wrong.
I take David's answer as giving me some such reasons. There are phenomena that Tomasello has no account for. The question is does he have no account because he did not get around working on these phenomena yet or because his framework could not even in principle account for them. If the latter is the case we should reject his account and if David has good reason to believe so he is absolutely right to do just that.
You say: "He simply can't manage such details in grammar as these guys do".
I am not familiar with all of Tomasello's work. My main interest is language acquisition [that's why as a philosopher I first got interested in Chomsky because he's one of the few one reads in philosophy of language who takes questions about language acquisition seriously]. The work on early acquisition Tomasello does seems solid to me. But again even if he can tell us everything about early acquisition and has no even in principle account for David's phenomena then we need at the very least a second account that can deal with those phenomena. Ideally we want of course an account that can both. i am unaware of any account that covers everything so i look at what people in different frameworks are doing.
Should we worry about biology? That really depends on whom you ask. When I asked Postal if he worries about biologic bases of language he replied 'no'. From his perspective this is an excellent answer, why should he? But if we are doing biolinguistics we should worry whether it is possible at least in principle that the structures we postulate for language are biologically realizable. Postal has argued at least some of them are not so as biolinguists we ought to find a convincing reply to Postal's arguments. It is the same problem as in the debate with Tomasello above: if there is one phenomenon X that clearly needs accounting for but is not biologically realizable that much the worse for biolinguistics.
Now to do some more putting cards on the table: yes, initially my research lead me to believe that Tomasello [but also many others who work in the 'empiricist' framework] is right and Chomsky is wrong. However, i always has the feeling that there was something missing in the empiricists' account and it was when i first talked to Postal that i was able to understand what that was. Now probably no one in this forum will deny that Postal has excellent accounts for the kinds of phenomena David worries about. [note I say kinds of not the exact examples mentioned above]. So I take his work very seriously. The phenomena he works on certainly are real and need to be accounted for. If someone like Tomasello can't do that so much the worse for Tomasello. If biolinguuistics can't ....
From Postal's perspective there are two independent research projects: one linguistics [that is what he works on] and one lets call it developmental psychology. The second aims to discover how children can acquire knowledge of language. Of course in order to do that the second needs to know what the first has discovered. If someone insists that language is part of biology then s/he needs to be able to account [at least in principle] for all phenomena of language in biologically realizable terms. One can display the kind of disdain for people who point this out as Seid Tvica does in his [hopefully] last post but that does not make the problem go away.
I'm confused about Postal's "ontological" criticism. Perhaps I am driven to distraction by the stridency of his writing (ling.auf.net/lingbuzz/001608/current.pdf is my reference; I don't have a copy of his [2004]). The central claim seems to be that NL can't be both 'biological' and 'infinite,' but I am sure I am misunderstanding this claim, since I don't see precisely what the problem is with a generative/recursive production system with the capacity for infinite expressions that is physically instantiated. (I mean, kids these days build Turing-complete computers inside Minecraft.)
ReplyDeleteIf the problem has something to do with 'Platonism,' i.e., the problem of correlating a finite physical system to a body of mind-independent truths, then there are two obvious rejoinders: first, that it is certainly not unprecedented for science to go ahead with little heed to metaphysical controversy (in fact, as Mark Wilson notes at length in 'Wandering Significance,' logic itself is malleable: mathematical descriptions of physical systems sometimes require rejection of even very intuitive inferences such as [A; B; therefore, A & B] His arguments are one of the reasons I am drawn to internalism). Descriptive adequacy typically trumps such metaphysical scruples.
But second, and more important, even we grant 'Platonism' it seems a mistake to worry too much about it since the semantic features of the theory are explicitly *internalist* and other mental resources are to be brought to bear when considering language-world relations. And Chomsky repeatedly confesses ignorance as to such matters. I can't help but read Postal's critique as a simple rejection of anti-representationalism--which is fine, but, you know, I can't be held to the standards of representationalists if I'm not one.
Again, perhaps I misunderstand Postal's criticisms and I would certainly benefit from a clearer explication, sans Postal's axe-grinding. (I used to think philosophy got heated, then I read things like this: http://theanti-chomskyanredoubt.blogspot.ca/2006/03/noam-chomsky-and-quest-for-social.html -- this doesn't help with my "not a crank" heuristic...)
I am probably ill equipped to explain Postal's position. And the piece you read does not really explain much because it refers to arguments he made elsewhere. A good place to get more information about what is at stake would be the piece Katz and Postal wrote. Here is a link to it: http://ling.auf.net/lingbuzz/001607. It explains why for example the Turing machine example is not a 'solution' of the incoherence problem. As for semantics, I am certainly no expert but know that Katz has worked on it for decades and I found reading his books quite convincing [but also rather time consuming]. Maybe you want to give his Sense, Reference, and Philosophy (2004) a try...
DeleteHi NIC M: This is a bit long so it will take two boxes.
Delete. It’s me, the strident axe-grinding subject of your claim:
I'm confused about Postal's "ontological" criticism.
I have to agree; you are confused. You attribute your confusion to my flaws. Maybe. But other factors are certainly at play. Anyway, you say:
perhaps I misunderstand Postal's criticisms and I would certainly benefit from a clearer explication,
so I will try to provide such.
One thing that could lead a person to be confused about an article is a failure to read it. I don’t know if you read my ‘Chomsky’s....Admission’ piece. But there is not a word in your comment which provides any evidence that you have.
One way to avoid confusion/error when reacting to a paper which purports to contain arguments for one or more conclusions, as mine did, is to analyze the arguments. First, one should consider what the premises are, overt ones and covert ones if any. One can then consider whether these premises are true, supported by evidence, etc. Then, one can consider the logic used in the arguments and ask whether there are any logical flaws. If at any of these steps, one has an objection, one can formulate it, make it public, and make some claim that the argument in question does not justify the conclusion. If you have done anything like that with respect to my paper, you have kept it a secret. Nothing about my premises or the logic of the arguments appears in your comment. If you really haven’t done the above, wisdom would had suggested you avoid commenting.
These methodological points aside, let us focus on your statement:
since I don't see precisely what the problem is with a generative/recursive production system with the capacity for infinite expressions that is physically instantiated. (I mean, kids these days build Turing-complete computers inside Minecraft.)
Here is the problem in a nutshell. If natural languages are biological objects, then natural language
sentences are biological objects, presumably brain objects/events/processes, whatever. One of my premises is that brains are finite. Therefore, everything that goes on in them, everything which they produce, the totality of all that, is finite. So how can natural language sentences, taken as biological hence physical, be infinite in number, which means minimally, capable of being put in one to one correspondence with the full set of integers. They can’t...a finite collection of brain things can’t be in one to-one correspondence with an infinite collection. That is the basic argument.
This comment has been removed by the author.
DeleteThanks to the links to further reading, Christina. I'll be sure to have a look.
DeleteThe response you imply, which is hardly novel with you, is that somehow the existence of a generative/recursive production system solves the problem. This is a hopeless point of view (I fear stridency is creeping in, but that happens with the unsavory), simply a nonsequitur. The reason is that generative/recursive production systems/Turing machines/algorithms are not physical objects but abstract/mathematical ones. If as many people do, one hates MICROSOFT WINDOWS, one can still not destroy it, any more than one can destroy the integer 351, although one can destroy any computer in which it is coded. So coding is the key. People carelessly might say that people are Turing machines or etc. But that is just a category mistake. Carving the names of some integers in a rock doesn’t make the integers rock objects. Like computers, at best, people can have some abstract systems coded in them. But the abstract systems are not the physical coding. So the fact that a Turing machine T can have an infinite output doesn’t mean that a computer, a physical thing, in which T is coded can have an infinite output. And it is no use saying, as alas some have, that if one supplies the computer with more and more time/memory, it can produce more and more physical outputs. Because this hypothetical process quickly requires positing more brain space, more energy, more life span, etc. etc. than the physical world begins to supply. In other words, at this point, the so-called biological concreta supposed to represent natural language sentences have become completely nonbiological/non physical. This is just surreptitious Platonism, the worst kind.
ReplyDeleteIt is a bit strange that you are confused about all this...since what I just said is spelled out in some detail in the paper you criticized and in other works referred to there. Perhaps, if I may return to my natural
stridency, the problem is that the position I criticized is so deeply absurd that it is difficult for many people’s minds to really grasp on to the character of the claim.
Also, I might observe that you don’t refer to the fact that in the work of Chomsky’s I criticized, he essentially admitted that his position was nonsense and said defensively that we ‘had to’ accept things which don’t make sense. Maybe he has to. But obviously the rest of us don’t and shouldn’t.
Norbert:
ReplyDeleteThis responds briefly to two comments of yours from 2/17 (9h14, 2h30). It will take two comment boxes.
First, I would like to congratulate you on taking the trouble to run the blog,
allowing comments freely, accepting material you probably were not all that
interested in seeing, etc. This in my opinion is a very good, generous and desirable
thing to do and could be of general benefit.
Now to the comments: You say/ask:
for it amounts to trying to discover the biological bases for language. Does anyone seriously believe there are none?
Of course, I believe there is no biological basis for language. How could there be?
Languages are, I maintain, collections of sentences, each of which is an abstract object. Actually, the latter claim needs a bit of nuancing with respect to phonetics but no matter. There is no biological basis for collections or any other abstract objects. There can’t be because there is no causal connection between physical things and abstract ones. For any platonist about language, the issue about language is identical to that for mathematics and logic. Is there a biological basis for logic, for mathematics? Are Gödel’s incompleteness results about biology? Can Godlbach’s conjecture be resolved by brain investigations? This does not seems serious. None of these domains can be biologically based simply because the objects involved are not physical. And so it is with sentences.
I suspect you are confusing language with knowledge of language, entirely different
things (like e.g. rum and knowledge of rum). Of course there could be some kind of
biological basis for knowledge of language; how could there not be. I fear that the whole basis of the biolinguistic movement is question-begging failure to distinguish language from knowledge of language.
indestructible? If you present an ontological alternative to a platonist view of sentences, I suspect it is already refuted somewhere. If not, I will give it a go to refute it here.
Then in the next post you say:
ReplyDeleteIf Postal is a platonist then he believes that minds can see forms.
You didn’t get this claim from any publication where I assert it, any lecture, any interview, any recorded remark, any overheard remark because there never were any such. I have no such belief.
My speculation is that you take the claim to be an entailment from the nature
of platonism. But this is mistaken. Obviously, there is the traditional problem of how can humans obtain knowledge of abstract objects. No platonist denies that this is a crucial issue. Some have proposals about it. I do not. Your claim is falsified by the observation that a platonist like me need only say that...I don’t know how such
knowledge is obtained, but I do know that it is, just as much in the case of language,
as in the case of logic or mathematics. Apparently Kurt Gödel held a view of platonism which might be in the realm of your claim about what I believe. But while he is the last person that I would want to disagree with about such matters, I don’t understand the position and don’t see that it makes any sense.
How about you? Do you believe there is a biological basis for logic (e.g. for
Modus Ponens, not, note KNOWLEDGE of Modus Ponens). If so, how does it work?
Does our biology make that a valid form of inference, or only makes us believe it
is? If only the latter, why should we take our biology as grounding a form of inference
as yielding truths any more than a color blind person should take biology as grounding
the truth of colorless world? And if the former, how do you account for the logical necessity property of logical and mathematical truths...nothing about biology is necessary.
Prefinally, on falsifiability. This is a complex notion. My own way of thinking about platonism in linguistics is this. Look at positions on the nature of linguistic objects,
sentences. Platonism is defended if it provides the best account of what a sentence is ontologically. I would be interested to learn your alternative which could have some hope of actually underlying what we know about sentences, for example, their indestructibility (not to be confused, with the destructibility of knowledge of them.) If you doubt my assertion about indestructibility, you could e.g. destroy the sentence ‘This sentence is indestructible’. Such a demonstration would make a nice video. But if sentences are biological, how could they be indestructible? If you present an ontological alternative to a platonist view of sentences, I suspect it is already refuted somewhere. If not, I will give it a go to refute it here.
In two parts for reasons of space:
DeleteI can give two kinds of responses to your points. The first is to engage in an involved disquisition about Platonism in general. The other is to restrict myself to Platonism in linguistics in particular. Let me confine myself to the second, as the first would take us too far afield.
Here’s my view: whatever the attractions of Platonism for logic and mathematics (confession: I have no considered opinion on these matters and though I have always enjoyed the mythic quality of Platonist discussion I have never found the overall metaphysics very comprehensible), I fail to see why we should (or must) extend any of this to grammar and language. Or more accurately, you suggest that only Platonism gives a consistent view of linguistic practice. I disagree. From where I sit, there is nothing inconsistent with the biolinguistic perspective. There is nothing incoherent about taking brains to be computing devices with abstract properties or assuming that utterances have abstract characteristics. Of course, whether they do have any interesting abstract properties and what particular kinds they have are difficult empirical questions and what I take linguistic inquiry (note: not Linguistic Inquiry) to be about. But the view that our utterances and linguistic usages are capable of having abstract properties seems to me an entirely anodyne assumption. More importantly, it does not commit to a Platonic metaphysics.
On this view, the object of study is the human mind/brain, particularly those features that undergird our linguistic practice. What’s linguistic practice? Well what we are doing now is part of it, what I do when I gossip with colleagues is another, what I do when I go to the movies is another still. You know. There is a realm of human activity, a part of the world, that we can informally identify as ‘linguistic’ (as Chomsky says, much like there is a part that is ‘chemical’ another ‘mechanical’ etc.) and my interest is in finding what its biological basis is. As it has turned out, doing this has involved isolating certain levels of analysis. Some of these are the sentence, the phrase, the morpheme, the phoneme, etc. Utterances (things we actually say and comprehend) embody this structure and in virtue of doing so have certain properties and relations to other concreta. Brains that produce and comprehend these utterances have properties that allow them to do so (in part) by assigning structure of a sentence/phrase etc.-like variety to them. Here’s an assumption: we can fruitfully idealize away from many features of utterances and concentrate on these properties. So too, we can concentrate on those features of minds/brains that are responsible for these sentence-like properties of utterances.
Does this commit to some kind of Platonism? No. It commits one to assuming that conreta can have abstract properties. But why should I not believe that? Or, more pointedly, why should this be more difficult for linguistic concreta than it is for things like quarks, spacetime, pendula, planets, plants, bees etc. All of the latter have been unselfconsciously described as having rather elaborate abstract structures. What makes linguistic concreta less suitable than these for being described as having abstract properties? Why is it not Platonistic to say that spacetime, say, is a 12 dimensional Calabi-Yau manifold but it is Platonistic to say that utterances have hierarchical structure or that brains are Turing Machines? I just don’t see it. Or, more accurately, I see no special problem for attributing abstract sentential structure to the products of linguistic practice than I see attributing abstract structure to the concreta of the physical sciences. And until I see a reason to make an invidious distinction with regard to linguistics, I will refrain from doing so.
DeleteNote, that none of what I am saying requires identifying utterances with sentences or thinking that linguistic concreta are entirely abstract. It only requires assuming that they can (and do) have abstract properties (among others) and that these can be studied by studying the properties of these linguistic concreta. One can study physical computational systems by studying their abstract features. One can study abstract properties of brains by studying their concrete products. That’s what I am doing. No Platonism required. Just the assumption that concrete objects can have abstract properties, a view that seems, to repeat, entirely anodyne.
So here is a question for you: do your Platonist scruples extend beyond linguistics? If not, what makes it especially problematic that linguistic concreta have abstract properties? If not, then it’s not the biolinguistic program that you find objectionable, but scientific practice more generally. At least that’s how it seems to me.
Professor Postal,
DeleteYou write:
"How about you? Do you believe there is a biological basis for logic (e.g. for Modus Ponens, not, note KNOWLEDGE of Modus Ponens). If so, how does it work? Does our biology make that a valid form of inference, or only makes us believe it is? If only the latter, why should we take our biology as grounding a form of inference as yielding truths any more than a color blind person should take biology as grounding the truth of colorless world? And if the former, how do you account for the logical necessity property of logical and mathematical truths.."
I'm wondering if you could tell us which logic is the correct one, or if you have written anything on the topic that I may consult. There are many options on the table.
Priest has argued forcefully than the proper resolution to semantic paradoxes is to adopt a paraconsistent logic and concomitant dialetheist ontology. This obviously makes certain classical inferences invalid (e.g., depending on the proposal, disjunctive syllogism, explosion, certain interpretation of the conditional). Nathan Salmon has argued against that S5 cannot be the right modal logic. There are still any number of constructivists and intuitionists who take exception to the LEM and interpret negation non-classically. And so on.
If you are able to resolve the issue and tell us which inferences are Platonistic and necessary, and which ones are merely well-supported inductive inferences, philosophy would be in your debt, and would help me understand somewhat the nature of your project (and assess your critical approach).
I am not sure that such arguments can convince a Platonist to convert. ;) In fact there are no such arguments for this is a matter of choice. While biolinguists, according to Postal, run the risk that there is no biological basis of language, he runs the risk, according to me, that the abstract realm is not anthropocentric and, hence, his theory of natural language is not quite a theory of human language.
DeleteYou are right that the abstract realm is not anthropocentric - how could it be? And Postal's theory of natural language is just that: a theory of NATURAL language. Humans can obtain knowledge about natural language. Chimps apparently can't. So it is reasonable to assume that there is a biological basis for the human ability to acquire knowledge about natural language. Postal certainly does not deny that.
DeleteNorbert -
ReplyDeleteJust noticing some of the things that you have said in reference to my work, which you do not seem to have read with any care. You certainly have never shown that Chomsky and I use different senses of the word "universal." We use it exactly the same way.
Let's take the purported absence of recursion in Pirahã as an example. To quote HCF (1569): "We hypothesize that FLN only includes recursion and is the only uniquely human component of the faculty of language." Or "FLN includes the core grammatical computations that we suggest are limited to recursion." (1570) That is, for any language, the core grammatical computations of any language must be recursive by this hypothesis - only in this way, the author's argue, can we capture the "open-ended" nature of language and "discrete infinity". There is no allowance made for an alternative, non-recursive language.
Recursion is not like other grammatical operations in this sense. Thus the passive is not claimed to be the sole member of FLN. In his remarks in the Grammar of Happiness, Chomsky makes this clear. He says on camera that I *must* be wrong, that it is clear Piraha must be "built on a recursive procedure."
On the other hand, Chomsky contradicts what he says in the film in an interview about my claims that he gave to the Folha de São Paulo on February 1, 2009 (where he refers to me as a "charlatan"). According to the interviewer, Chomsky says that the implications of Piraha grammar for UG are "zero."
He claims that "Everett hopes that readers do not understand the different between UG in the technical sense (the theory of the genetic component of human language) and the informal sense, concerning properties common to all languages. The speakers of Pirahã have the same genetic components that we do, so their children can construct a normal language. Suppose Pirahã doesn't allow this. It would be like encountering a community that crawls but doesn't walk... The implications for human genetics are null." That is, if recursion is the core cognitive mechanism underlying human grammars, then the Pirahas lack a grammar; they "crawl" rather than "walk."
The idea that recursion is just something that, like the passive, may be absent in any given language, would neither square with HCF 2002, nor with Chomsky's conclusion that if the Pirahas lack recursion, then they lack a fully developed grammar (because, again, his remark assumes that a grammar without recursion is not a fully-developed grammar, showing that all "real grammars" must have recursion). Here we have a case where Chomsky clearly recognizes that languages without recursion in the phrase structure are counter-examples to his claim of the place of recursion in the characterization of natural language. Although Chomsky has more recently said that recursion/Merge might be set to 0, all this new "setting" does is render what he says unfalsifiable.
But you claim, Norbert, that Piraha is irrelevant and that I fail to realize that a surface exception has no bearing on the underlying capacity.
This is still false. Let's say that the Piraha case is an exception to the idea that grammars (not 'crawling grammars' but 'walking grammars') are based on a recursive procedure. Next let's say that Riau case (as claimed in Jackendoff and Wittenburg to also lack recursive syntax) is also an exception to this claim. How many exceptions can there be? Well in principle all living languages could lack recursion, yet one might still maintain that recursion underlies human language. But there is no empirical sense in that. Again, the idea that one language is an exception in this case rather than a counter-example renders the claim vacuous, because once exceptions are allowed in which the property underlying human language is missing from actual languages, then no claim is being made.
Dan Everett
Let's say that the distinctive characteristic of FL/UG is recursion, does this imply that every G displays recursion? No. It implies that every UG endowed individual can acquire a recursive G. UG specifies the acquirable Gs. Among these are recursive Gs. If a UG endowed individual does not have a recursive G it does not imply that s/he cannot acquire one. Thus, FLN is not a G it is a theory of the class of possible Gs.
DeleteGiven this, your claims about Piraha, if correct, and I must tell you I favor your opponents on this matter, are simply irrelevant. They say nothing about UG or FLN. Sorry. You are wrong. Period.
Isn't this discussion ignoring the difference between the two sense of 'recursion' that are unfortunately now floating around, the classical sense where it's the ability to put a phrase of type X into one of the same type, which the Piraha appear not to do, and Fred Karlsson has documented interesting historical features of (when writing starts, people do more of it), vs. the innovative sense in which its the capacity of Merge to apply to its own output and thereby produce structures containing three or more elements, which the Piraha are perfectly capable of doing.
Delete[Part 1 of 2]
DeleteI agree with AveryAndrews here, that there are two importantly distinct things being run together. For concreteness let's call them hierarchy and self-embedding.
Hierarchy: putting one constituent/phrase inside another.
Self-embedding: putting a constituent/phrase of type X inside another constituent/phrase of type X.
There's a straightforward sense in which self-embedding is usually thought of as a special case of hierarchy: you can't put a constituent of type X inside a constituent of type X, if you can't put constituents inside each other at all. (This leaves aside the logically possible option of a grammar that only allows X-inside-X hierarchy.)
My understanding is that the empirical claim about Piraha (I haven't read the original papers) is that it doesn't do self-embedding. One imaginable response to this is to say: "It doesn't matter; a claim about UG is just a claim about what humans can learn to do, it's not a claim about what every human will learn to do. In particular, the Chomskyan claim is only that humans can learn to use self-embedding, not that every human will learn to use self-embedding."
Now let's suppose that tomorrow someone claims to have discovered a language that does not use hierarchy; it shows no evidence of "constituent structure" in the classic intro-to-linguistics tests, and does not seem to have "structure-dependent rules". Again, in principle one response could be: "It doesn't matter; the claim is only that humans can learn to use hierarchical structures, not that every human will learn to use hierarchical structures."
This response to a non-hierarchical language would not be in keeping with the way I understand the Chomskyan view. A language which is non-hierarchical in this sense would seem to be a counterexample to what I take to be a standard claim of generative linguistics; this is why it's relevant to point out, for example, that children never hypothesise a "move the second word of the sentence" rule for forming questions. Similarly, the response of "we only said humans could do self-embedding, not that they must" to the Piraha claim seems to be backing off further than (I suspect) generativists might really intend to.
[Part 2 of 2]
DeleteIf we bear in mind the sense in which self-embedding is a special case of hierarchy, however, then I think it's possible to make sense of things: one precise version of the UG claim (which I suspect many generativists would find plausible, though I may be wrong) is that all human languages will use hierarchical structures of one sort or another, but these hierarchical structures may or may not involve self-embedding. Once we say this, then a non-hierarchical language that uses a "move the second word" rule for question-formation would be a genuine counterexample, but a language that has no self-embedding (such as Piraha might be) would not be.
This precise version of the UG claim (i.e. "hierarchy yes, self-embedding maybe") is not too distant from standard syntactic theory. It corresponds to, for example, accepting Chomsky's original arguments for moving beyond finite-state machines and adopting (say) context-free phrase-structure grammars as the "formalism" in which human grammars are stated, without making the extra claim that every grammar has to include rules that "take you round in a circle" and let you put an X inside an X. Or, in somewhat more recent terms, you claim that structures are built up via X-bar schemata or merge in ways that are constrained by subcategorisation/selection requirements of particular lexical items, but make no claim that every language must include lexical items that select an X and yet can also end up as part of an X (e.g. no claim that every language must include, along with many verbs that subcategorise [__ NP], other verbs that subcategorise [__ CP]). In other words, it's not unnatural to see hierarchy as the claim that is made by adopting the kind of formalisms/frameworks that generative syntacticians usually adopt, while seeing self-embedding as a somewhat more "accidental" property that a particular natural language may or may not have.
As I hope is obvious, I don't mean to make claims here about who might endorse this "hierarchy yes, self-embedding maybe" version of a UG claim. (Note that you need self-embedding, not just hierarchy, for a grammar to generate an unbounded number of sentences, so it is consistent with there being human grammars that only generate a finite number of sentences, and some may not like this.) But it does seem to be an in-principle option that could get lost if the two notions are not kept distinct.
Why do we think that UG requires phrasal hierarchy? Because we assume a principle something like UTAH, a mapping principle between thematic structure and syntactic structure. Let's make the general assumption explicit: If A is an argument of B then A and B form a constituent. If A modifies B then A and B form a constituent. If we adopt this, a very plausible principle of UG, then if we assume that predicates mean more or less the same thing cross-linguistically, then if you have a verb like 'eat' and a noun like 'apple' then 'John ate an apple' would have to look like [John [ate an apple]], viz. hierarchy.
DeleteWhat of self embedding? Well this would require that we have verbs like 'say' and that they mean what they do in English and so clauses are complements to 'say' and we get something along the lines of [say S]. Now, must every language have a verb like 'say' and must it take a propositional complement? Well, no. Even English allows 'say that' and this is not a propositional complement. So, one can imagine a language without verbs like 'say' or even without relative clauses but it is very hard to imagine a language where thoughts are expressed but there are no arguments of predicates at all. But if there are, then we will given this principle find hierarchy.
Now back to Everett: There is no argument that Piraha has no phrases. There is a claim that it has no recursion in the second sense that you and Andrew identify. Here would be a counter-example: we have verbs like 'believe' they mean what they do in English, they have propositional complements but there are no objects like [believe clausal complement] as a unit. Note, they may be able to say things that are similar, e.g. 'John left. I believe it' for I believe that John left. But try 'John left. Harry said it. I believe it' cannot mean 'I believe that Harry said that John left.' If the Piraha can express this thought and they don't use recursion to express it then this version of UTAH would be in trouble. Given that it is a simple central assumption, then we would have to think hard about what is happening. Until then, the absence of self embedding would not violate any UG principle I know of and so does not bear on UG. UG is a description of the LAD. CHomsky's point is that UG enables the acquisition of structures with self embedding. Can the Piraha learn such languages? Yes. Do they have UG? Yes. The only interesting question is why they choose not to use this in Piraha. But whether they do or not is not relevant to UG on any formulation I know of.
This comment has been removed by the author.
DeleteSome people assume a UTAH but not everybody. LFG for example has no UTAH, each verb has some collection of syntactic strategies for expressing arguments, but there is no motivation for assuming more uniformity and system in these than the data warrants. But I don't think your example is a problem anyway even with the UTAH, since in the long parataxis 'John left. Harry said it. I believe it', each individual verb has an expected argument in a UTAH-compatible place, the problem is that the antecedence of the pronouns isn't signalled unambiguously. The expressive problem could be evaded by replacing the last 'it' with 'him' - not the exact same meaning, but close enoughl
DeleteThis is a very brief reply to Tim:
DeleteLast year I e-mailed Chomsky a version of one of your questions: If we were to find a tribe that speaks a language that has a 'move second word' rule - would that be a counter example to UG. I asked because I was under the impression one specific claim was that humans could not learn such a language. Seemingly I was wrong because he replied such hypothetical tribe would not be a counter example...
It may be claimed that either I did not formulate my question clear enough and Chomsky misunderstood what I was asking or i did not understand his answer. So, if you think his answer is important [I thought it was] then you should not take my word on it but ask him yourself. He usually replies to e-mail very quickly - something I find quite amazing [in the most positve sense this word has] given how much he has to do
I think it's important that the logic of the situation is clear, i.e. that we understand which views would be contradicted by a language that does such-and-such and which views would not, and which views are compatible with the various toolkits that linguists adopt for doing their everyday theorising. But I don't (personally) think it's so important to know exactly which view Chomsky holds.
DeleteI agree that it is important to specify for theory T which properties of languages are N [necessary = every language has to have N], O [optional = individual languages either have O or they do not] and I [impossible - no language can have I]. Then if we find a language L that either fails to have [one of the] postulated N or has a postulated I we should either [i] give up T or [ii] move N or I into the O category. Now if we do always do [ii] at one point we will have no I and N properties. And at this time no matter what the evidence T is no longer falsifiable.
DeleteI think this is what happened when the Piraha data turned up. Chomsky [and not he alone] said 'even if Everett is right this does not matter. Now at that point recursion was the last N property. Moving it into the O category 'saves' T but it also makes T no longer falsifiable. I think it is really not relevant whether or not Everett is right about Piraha. What is relevant is the claim [repeated several times in print] that even if he were right it would not matter.
I think it matters to many people [especially on this blog] which view Chomsky holds [e.g. what he would accept as falsifying his theory] . And given the influence he has on the field i think it matters to many people not posting here as well. That overrides my personal preference...
My post above leads me to say something about the problems of UG more generally, as discussed by Phil Lieberman (The Unpredictable Species, 2013) and me (Language: The Cultural Tool). As we discuss in these works, UG (ironically) makes only one prediction, namely, that *not* all languages can be learned by all people.
ReplyDeleteIf we can find a population that cannot learn pro-drop we have potential knock-down evidence for UG. If not, its single prediction remains unsubstantiated. Though Phil and I give the reasoning in more detail in other places I briefly summarize it here (but see also my New Scientist article: http://www.bentley.edu/files/Big_Idea_Everettlanguage%20Read-Only%20pdfNEARFINAL_0.pdf)
If there is a grammar based on the genes, then it is subject to mutations and local selectional pressures. We know, for example, that lactase persistence has developed based on cultural pressures in something like the last 7500 years and that genetic mutations shaped by local pressures have improved the oxygen processing abilities of some Tibetans in just the past 3000 years.
We further know that some linguistic features have a time depth that would allow them to be plausibly subject to evolutionary change. Pro-drop has been around in Indo-European and its offshoots for about 6500 years at a minimum. Taking that as an example, we might expect a genetic mutation that makes learning languages with obligatory subjects impossible, reinforced by the local pressure to learn the pro-drop language. (We do know that speakers of tone languages have undergone genetic evolution in the ability to perceive tone - Bob Ladd's work.) A genetic-based theory of language predicts this. A non-genetic model of language, e.g. Postal'sor mine, does not. The evidence so far is against the genetic model.
I further develop the biological and cognitive bases of human language and why these are not UG in Language: The Cultural Tool.
The most recent work discussing much of what I summarize above, with the evidence is found here: http://wires.wiley.com/WileyCDA/WiresArticle/wisId-WCS1195.html
- DLE
UG is the name of those characteristics that specify the class of human Gs. Unless further specified it makes very few claims beyond saying that human G acquisition is partially a function of some native endowments that are distinctively human. the more interesting theories, the ones that I think there is a lot of evidence for, is that UG can specify the characteristics of those acquirable Gs. I actually think that GB gives a pretty good description of the kinds of characteristics those Gs will have (e.g. movement obeys locality conditions, binding requires c-command etc.). The minimalist question is whether these characteristics reduce to more fundamental ones. Maybe. I think that it is plausible that they do, but so far the evidence in favor of this position is at best incomplete. However, to the degree that they can be, to that degree properties of Gs reduce to properties of cognition more generally. I assume, like Chomsky does, that there is something linguistically parochial or otherwise we would expect other animals to "talk," which they don't.
DeleteAs for your genetic speculations, I have nothing to say. There are many ways to enter the gene pool. Chomsky has noted that there is relatively little UG variation despite a lot of isolations. His example are Papua/New Guineans being able to learn any language that any other kid can learn despite isolation for 30k years. thus, whatever UG is it has not changed for 30k years. If so, then whatever variation there may have been was squeezed out before the trek from Africa. I have no idea what this entails.
Ted Briscoe has been talking about stuff like this since about 2000, when he gave a talk on it at the ANU, the basic point being that languages evolve at about the same rate as human parasites, so that the principles of host-parasite evolution ought to apply. Then Dan's point is that they don't seem to, which strikes me as very important.
DeleteCould you elaborate. Say Gs change at the rate indicated. Does this imply that UG changes at all? What's the host, what the parasite in the analogy?
DeleteShould we be expecting UG to continue evolving? Should we expect to find isolated populations incapable of learning the languages that non-isolates do? I have no idea. If you understand this, please expand as I too find the notion that there has been no variation in UG for the last 100,000 years surprising, though I can't tell you why.
Let me try to explain. note this is a cartoon with no claim to accuracy. Natural selection can only work based on heritability with variation. Heritability ensures roughly that we're all humans and variation that we're not all like identical twins. Some traits [NT] are necessary for survival [if an organism does not have NT it'll die - hence not reproduce]. Lets assume a gene [G1] codes for one NT. Then every living organism has G1 - there is no variability. Most traits allow for some variability. Lets say these are encoded by gene G2 but the organism also survives when it has G2* or G2** or G2*** etc. And other traits allow for even greater variability - they are encoded by G3 and some organisms may have G3 and others not.
DeleteObviously things are a lot more complicated than this [and there is no 1:1 relationship between a gene and a trait]. But my point is this: unless you have very stromg pressure that prevents variation [as in the case of G1] you will have some kind of variation in the population. Now assuming we have the kind of language faculty Chomsky postulates it seems clear that it is not a NT [people can survive without LF]. So it would not be encoded by G1 genes. Now this is where the problem is: we should then expect some variation in the LF of individuals [and over time indeed some individuals who are NOT able to learn all human languages. But according to Chomsky no such people exist. So what is it that prevents variation?
Briscoe's thinking may well have moved on from where it was when I heard him 13 years ago, so here's the relevant web-page: http://www.cl.cam.ac.uk/~ejb1/evol.html
DeleteChristina,
DeleteI'm not sure about the specifics of your evolutionary argument, though there are some really interesting things to think through here. (I should stress though that such musings always should be made against the backdrop of the surprising genetic uniformity of the human species, probably due to a recent-ish population bottleneck).
I don't think it's safe to assume that not being 'necessary for survival' implies an expectation of hereditary variability, especially at the population level. We don't see very high rates of polydactyly or oligodactyly, for instance, though arguably not being able to acquire language is a more dire predicament than having supernumerary digits. Since having five fingers is not necessary, we'd expect variation; but the 'variation' is often always linked to ontogenetic pathology, not simple 'mutation', and thus is not hereditary. (Hereditary polydactyly is actually quite rare).
The point is we need a rather subtle model to generate any useful predictions about what we would 'expect' if some trait had a genetic or hereditary component. The uniformity of the human species and the messy nature of evolution makes it difficult to assert that we'd 'expect' variation in any putative language faculty. In particular it is unclear that we should predict that populations would 'adapt' to their languages (say, by 'selecting for' pro-drop) since linguistic change is prima facie far too fast for selection pressure--presuming that parameters are even up for selection, rather than the underlying architecture that accounts for parameterization.
A more fruitful line of evolutionary inquiry might proceed by examining pathologies of language, for example, whether certain forms of aphasia have a genetic component. There is some evidence for this. Then we'd expect that, much like polydactyly, atypicality of language would not be a 'variation' within a population but a kind of pathology that sometimes finds itself expressed (especially if something like minimalism is the case and GnB/P&P is subtended by a few very simple mechanisms).
Suppose language acquisition and use was indeed just a function of general intelligence, statistical learning, and so on. Then we'd expect pathologies of language to be pathologies of cognition more generally, and vice-versa. This doesn't appear to be the case. Of course, the dominant causes of aphasia are strokes and tumors, and genetic causes have not been definitely established (though there is some work here in the case of primary progressive aphasia). This doesn't *prove* anything one way or another, of course, but scientific confirmation is holistic. ("Crucial experiments" are a bit of a myth. A counter-example is not always disconfirmation: theories are abandoned when they become too baroque to account for the data, and, alas, there is a somewhat subjective component to this process).
Nic,
DeleteThank you for the thoughtful reply. Note that what i had provided was a cartoon with no claim whatsoever to accuracy. You are absolutely right that 'the real thing' is incredibly more complex and that it is also possible to have uniformity in non NT - like say if we have a large population of Chinese speakers who over centuries only receive Chinese input we expect a uniform 'trait' of speaking Chinese even if this is neither necessary for survival nor determined by LF [we assume everyone born in this population COULD have learned English if provided with the relevant input].
So my point [maybe badly made] was that unless we have an explanation like NT for uniformity we need another explanation if, pace expectation of variability, we observe uniformity. I agree with most of what you say above but would like to draw your attention to one point you make:
"In particular it is unclear that we should predict that populations would 'adapt' to their languages (say, by 'selecting for' pro-drop) since linguistic change is prima facie far too fast for selection pressure--presuming that parameters are even up for selection, rather than the underlying architecture that accounts for parameterization."
This strikes me as an utterly odd line of defence for a Chomskyan account. Why? Remember for Chomsky ex hypothesis there was just one single mutation needed to bring our wonderful language faculty into existence. Norbert called it somewhere a miracle, I gladly adopt the term here. This miracle occurred in ONE individual at one point and then spread through the population like wildfire. No prolonged slow change just one event and we had our wonderful LF in place.
Now why, the critic asks, can such a mutation not happen again? Say after the treck from Africa? And this second mutation results in a LF that is perfectly adapted to the local input [whatever that is] but no longer able to acquire any other human language? I think it is a fair question to ask given how little we actually know about LF. So the 'language change is too fast for evolutionary change to act' argument alone cannot account for uniform LF across the species. We need something else...
I completely agree with you that the most fruitful way to procede is doing work in the areas you suggest. Some of this work suggests that the disassociation between 'general intelligence' and language is not as great as we initially assumed. But I think it is too early to draw implications. It is however not too early to think really hard about what we would be willing to accept as disconfirming evidence for LF [or as you put it as reason to abandon the theory]. And that is obviously a lot easier when we know a lot more about the biology of LF than we currently do...
Hi Christina,
DeleteThanks for your comments! You say:
"Remember for Chomsky ex hypothesis there was just one single mutation needed to bring our wonderful language faculty into existence ... No prolonged slow change just one event and we had our wonderful LF in place."
That's not how I read the theory, though I could very well be wrong on this. In particular, I thought I understood MERGE as the 'computational core' that made language possible, though not necessarily immediately actual (for obvious reasons: what language did the first mutation-holder speak? none, of course!); and that the 'single-mutation' theory was not meant to cover every aspect of the LF, only the 'kernel' (to borrow a term from programming).
I'm basing myself here on Bolhuis, Chomsky, et al., "Evolution, Brain, and the Nature of Language" [Trends in Cognitive Science 2012], where the authors argue that the 'externalization' of language is an "ancillary aspect ... not its key function," which is instead to be an "internal instrument of thought." With this computational core in place, though 'public' languages could emerge, shaped by a number constraints placed by externalization, which didn't happen all at once (e.g., linear order, complexity control, that don't figure in internal thought):
"...much of the apparent complexity of language flows from externalization, with variation from one language to the next corresponding to different solutions to the way that internal syntactic representations ‘surface’ as sentences. These are precisely the aspects of language readily susceptible to variation and historical change, where models drawn from evolutionary biology have a role to play in accounting for language variation."
Presumably because the initial, enabling mutation was such a simple mechanism (with significant eventual downstream effects) there is little room for further variation *there*, especially of the kind that would allow for a different carving-out of the space of possibly acquirable languages. What kind of 'second mutation' could do the trick? If MERGE isn't operating properly, there will be no language at all, internal or external; if some variant on MERGE appeared no existing (public) language would be learnable to those with MERGE2. For instance, imagine a mutation that required MERGE2 to take only *ordered* pairs (I have no idea whether this is even coherent, but bear with me). This would completely change the conditions placed on the externalization of internal thought processes--themselves now radically constrained. The kind of piecemeal adaptionism just wouldn't be able to occur.
I don't think it is right to think of a single and sudden mutation event as carrying with it the full complexity of some higher-order description of LF, with some complex of genes coding for island constraints, another for c-control, etc.--that would be both incredibly implausible and open to the precisely the objection you raise: why hasn't selection pressure winnowed out some parts, tweaked others, made local adaptations, given that we already know rapid and sudden change is possible? But what made language 'possible' is not the same as 'everything that goes into language', as I read it. Unfortunately there are some features of this reading that make empirical confirmation/disconfirmation rather tricky, given how little we know about the 'software' side of things. But I don't see a knock-down punch, only a need for more work (but I'm speaking as an unabashed methodological pluralist, too).
You are right that for Chomsky externalization occurred later. But then it is not part of the core that is in need of an evolutionary explanation [I disagree but that is a different matter]. What the miracle mutation did [must have done] was conferring IMMEDIATELY such an important selective advantage that over a very short period of time everyone in the small breading group had Merge. It is not important here to speculate what exactly this miracle mutation installed and whether it took a long time for externalization to occur, why externalization occurred etc. etc.
DeleteMy point is if someone accepts a miracle mutation having such a profound and immediate effect could occur once then for that someone there is no good reason to deny that a mutation with a similar effect could have occurred later. As another cartoon imagine in a small breeding group [say the ancestors of the Piraha a second miracle mutation occurred that installed a 'grammar of happiness' this could explain for example why Everett did not find the same kind of recursion that is found in other languages in their language. Maybe for this small group 'being happy' was more important for survival than 'having recursion' that's why the grammar of happiness spread among them. [note the Piraha have recursive thought but apparently so do also some non human animals]. Again: Cartoon alert I do not think anything like this actually did happen. But then I also do not think the initial miracle mutation happened. I find non-Chomskyan evolutionary explanations more convincing.
Now I also did not try to provide a knock down argument. The burden of proof is on anyone proposing there was one miracle mutation that installed everything that is at the core of UG to show that another mutation of similar magnitude [or even far lesser magnitude as in my hypothetical Piraha case] could NOT have happened. Anyone proposing that language evolved gradually does not have to worry about the 'second miracle mutation' counter example only someone proposing the first miracle.
Hi Nic,
Deletethis is a p.s. to my previous reply.
You say: "Presumably because the initial, enabling mutation was such a simple mechanism (with significant eventual downstream effects) there is little room for further variation *there*, especially of the kind that would allow for a different carving-out of the space of possibly acquirable languages."
This is an argument by Chomsky that has too little concrete information to allow a meaningful evaluation. What ARE the significant downstream effects and how can they be connected causally to the 'initial enabling mutation' [which did exactly WHAT?].
Chomsky claims his speculation is based on similar accounts from evo-devo. But in the cases that are accepted by the evo-devo community we know exactly what the downstream effects are. Some have argued that we have good reason to believe that language is not a case like those. For one such argument see
Christiansen, M.H. & Chater, N. (2008). Language as shaped by the brain. Behavioral & Brain Sciences, 31, 489-558.
Unlike C&C I am not rejecting Chomsky's account because I think at the moment there is nothing to reject. To move away from his circular arguments in language evolution, which when cleared from all the empty assertions as the ones you cite boil down to: 'Merge' exists therefore Merge must have evolved
Merge could have evolved therefore Merge exists
he needs to supply the kinds of details I have been asking for in this forum many times. Then people can begin to look for evidence that confirms or disconfirms his account. No one expects Chomsky to do all the work himself. But he needs to tell people who are capable of doing such work what to look for. Unlike in say language acquisition where Chomskyans provide real phenomena that need to be accounted for, in language evolution no such detailed information exists because no one is working seriously on the biology of UG. But this is what you need before you can think about giving an account how the putative biological organ could have evolved.
Here is one way of giving content to the issues: Do (or can) the development of different, or new, or modified, means of production, means of transportation, of communications, forms of art, of architecture, social and political and commercial institutions, scientific theories, ontological presumptions, religious dogmas, diets, mores, discoveries, cross-cultural encounters, etc. etc. etc. affect the nature of (and limits imposed by) the (initial?) state of human organisms that determines humans – and only humans – to acquire a lexicon, a syntax, a phonology, a morphology, a pragmatics, a semantic; or can any of these contingencies affect the restrictions imposed at the interfaces to the means of production and recognition?
ReplyDeleteThis strikes me as a set of empirical issues that we are simply not in a position to answer at this point. A sensible default position – which I understand to be Chomsky's position when cleaned of all the polemical noise – is that the answer is “no”. Maybe facts that require departures will emerge, but that seems like a poor bet at this point.
The problem I think is that the task-specificity of what I like to call 'the child's cheat sheet for language acquisition' is widely presented as something we know, when it is, at best, a somewhat plausible speculation.
DeleteI don't know why the above comment seemed to come from a suburb of Antwerp. It was written by Sylvain Bromberger.
ReplyDeleteI agree that if you assume the existence of a language faculty having an initial state that determines that humans – and only humans – acquire a lexicon, a syntax, a phonology, a morphology, a pragmatics, a semantic, then your conclusion follows. If you are convinced such LF is necessary, then there is no argument no matter what kind of empirical evidence turns up. But if for whatever reason you have at least doubts about the necessity of such a faculty then things may change.
ReplyDeleteEverett seems to claim that his findings provide counter evidence for LF. I do not know a single word Piraha and am in no position to evaluate his evidence [and the claims made by for example Nevins et al]. So what follows is HYPOTHETICAL: IF Everett is right about the lack of recursion in Piraha [and if there is no equivocation] would this count as counter evidence or not? I take Chomsky claiming it would not. In that case recursion [properly understood] becomes an optional property of LF and one has to ask whether LF has any necessary properties and what would happen if counter evidence turns up regarding those. If there is nothing that could disconfirm the LF hypothesis are we still doing science?
Note that IF we entertain the possibility that, as Everett [and not he alone] claims, there is no LF, then your either or scenario would not hold. Instead we would have to ask: what can account for the fact that humans – and only humans – acquire a lexicon, a syntax, a phonology, a morphology, a pragmatics, a semantic.
My reply seems to have vanished. I will try again later.
DeleteNorbert,
ReplyDeleteThis is your blog. I don't want to be the party pooper. But you clearly choose doctrine over reading. The genetic discussion I raise is not speculation. It is an entailment of the theory. Recursion (in fact, all my claims about Piraha) is about as well-established, now with two sets of independent tests based on a large body of data as any claim for any other language.
But the peculiar idea that the ability to learn recursion means that it is part of UG is at the same level as the ability to learn the use of Dutch ovens for a UG of kitchen tools. I am not going to dwell on it. I have written a book on the subject and another (U of Chicago Press, 2014) to appear. And there is an entire journal issue (Pragmatics & Cognition 20:2) dedicated to my new book. I won't try to duplicate those discussions. The "faculty of language" or UG is about as real as the "faculty of hamburger eating" or "the faculty of speaking in tongues."
-- Dan
Aren’t two things mixed together here – the “deep structure” such as body morphology and characteristic cognition abilities vs. “surface structure” such as skin color, hair on the chest, lactose tolerance or adaptation to high-mountain locations or, hypothetically, to constant exposure to tones? Syntax (plus the capacity to acquire complex phonological structure) is a joint between the “deep” and “surface” structures. Should it be more like the former or the latter? I guess it’s a part of what, unlike the skin or hair color, makes us human. Wouldn’t we find genetic differences between hearing speakers and congenitally deaf signers?
DeleteYou said: “We do know that speakers of tone languages have undergone genetic evolution in the ability to perceive tone - Bob Ladd's work” If it’s so than it must be something quite new for the 2007 paper is about correlation which doesn’t mean causation.
Perhaps you don't know that what you quote from HCF is not what they meant (though the Science paper and the 2005 attempt to “explain” it are all but not transparent readings).
Dan, I have no idea what you are talking about wrt Dutch ovens. Look, UG is a specification of FL and this is about what kinds of Gs are acquirable. It is a property of humans that have a UG that they can learn recursive rules of grammar (i.e. where categories of type A are embedded within those of type A). If we find a language that does not display recursion this says nothing about whether speakers of this language could acquire a language that does. Your interpretation of 'universal' here is Greenberg's, not Chomsky's. An analogous case: Islands are part of UG. This means that no G can have movement rules across islands. Does this imply that every language has movement rules? No. UG says that IF there are movement rules then islands will be obeyed by them. UG says that if there is recursion the LAD can acquire it.
DeleteSee the discussion above by Tim Hunter. I think he puts the point well. I have a comment there as well.
As for the genetic argument, as I said, I have no idea what you were driving at. Sorry.
Norbert,
DeleteI'm sorry to crash into this wonderful exchange (thanks, btw: this is the most valuable tool for linguistic discussion *ever*!) but I'm a bit confused by your last remark. In his lecture at OCP Istanbul (but also elsewhere, for instance here in Leiden two years back), Chomsky made the following statements: "UG is not to be confused with Greenberg's universals. Greenberg's universals are generalizations - they can have exceptions. UG has NO exceptions except pathologies", which I interpreted, wrongly, as "UG is reflected in the possible grammars, and there are no exceptional grammars, except if you have a pathology, in which case you create deviant grammars".
The fact that he compared the two (Greenberg and UG) made me think that he was comparing models of possible grammars, not of FL, because he would have never mentioned Greenberg if he was comparing theories of what a child can acquire. And then I thought that if it was about possible grammars, this means that if you find a language in which negation is obtained by eliding the 3rd word of a sentence you would have found a falsification of UG.
Now I read that what he meant instead was (at least if I got you right): Greenberg made generalizations about languages/existing grammars (hence, possible grammars), UG is about what a child can acquire. But then I must stay the structure of Chomsky's argumentation is a bit confusing, as he himself is comparing two objects that are not comparable.
Greenberg's universals are generalizations across the surface patterns of language. Some are statistical, e.g. If A then a tendency to B. Chomsky's universals are intended to describe the structure of FL. These universals are generalizations over grammars; what structures can grammars have. To my knowledge they have never been taken to have statistical properties, though frankly I don't know why they couldn't. So, Greenberg's universals describe properties of languages, while Chomsky's describe properties of FL (and so, properties of grammars). Chomsky's conception is a step more abstract.
DeleteMaybe this reply will go through, but I am glad that Dan took the opportunity to put his comment in in reaction to yours.
ReplyDelete"Language faculty" is an very infelicitous expression that invites mostly cross purpose. The fact that people can meaningfully produce, parse,understand, and entertain linguistic expressions is not a necessary truth like two plus two is equal to four, but it seems fairly well attested. The conjecture that those capacities are grounded on specific biological endowment is as plausible as is the conjecture that the capacity to see objects whether for the first time or not is grounded on biological endowments. The interesting and challenging issues concern how to model the capacities grounded on those endowments and which are necessary consequences of those endowments. Do-insertion on its own presumably is not. Whether recursion is or is not ultimately implicates whether there are limits to the number of phrases that one can process from a finite lexicon. That is orthogonal to the issue of biological grounding. I am not sufficiently familiar with the Piraha data to have an informed view. But the best way to do the modeling is still far from achieved. Chomsky's own view have evolved and continue to evolve under the so-called minimalist program. See what happened and continues to happen in phonology. That egos get scratched in the process, well that is the history of scientific progress.
By the way, like you, I am not a linguist, but I love the field and its controversies.
Thanks for your reply.
DeleteYou say: "The conjecture that those capacities are grounded on specific biological endowment is as plausible as is the conjecture that the capacity to see objects whether for the first time or not is grounded on biological endowments."
Possibly. One thing i notice as [former] biologist is that we have visual systems all over the place, language on the other hand just once [I may disagree with Chomsky on a lot of things but like he i believe no other species has anything comparable to human language]. So if language is grounded on specific biological endowment i would think it would have to be something unique to humans. Chomsky is right that we can't experiment on humans the way we can on other species. But we certainly can do non-invasive genetic analysis etc. etc. So far there is no evidence that there's anything human specific that's found no where else in the animal kingdom. Chomsky did not come up with his theory [hypothesis, program] of innate endowment yesterday but decades ago. Look at the things we HAVE discovered since the 1960s. We can tell that some people are related to [descendants of] one group of Africans while the vast majority of people is related to another group. No invasive work was needed to figure that out. It would seem to me like a very promising research program to look for genetic differences between humans and say chimps that could provide answers. I would expect Chomsky to spend his energy on such a project instead of speculations about nematodes or bacteria.
Regardless of what the reason are there is no specific research for language related genetic endowment at the moment and we have no *hard evidence* but conjecture from what Chomsky calls E-language. So findings like Everett's on Piraha [again IF he is right about everything] SHOULD make a difference...
Chomsky is not young, he is involved in many things, he has no formal training in the techniques of genetics, or of neurology etc. Obviously that work should be done. I hope it is being done somewhere by qualified young people and that the results and debates in linguistics will provide relevant questions. The issue is surely not how Chomsky should be using such energy as he has left, but whether the sciences and technologies have reached the point where such research can be fruitfully done, and if so, how.
DeleteSylvain
You are right, my formulation was careless. I did not mean to suggest that Chomsky personally should work in genetics still less lead the way making discoveries there.
DeleteBut he could make suggestions to people capable of doing such works what exactly it is they need to look for. [Suggestions similar in kind as those David and I discussed in the context of Tomasello's work]. In his earlier work Chomsky has done this and provided fairly detailed descriptions of what LF might look like. At that time we had no way to carry out any genetic research program of the relevant kind. By now Chomsky has drifted further and further away from providing detailed descriptions of LF and focusses on emphasizing commonalities with ALL living organisms. No one denies that any biological organ has some of the latter. But LF certainly has at least some properties found no where else in the animal kingdom. And these would seem to be of main interest. Until Everett's work came along Merge seemed to be the last remaining LF specific property. But by now even this seems up for grabs. So WHAT are the qualified young people to look for?
Although I distance myself from Postal's bitter *ad hominem* attacks on Chomsky, I sympathize with his Platonism. Norbert, your question how linguistics differs from physics has several traditional answers, which are about knowledge vs. physics in general. So, assuming that linguistics is about a form of knowledge, it should not be compared to physics but to *knowledge* of physics. Naturalizing knowledge (as the biolinguistics program does) makes one, ironically in this case, a follower of Quine's "naturalized epistemology". Well, Quine could be right of course, but I think he isn't. So, how does knowledge differ from the physical word?
ReplyDeleteOne traditional answer, going back to the medieval scholastics, was formulated by Brentano in the 19th century. He formulated the difference in terms of "aboutness": unlike the physical world, knowledge is "about" something. As you know, this is what philosophers call "intentionality." Ultimately, I believe, intentionality can be seen as a variant of Plato's insight that ideas cannot be located in the physical world. One way to see that is that ideas (and/or intentional objects) are *normative*. A norm cannot be reduced to something physical in the world because, being *about* physical things, it is at a higher level of abstraction: it allows you, for instance, to judge whether the physical thing is in accordance with the norm. General definitions (or physical realizations of them) don't help because those can be judged from the normative meta-perspective as well.
The problem of Platonism is how to connect the abstract world to the physical world of concrete things. Plato only had mythological answers to offer. No real progress has been made ever since, which reminds me of a friend of ours who made a distinction between "problems" and "mysteries." Be this as it may, Quinean/Chomskyan "biolinguistcs", which conflates ontology and epistemology and "naturalizes" them, does not offer a solution that convinces me.
You begin your comment with "Although I distance myself from Postal's bitter *ad hominem* attacks on Chomsky"
DeleteGiven that I have been harshly criticized here for incorrectly attacking a person and had to provide extensive evidence to support my claims I am wondering based on what you say that Postal's attacks on Chomsky are ad hominems?
The way i understand the term is that it denotes the fallacy to dismiss the view of a person by illegitimately attacking an irrelevant character flaw of that person. Based on Postal's published work I am not able to locate any instance of such fallacy. Postal has certainly done both: he has attacked views of Chomsky and he has attacked what he considers to be character flaws of Chomsky. In both cases he has provided evidence that seems relevant to the charge.
You can of course distance yourself from those criticisms. But calling them ad hominem seems inappropriate. Obviously it is quite possible that I have missed an ad hominem in Postal's published work - so would you mind providing an example where he commits this fallacy? Thank you.
Jan, from where I sit you are running together two different issues. the first concerns the status of linguistic objects. For Paul, I believe, he wants to treat them like numbers and urges a Platonist ontology. Why? So far as I can tell, because they are abstract and concrete objects cannot also be abstract. My problem with this is that I don't buy that concrete objects cannot have abstract properties and that this is enough to ground the Chomsky Program. If this is correct, then the problem Paul notes carries over to any attribution of abstract properties to concreta, including those in the physical and biological sciences (e.g. hearts are not pumps, arms not levers, spacetime not Calbi Yau manifolds etc.).
DeleteYou identify a second much more difficult problem for the mental sciences; how do representations represent. You are right, at least from where I sit, to date this problem has not made much progress. Fodor following some mid 80s philosophy points to causal theories, but these seem to me to name the problems, not solve them. The only people I have read that actually say anything worthwhile are Jackendoff and Paul and collegues. They discuss how ling reps line up with those from other cog systems, mainly vision. This is very sketchy, but none the worse for that.
I agree about the Platonist metaphysics but I love the writing and many of the ideas concerning "recollection" seem spot on.
I consider discussion of somebody's "character flaws" in a scientific context "ad hominem" and fruitless.
ReplyDeleteThank you. We use a different definition for ad hominem then and i'll have to remember yours. Just to make sure I completely understand: you would then also consider the discussion of character flaws leading to say Marc Hauser's scientific misconduct as fruitless?
DeleteBTW I do agree with what you say in the remainder of your post and was merely surprised by the opening line which seems quite irrelevant to anything else you say.
The accusation of scientific misconduct was embedded in formal procedures especially designed to establish that, plus there was gossip in the press. I can't remember discussion of character flaws in scientific journals or books.
ReplyDeleteThanks for the further clarification. I see for you 'the problem' seems to be where the criticism is issued not that it is issued. I am glad i asked and did not misjudge what you said based on how I use the term ad hominem.
DeleteNorbert, thanks for your reply. Paul Postal should speak for himself, but I think you underestimate the depth of the ontological problem. The problem has little to do with what you say about abstract properties of concrete objects but more with the very fact that I-language is considered to be a simple algorithm ('merge'). The problem, as I see it, is that algorithms are not physical objects but mathematical objects that have, at best, a (non-essential) physical realization. More generally, algorithms (with or without a physical realization) raise the same ontological and epistemological issues as other mathematical/intentional objects. Within the standard spectrum of opinion about these, Platonism is an entirely respectable position. Also its epistemological variant, Kantianism, sees these objects as transcending physical reality.
ReplyDeleteWhat do you expect to find in the brain? Physical factors that *cause* algorithms? That does not make sense. A Gallistel-style molecular computer that *runs* algorithms? From where I sit that would be only marginally more interesting than an announcement by Apple that 'merge' can be run on the next iPhone!
Jan, two points:
DeleteFirst, I am not arguing that Platonism is not respectable. I have said nothing at all about whether this stance is reasonable or not. I have been doing something more modest: defending the view that the bioling program makes sense and this includes defending the view that abstract predicates can be applied to concrete objects e.g. FL is a generative procedure with one basic operation Merge and a series of restrictions on its applications such as...This is an abstract characterization of a physical object. The project can proceed along lines outline by Marr and it entirely straightforward conceptually though unbelievably hard empirically.
Second, I expect to find in brains exactly what I expect to find in a desk top computer: programs, hardware implementing these with all the complexity this involves. This is the Marr program in the domain of language and it makes sense. Physical factors embody algorithms, sometimes transparently sometimes not. There are many steps between programming language and machine language. A Priori I see no reason to think that the relation between our theories, procedures, algorithms, etc. will be any simpler. However, I see no reason to think that this way of looking at matters is inappropriate and conceptually flawed. So yes, I do expect to find them (or hope to). As for Apple implementing merge, I am pretty sure that they can do so already if they want. I can see your excitement rising from here.
Norbert:
ReplyDeleteThis will be another two box deal.
I want to thank you for your hospitality and for your taking the trouble to respond to my comments at length. I will make two comments on your remarks about platonistic views of natural language and then disappear for good.
You say:
Or more accurately, you suggest that only Platonism gives a consistent view of linguistic practice. I disagree. From where I sit, there is nothing inconsistent with the biolinguistic perspective.
Perhaps then you are sitting in the wrong place. Some seven months ago I posted at LINGBUZZ an article (‘Chomsky’s Ontological Admission’) arguing extensively that Chomsky’s biolinguistic position was incoherent, literally contradictory. Later I posted other older papers of mine, referring to work of Jerrold Katz and Terry Langendoen and I, making similar points. I quoted a statement by Chomsky admitting that his position made no sense, in effect, admitting the correctness of my general claim. Since then thousands of people have downloaded these papers; maybe some have even read them.
Now, I am willing to discuss the issues further, although I really have nothing much more to say, but after these remarks today only under one condition. Anyone in the discussion will have to have addressed publicly by posting at LINGBUZZ or in some journal or some other forum a response to the arguments I and others have produced over decades. I note that neither Chomsky or the major advocates of his position, including the illustrious posters on this blog have ever attempted that as far as I have been made aware, and have certainly not done it at LINGBUZZ. They are, obviously, under no obligation to do so.
But until such a response is provided, and frankly, inductively, there is no reason to expect it will be ever, my conclusion is, has to be, the following very unpleasant one. The biolinguistic position which I claimed to be incoherent is incoherent, the arguments to that effect are unanswerable which is why they have not been answered. At the same time, neither Chomsky nor his major advocates are willing to modify or abandon the contradictory biolinguistic doctrine. So they remain silent. To the extent this continues, I can only conclude that biolinguistics in Chomsky’s sense, despite its pretensions to have something to do with science, has descended into being little better than a cult.
Notice how easy it would be for major defenders of these views to prove that claim is wrong if it is wrong. They need only address and refute the arguments I have posted. Chomsky himself has privately in e-mail addressed his reason for not responding. He told one of my correspondents this was due to his wish not to embarrass me. Really, what touching concern for others and lack of personal self-interest, strangely lacking when, without the slightest ground, he called Dan Everett a charlatan in of all places a Brazilian newspaper. Since you have personal communications with Chomsky, feel free to tell him though that he has my assurance that whatever he or someone speaking for him might say in this regard, I promise not to cry. So that hyperlame excuse has dissolved. No doubt he can invent others.
On your own remarks, one brief comment. You talk about assigning abstract properties to physical things. This is an obscure idea to say the least. Moreover, biolinguistics purports to be science. Scientists don’t assign properties to things, they try to discover the properties they have.
ReplyDeleteI don’t see any sense whatever in talk of assigning abstract properties to physical things other than super general properties like existence (if it is one) and self-identity. Since these characterize everything real, they can hardly be the basis for linguistics in particular.
Take the largest tree on my front lawn. What could it possibly mean to assign this properties of abstract objects such as nonlocation in space and time, innertness with respect to causation, indestructibility, etc. Nothing, such talk would be senseless. This has nothing to do with the particularities of trees but rather only with the unbridgeable differences between concreta and abstract objects, which, I fear, you give no sense of grasping.
I really have said too much already. So this is the end of my remarks on this blog. Good luck.
I also want to thank all the participants to this discussion for the vigor and (relative) congeniality of the give and take. When I started the blog I was hoping that it would instigate discussion of topics that I believe are central to the enterprise but to which working linguists dedicate little thought. So thanks to all.
ReplyDeleteThis will also be my last post on this thread. I want to make two brief remarks in ending.
First, Contrary to Paul, I see no problem assigning abstract properties to concret objects. It's entirely coherent and fine to say that a certain compound has a helical structure, or that it is icosahedral. Similarly for space-time being a 12 dimensional manifold, or a lap top being a Turing or von Neumann machine. This is all fine. Similarly for the study of brains/minds and FLs. The Marr program outlines a reasonable way of thinking about these issues, one that is entirely congenial to the study of FL in my view. So, the bioling program strikes me as a perfectly reasonable project, albeit one where we have only begun to scratch the surface of the problem of how brains embody an FL, how FLs build Gs and how Gs get used by humans. The program is sensible and not conceptually threatened. Paul seems to think that it is. I think he is wrong.
There is a remark that Ian Hacking made about electrons: if you can spray them they are real. I think the same thing can be said about computation. We can build computers that embed algorithms and codes that have specifiable properties whose function we can COMPLETELY understand. It is complicated, but fully comprehensible how machines can embody algorithms that compute functions. As Jerry Fodor said, this is Turing's great contribution to the study of mind: how to mechanize (at least part of) thought. To the degree that what linguists do can be understood in computational terms I don't see that there is anything in principle problematic about postulating sentences, generative procedures, algorithms etc. and understanding these as abstract descriptions of brain properties. There is nothing mysterious or hard to understand here. Paul's last substantive paragraph suggests that he thinks this is far more problematic than I do. We disagree. However, I also sense given his discussion of trees that he finds it senseless to ascribe mathematical properties to objects other than brains. If so the problem is not special to biolinguistics, but to "materialism," the idea that the physical world can have mathematical structure. Interestingly, this is NOT something that Plato and later Platonists would have denied. For them, concreta could have such properties in virtue of participating in the forms that made them what they were. What participation consists in is hard to fathom, but it is clear that the intent was to be able to apply mathematical reasoning to "real" world objects. So in this sense Paul may not be a Platonist at all. However, my concern is not to evaluate whether Platonism is a defensible position. My only concern is to defend the idea that CHomsky's bioling program is coherent and defensible. So, contrary to what Jan Koster suggested, I am NOT arguing that Platonism is indefensible but that Marrianism in linguistics is. That suffices for my purposes.
Let me put this another way: on these matters I am a methodological pluralist. People study language for many reasons. I am interested in one set of questions relating to what they tell us about brains and minds. You need not study these questions. But, in my view, it is entirely possible to study language from this vantage point. This is a modest position. Paul's is not. Right now, I have no reason to change my mind given that I suspect that where Paul and I part ways concerns matters that go far beyond the whether brains can code for language.
I think if disagreements are so deep it is probably best to agree to disagree. Given that Norbert's disagreement with Paul seems at least in part be based on misunderstanding the position I just would like to point out that [seemingly] Norbert misunderstood the point of the 'tree-example'.
DeleteTake the largest tree on my front lawn. What could it possibly mean to assign this properties of abstract objects such as nonlocation in space and time, innertness with respect to causation, indestructibility, etc. Nothing, such talk would be senseless. This has nothing to do with the particularities of trees but rather only with the unbridgeable differences between concreta and abstract objects,
The point of this example was to demonstrate that trees do not and can not have any of the properties of abstract objects. Trees are located in space, abstract objects are not. Trees come into existence and cease to exist, abstract objects do neither. Exactly the same as for trees is true for brains. Brains are biological objects and as such concreta just as trees are.
So "I also sense given his discussion of trees that he finds it senseless to ascribe mathematical properties to objects other than brains" does not describe Paul's position because Paul does not acribe mathematical properties to brains any more than he acribes them to trees. One difference between brains and trees is that a human brain [unlike a tree] does have the ability to learn about mathematical properties. But a brains does not HAVE mathematical properties. Given that this seems to be the point about which the most confusion exists I recommend reading the paper Realism vs. Conceptualism in Linguistics by Jerrold Katz&Paul Postal 1990
http://ling.auf.net/lingbuzz/001607. One can of course reject Platonism but I think it is important that one understands the position before rejecting it.
Now regardless what one thinks about Platonism, it is certainly an interesting, worthwhile project to study what allows brains to learn about mathematical properties or natural languages. I suspect Norbert and Paul would agree on that.....
"The point of this example was to demonstrate that trees do not and can not have ANY of the properties of abstract objects. Trees are located in space, abstract objects are not. Trees come into existence and cease to exist, abstract objects do neither. Exactly the same as for trees is true for brains. Brains are biological objects and as such concreta just as trees are." (my highlighting)
Deletethis doesn't strike me as a good example. if anything, this shows that trees can't have some of the properties abstract objects can have, doesn't it?
I am not sure if this is a serious objection or just a humerous attempt to criticize that I forgot to insert "etc." after "neither"? I hope it was the later so my apologies for forgetting. If your point was more serious than that you need to read the Katz&Postal paper I mention where the authors explain why the question you ask does not make any sense. Postal assumes that people who challenge his view here have already read all he has published on platonism - so the tree example pre-supposes you have done that. As 'shortcut' try to come up with a property that a given tree T1 has. Say it weighs 444 pounds. Now are there any abstract objects A that have this property? NO. Do this for any property of T1 you can think of. Postal says you will find none that is shared by T1 and any A [except for say self-identity and existence]
Deleteyou're right, this wasn't a serious objection, merely nit-picking. I don't want to get involved in a discussion about properties, but thanks for taking your time to reply and I'll try to keep my nit-picking to myself next time.
DeleteThanks. If you have a serious objection you actually should 'get involved'. Postal has put forward a criticism of the foundations of biolinguistics and if he's right these foundations can not be repaired, not even in principle. It is of course up to Norbert [or anyone else] to say; I don't care about such foundational issues I am still sticking to biolinguistics. But for anyone who does care about whether biolinguistics can at least in principle be put on a coherent foundation it is important to show that Postal is wrong. I think many people do care, I know from my correspondence with Chomsky that he does. And from Postal's comment we know that Chomsky thinks Postal is wrong. So why would Chomsky not put forward some knock down arguments for the sake of all those who do care?
DeleteNorbert says above that Postal's argument, if accepted, is not just a problem for biolinguistics but for any materialist position. Is he right? Yes and No. A framework of someone like Tomasello does not have the foundational incoherence because he rejects UG. He works [on Postal's view] not on language but on how brains acquire knowledge of language. But, enter David's arguments from a while back: Tomasello's account so far is incomplete: There are some phenomena X that need to be accounted about which Tomasello 'has nothing to say'. Now if Postal is right, then Tomasello's account will always remain incomplete, it is not a matter of 'we just did not get around to work on X yet' but 'there are some X for which we cannot even in principle account'. It seems David believes the latter is the case and if Postal is right [that languages are Platonic objects] that is an argument to support David...
The whole discussion of a difference between the abstract and the real is based on an understanding of physics that is at least 300 years old. Joseph Priestley showed that Newtonian insights led to a rejection of the divide between the physical and the non-physical/abstract.
DeleteFor example, Christina said let's use weight as a character of the physical. By this metric any "force" (gravity, nuclear, van der waals...) is not "physical" (on the assumption that Christina means "mass" when she says "weight", and is thinking Kgs not Newtons). Yet, these concepts have been crucial to our understanding of the "physical", and by incorporating them as real alongside what are considered real by our folk-physics, we can better understand nature.
I am not sure why we can't employ the same strategy in Linguistics. Before we understand any work from the 1980's-1990's let's make sure we understand work from the 1700-1800's.
I mean the concept "tree" itself is an abstraction! It crucially depends on the level of physics you are dealing with. The concept of weight is an abstraction! So, we are clearly in the business of identifying reality based on "abstract" concepts. Rather comically, by the logic laid out above, this should be incoherent. What exactly is this divide people are talking about?
You mistake my example. I said a tree has for example weight. It does not follow that EVERY concrete object has to have weight. Rejecting the view based on an example used for illustration seems a mistake. i have pasted below the links to papers by Postal which I assume you have read before putting forward further objections.
DeleteChomsky's Foundational Admission - Paul Postal July 2012 http://ling.auf.net/lingbuzz/001569
The Incoherence of Chomsky's 'Biolinguistic' Ontology - Paul Postal December 2008
http://ling.auf.net/lingbuzz/001608
Remarks on the Foundations of Linguistics - Paul Postal October 2002
http://ling.auf.net/lingbuzz/001606
Realism vs. Conceptualism in Linguistics - Jerrold Katz, Paul Postal November 1990
http://ling.auf.net/lingbuzz/001607
If you dislike polemics stay away from the first two and only read the others. In addition you can read the books by Katz [Postal references them].
Further, the comments by Jan Koster may address some of your questions.
Finally you use an equivocation between abstract concept [such as 'tree' or 'weight' and abstract object such as numbers or [if Postal is right] sentences.
My point was simply that you can't maintain the distinction that you are trying to if you take a post-18th century view of physics. You haven't addressed the basic issue. What justifies the dichotomy between the physical and the abstract?
DeleteAnd just like you, I used the tree/weight as an example. You mistake my continuation of your example for an argument. note "For example" at the beginning of the relevant paragraph.
But these ARE the questions that are answered by Katz and Postal. Katz raised the arguments over 30 years ago. If it were as easy to address them as you seem to think don't you think by now we had a nice publication by Chomsky pointing out in excruciating detail where Katz and Postal went wrong. He is famous for his 'take no prisoners' type of argument - have a look of his review of Skinner. Then take a look at the uncompromising criticism Postal has put forward in his 2004 book and the pieces he posted on LingBuzz.
DeleteWhat ever we may think about Chomsky and Postal; both are brilliant. So if one puts forward a position that seems to me as if i could refute it without thinking much i assume i have maybe not understood the position completely and read as much as i possibly can about it. [I have done this with Postal&Katz and in the process all my 'this just MUST be wrong' arguments evaporated]. If the other never defends himself against the relentlessly repeated attack I gather the defence can't be that easy.
But as I said in my piece http://ling.auf.net/lingbuzz/001573 don't let any of this stop you. If you think you can do what Chomsky said cannot be done [at least not at the moment], provide a coherent foundation for biolinguistcs and at the same time a refutation of Postal. Submit a publication to a journal. I am sure many people would love for such a publication to appear...
This comment has been removed by the author.
DeleteChristina, Chomsky's critique of Skinner was that against the ruling paradigma. Postal's is far from being the ruling paradigma. Moreover, Postal is not right when saying that his critique was not addressed - see Collins (2009).
DeleteBut I can speak for myself only. I agree with Platonists to the extent that I assume the realm of abstract objects as a cognitive construct of an observer (not necessarily human). I’d like to make it clear that this is an ontological rather than epistemological point: there are no abstract concepts without observers. From this point of view, the type-token problem is not topical and Postal’s notion of the nondestructability of abstract concepts is artificial. Let it aside, I don’t think that, say, triangle will have different properties depending on one’s philosophical position.
Unfortunately, I’ve read very little from Postal. Not that I wouldn’t try but it’s hard to concentrate on papers driven by the desire to show that someone else’s is a junk linguistics. My yesterday’s comment, which you, Christina, responded to, was somewhat related to Langendoen & Postal’s (1985) attempt to invalidate generative theories of language. They show that, from a formal set-theoretical point of view, the choice of Chomsky’s discreet infinity (finite but unbounded length of sentences) is an arbitrary choice from infinitude of possibilities. However, this conclusion holds only if there is no constraint on sentence length (btw, if we admit that a language sentences need not necessarily have an end, why not to allow for those having no beginning). Under such a general assumption, language’s (the set of all formally possible sentences’) cardinality is higher than Aleph null; in fact it has no cardinality at all, being a proper class. Sentences of such a language cannot, of course, be generated, ie. the generative grammar can’t describe it. That's true.
However, it’s an empirical fact that each sentence is composed of discrete constituents and has not only a beginning but also an end, so that the choice of discrete infinity does not appear arbitrary at all.
February 21, 2013 at 11:27 AM
I am not sure if you are serious about your first paragraph. If you are look at
Deletehttp://www.chomsky.info/articles/20071011.htm
This is a 2007 reply by Chomsky to comments made by Margaret Boden. Boden is hardly of Skinnerian proportions. It is irrelevant here what i think about her comments. The point is Chomsky goes one by one through what he perceives incorrect arguments on her part. If anything, compared to the Skinner review this piece is more devastating. In comparison to the attacks Postal has put forward again and again what Boden actually wrote about Chomsky was almost adoring. If he feels incorrectly attacked he will react forcefully.
Further, I am amazed that you would mention Collins because Postal 2009 { http://ling.auf.net/lingbuzz/001608 ] is the REPLY to Collins. The very first sentence of Postal's paper reads: "I am indebted to the editors of Biolinguistics for their unsolicited invitation to comment in this forum along with John Collins." And Postal expresses [justified I think] surprise that anyone but Chomsky would reply to his arguments and he refutes Collins' arguments. So again I am not sure what to make of you mentioning Collins.
I also do not understand your last sentence. You say: it’s an empirical fact that each sentence is composed of discrete constituents and has not only a beginning but also an end, so that the choice of discrete infinity does not appear arbitrary at all.
How can anything that has a beginning and an end be also infinite [discretely or otherwise]? If you mean, as i hope you do, there are infinitely many sentences of finite length then WHERE are they. In any given brain can only be a finite number of them [or to be more precise of tokens of them]. The number of brains is finite. The time humans have been on this planet is finite. So how do we get to infinity?
Now for your claim about sentences always having to have an end. Take the sentence 'My father's father was bald"
next take
"My father's father's father was bald"
"My father's father's father's father was bald"
"My father's father's father's father's father was bald"
If you are the hypothetical Turing machine with infinite tape and time you can go on forever adding another 'my father's' to this sentence. There is no longest sentence here that is such that you cannot add another 'my father's' to it.
There is however a longest token of this sentence that could be stored in a human brain which is finite.
Again PLEASE read the arguments by Postal instead of relying on what I say. I am NO Platonist but attempt to present a position I do not fully understand. Even if something [everything] I say is wrong, it does not mean the position is wrong...
This comment has been removed by the author.
DeleteDéjà vu. See my February 16, 2013 at 4:19 PM.
DeleteIf I am the hypothetical Turing machine with infinite tape and time then I am not a speaker of natural language. I am able to repeat your schema only up to a point. If I say it's one billion cycles, you can argue what if I'd learn it twice as faster or what if I'd live twice as longer than I assume etc. That's why one can't set no upper limit. But it's always be finite.
I am glad we agree on a few things:
Delete1. You did not object to my observation that Chomsky's desire to refute those who attack his position is alive and well. So maybe i can suggest the "Chomsky test". If you find what you consider to be an objection to Postal ask yourself: Would Chomsky have thought of this and used it for a public rebuttal of Postal's position? Chomsky is very smart and he had roughly 30 years to come up with a rebuttal. What are the chances he never thought of what i propose right now?
2. We agree that infinite Turing machines are no models of speakers of natural languages.
Now for the rest of your post i am again not sure what to make of it. It strikes me as a version of the countless Zeno paradoxes philosophers entertain themselves with. They are all based on the fact that a finite distance A--B can mathematically be divided into infinitely many segments like the series 1/2, 1/4, 1/8, and so on. Then depending on the version they claim that an arrow could not move from A to B or not even move at all, Achilles could never overtake the tortoise etc. etc because that would involve moving over infinite numbers of segments. But since obviously arrows [and people] can move and fast runners do overtake slow runners [did I ever hate that in high-school!] the mathematically possible division into an infinite amount of segments is not the right 'math' to apply to those kinds of problems. In the physical world there are not just upper limits but also lower limits to how many times you can divide a physical object say. [There are more elegant solutions to the Zeno paradoxes so please do not use my sketch as another counter example].
Now when I look at what you say: maybe I can speak twice as fast, maybe i can speak 10 times as fast. But there will be a limit, if I attempt to speak faster it will be just noise. You can maybe built a computer that is better than me but it will also have physical limits. As long as we are in the physical world you have to account for those limitations. That's why I keep on asking for biological realizability.
Since it was not really clear to me if you are trying to argue for or against Platonism or defending Chomsky I stop here. Just let me repeat one thing [which i also pointed out here: http://ling.auf.net/lingbuzz/001573 ] that seems still not clear:
Postal makes two independent claims:
1. Chomsky's biolinguistics is internally incoherent.
2. Platonism can provide a coherent foundation for Linguistics.
Most people seem to think if they refute 2. they also refute 1. But that is NOT the case. Even if 2. is utterly false you still need to refute 1. So the focus should really not be on refuting Platonism but on providing a coherent foundation for biolinguistics [or at least showing that it can be done] - at least if one is a biolinguist.
Ad 1: It is not a topic I’m interested in. I’ve just explained why I can’t get through recent Postal’s texts.
DeleteAd 2 We have to keep in mind that it is words and rules (rather than sentences) what’s on the tape and, therefore, there is no need for it to be infinite.
The Zenon paradox is about the finite limit of an infinite series. If we conceptualize Achilles and the Tortoise in terms of relative speed, the solution is straightforward. But if one doesn’t have this concept and conceptualizes the task in terms of an infinite geometric series, one will arrive at the same result. Both ways are equivalent.
I mentioned, as an example of idealization in natural sciences, the macroscopic limit for open systems in statistical thermodynamics which is conceptualized as an average over an infinite set of systems, the occurrence of far most of them is extremely improbable, either virtually or completely impossible. Moreover, the system is assumed to be in equilibrium with an infinite mass and heat reservoir (the implementation of which would require unlimited mass and energy resources). What wrong then on a concept of an idealized speaker capable of speaking comprehensively at any finite rate and during any finite time?
Finally I mentioned a conceptualization of language as a countably infinite set of all possible (finite) sentences, each of which can occur with a certain probability. Given a finite time, some sentences will occur many times while other will never do. Though this conceptualization of language is different from the generative one, the results will be the same in both cases.
Ad “Platonism can provide a coherent foundation for Linguistics”. I agree as far as the descriptive issue is concerned. For example, I like European structuralism. What is contentious is the explanatory issue.
I am glad we are clearing up some misunderstandings. If you are not interested in who is right in the Chomsky vs. Postal debate, then 1 is in fact not of interest. Given that for most people on this forum it does matter who is right in that debate, I still think it was not a complete waste of time to propose the Chomsky test - it's a useful tool for anyone who cares about that particular dispute.
DeleteI have little to say about your other points because i either do not understand what you are trying to establish or how they are relevant to linguistics. E.g., there is nothing wrong with an idealized speaker as long as the idealizations are such that they do not violate biological realizability. Since it would be impossible to speak comprehensibly at ANY finite rate [even if 'speak' is just internal speech to yourself'] I do not see what the point of such idealization is if we are interested in a language faculty that is biologically possible.
For your last point: recall Postal sees two research projects: Linguistics and Psychology. He does not claim that HIS project [linguistics proper] solves explanatory issues for psychology but suggests a project different from his is needed to address those issues. As for his own project [linguistics proper] he claims to provide both descriptions and explanations. You can challenge these claims if you have evidence to believe they are false for HIS project. You also can challenge particular descriptions and explanations he provides.
You can also adopt the other currently existing research program that does not rest on an incoherent foundation: a form of what we call over here empiricism [Katz called it nominalism]. Here you do not have to worry about the infinity issues at all because you can insist that all there is are tokens, not types at all. Note however that such a position is of the same kind as the structuralism Chomsky replaced with his 'cognitive revolution' in the 1950/60s. Virtually everyone agrees that the shift from nominalism to conceptualism was a hugely important contribution and there is also broad agreement that Chomsky should be credited with this contribution. Chomsky justified the shift with claiming that language has some properties that cannot be explained based on nominalism. Platonists like Katz and Postal agree with this but say Chomsky did not go far enough. So one can see their project as continuing on a path Chomsky's work had made possible. Adopting Nominalism now would be like returning to where we were before Chomsky's great contribution. It is of course possible that his most innovate work resulted just in a huge detour and not in genuine progress. At the moment I am not prepared to believe this but others may disagree...
A few remarks about Norbert's helpful reference to David Marr. Although I like Marr's three-level approach to vision, I have never understood why FLN (recursive 'merge', etc.) is like the visual system, or any other "organ" for that matter. It is a false analogy. Here is why. An algorithm does not have an intrinsic function. For it to have a function, biological or otherwise, there must be an *extrinsic* context that determines the function. Well, the functional context of the algorithms of vision (if there are any) is fundamentally different from the functional context of 'merge'. In the case of vision, this context is autonomous and provided by biological causation (like natural selection). However, the linguistic functionality of 'merge' is not provided by natural selection but by human agency, viz., by the purely human invention of signs (words, etc.) that live on as part of our cultural record. Blurring this distinction is blurring the distinction between biology and *applied* biology, a fallacy known as "Panglossianism."
ReplyDeleteAn analogous situation we see with structures like the lungs. If we limit ourselves to the body, their main function is breathing (oxygenation of our blood, etc.). However, in a certain context, like playing the trumpet, its function is providing the air flow for tone production. The visual system, in Marr's sense, involves purely biological, non-agentive function assignment (like the lungs in breathing), while the function assignment in the case of FLN is agentive and therefore within the purview of human culture (like the lungs in the production of wind music). Both cases involve the same innate structure but what distinguishes them is the mode of function assignment.
There is a second difference. Marr's visual algorithms cannot be seen as knowledge, at least not beyond the metaphorical sense in which thermostats have "knowledge" of temperature. FLN, however, is usually seen as a form of knowledge in the non-metaphorical sense. As human knowledge in general, it is normative in the sense discussed before. It allows us not only to generate sentences but also to judge sentences or to deliberately deviate from the norm in sentence production. This brings us to Plato again, who understood that the normative aspect of ideas cannot be located in the physical world, which also includes the world of biology.
Minimalism added a question to the general linguistic agenda. I talked about this in an earlier post (which I would link to here if I knew how: it's 'Minimalism and the Language Specificity of UG.'). One generative project is to describe the properties of FL (that which undergirds the human capacity to acquire grammars/I-languages), the minimalist project os to isolate what in FL is uniquely linguistic. So FL will have two kinds of operations/circuits/features; some uniquely proprietary to language and not shared with other cognitive organs and some that are shared across cognition. FLN is the project to identify the linguistically parochial and there is a hypothesis that Chomsky has advanced that there is only one parochial operation/circuit/feature, viz/ Merge. That's his guess. However, even if this is so, it remains to ask how merge combines with the cognitively general to form FL, i.e. that part of the brain that acquires grammars/I-languages. FL so conceived is fit for Marrian analysis, in pretty much the conventional way. We can ask what underlies the observed pairings of sound/meaning that are characteristic of language use. We can attack this problem at various levels of abstraction with the "full" answer involving relating these different levels through more and more intricate mappings.
DeleteSorry, but I simply don't understand what you are talking about in the end of paragraph 1 and all of paragraph 2.
'Knowledge' is not a technical term of art in either Marr or Chomsky. It is a casual way of elaborating some crucial distinctions (i.e. the difference between having a certain structure and using structures for particular ends). The Marr project has not been completed for it bottoms out in object recognition, not something that we know how to get to yet. The early sketches and 1.5 D reps are all in service of this. So too with grammatical reps: they ties observables sound and meaning. I don't see why the analogy won't go through.
Norbert,
DeleteThe only part of language we are sure of it is "linguistically parochial" is the lexicon. It is hard to believe that something as simple and unspecific as Merge carries that badge of linguistic honor as well. Even Chomsky himself does not exclude that the content of FLN is zero. Do you have any idea how that could be constructed as an empirical issue? It is far from established that something like Merge ONLY shows up in language, but even if that were the case, it would be odd to call it FLN, as there is no intrinsic relation with language. Its linguistic status is crucially determined by the wider contest, which always minimally includes the lexicon. The lexicon is, by and large, not individual/biological but the cultural heritage of a community.
What you say not to understand in my post is very simple. Hauser, Chomsky and Fitch relativize the language-specific character of Merge by suggesting that it originally could have had a different function, for instance as computation in navigation systems. This involves what biologists call 'exaptation.' It is a slow form of *Funktionsverschiebung*, supposed to have happened by slow processes like natural selection. What I propose is a faster, agentive form of this kind of function change, brought about by human invention. This is what is called "recycling" (of brain structures) in Stan Dehaene's great book on the neurobiology of reading (http://readinginthebrain.pagesperso-orange.fr/intro.htm). Dehaene discusses a part of the brain that is active in all writing systems of the world, which he calls the "letterbox" (in the occipito-temporal part of the left hemisphere). A homologous area in the brain of Macaques is used for shape recognition in environments like trees. The human letterbox cannot have evolved by (slow) exaptation because the earliest writing systems were invented only 5500 years ago. So, Dehaene proposes that the brain area in question received its new function by associating it with a human invention --writing systems. Similarly, I propose that (the capacity underlying) recursive Merge originally evolved for a different biological function (if it is not a 'spandrel'), but that the linguistic function was brought about not by exaptation (H, Ch & Fitch) but by recycling (Dehaene) --associating an old brain mechanism with the more recently invented words.
Chomsky seems no longer to hold the position on language evolution you suggest he did in HCF 2002 and 2005 [I think it was only then that the 3 proposed LF could be empty?]. Fitch continues to hold that human language evolved gradually and via natural selection [see especially his discussion of the importance of Darwin's work on the matter in Fitch 2010]. He possibly still would endorse roughly the account you acribe to Chomsky above [but i can't speak for him]. But Chomsky has said explicitly in many more recent publications that Merge is the result of a one time mutation event and that only this event made it possible to combine {or exapt or recycle I have no idea if he would draw such a distinction] other cognitive structures and combine them into human language. So Merge itself cannot be an exaptation on Chomsky's current view. I think [but could be mistaken about this] Norbert pre-supposes Chomsky's current view and if so this may have caused misunderstanding of what you said.
DeleteRegardless whether I am right about this I also want to thank you for drawing our attention to Dehaene's account, which, in my opinion, has all the hallmarks of what a good scientific account should be [This does not mean i agree with everything he proposes but that would be probably the topic for a different discussion]
Thanks, Christina. I am familiar with Chomsky's one-mutation account, but did he explicitly reject the HCF story? Even if the one-mutation story is true, the functional integration of the resulting structures would still involve the invented lexicon as a necessary component. It seems to me that something along the lines of Dehaene's recycling is the only reasonable story available to solve the kind of nature-nurture problem at hand. It evens does justice to traditional Saussurian ideas about language or to Chomsky's poverty-of-the-stimlus arguments. So, maybe an interesting synthesis of viewpoints is emerging from the battlefield here!
DeleteI do not know off hand if Chomsky ever explicitly rejected that story. But i am not aware of many instances where Chomsky explicitly rejects a view he endorsed earlier - so i do not think it is necessary for him to explicitly reject to assume he has changed his view. My assumption is further supported by the comments he makes about the role that natural selection could have played in language evolution [e.g. in Chomsky 2012, 2011, 2010, 2009). He consistently rejects any role for natural selection to play except in the case of the Merge saltation.
DeleteI am also not sure what you mean by 'the invented lexicon'? If you refer to Chomsky's view he does not think the lexicon has been invented but holds [very consistently as far as i can tell] that it is innate. Since i find this position very strange i have specifically asked him about it and he confirmed that this is his view. He defends it with what he calls the massive poverty of stimulus [we could never have learned everything we know about items of the lexicon such as 'river' or 'psychic continuity' etc. etc. but nevertheless these are shared virtually by all humans [he discusses this in science of language and elsewhere]. referring to feminist literature I asked him also specifically if he disagrees with the claim that the term ‘sexual harassment’, for example, is a recent feminist innovation. He replied it could not have been an innovation because it always had existed just as insect harassment. So i very much doubt that on his view the lexicon could have been invented. His reply about sexual harassment struck me almost as 'Platonist"...
So given that Chomsky does not hold the lexicon was invented I am afraid i do not follow your point about the potential synthesis?
Reading, dancing, bicycle riding are cultural phenomena / cognitive technologies / surface functions built on the basic biological ones. Reading is based on language, bicycle riding, dancing on locomotion (for rhythm we are equipped with genetic clock). If language is such a surface structure / cultural phenomenon / cognitive technology, (1) what basic structure is it built on and (2) why is there a sensitive period for language (including syntax) acquisition?
DeleteThe discrimination of different syllables and even phonemes is not unique to humans. It’s present in some other mammals, such as apes, chinchillas, and some birds. What is unique (at least if we forget Alex the parrot) is, for one, the “perception-action” interface between speech perception and speech production which develops during the first moths after birth and, for another, rich combinatoriality which comes soon after the “perception-action” interface.
The notion of words as an invention is at least as speculative as the notion of genetic mutation enabling the combinatoriality. Rich lexicon seems to be a consequence of free combinatoriality rather than the cause. Or at least there is an interdependence.
This is not to downplay cultural factors, just to note that genetic endowment and cultural factors are not mutually exclusive, each side emphasizing its part.
We have a highly developed instinct to imitate which hardly is a cognitive technology.
Just to lighten the contention, there is a quotation from Koster (2010, “Recursion and the Lexicon”): “[Chomsky’s] emphasis on the (unknown) genetic instruction and his ignoring of culturally determined functionality brings his ideas closer [than desired] to sociobiology … it seem to me [close] to views that Gould and Lewontin (1979) characterized as ‘panglossian’, after Dr. Pangloss in Voltaire’s Candide, who praised the human nose as a structure to support glasses.” A beautiful optimality theory. ;)
Christina, a possible synthesis, as I see it, involves some Chomskyan ideas but does not depend on Chomsky's permission. What you refer to is Norbert's point 7, at the inception of this thread (that Quine's 'museum myth' is a reality for Chomsky). Note that even for Chomsky the lexicon involves, admittedly very minimal, invention in that the innate concepts have to be LABELED by more or less arbitrary, culturally conventional signs (to be 'spelled out' at the SM-interface). This is the achilles heel of Chomsky's entire internalist program.
DeletePersonally, I find labeling a totally misguided view of lexical semantics and it has been severely criticized for centuries. In my last communication with Chomsky (it has been a while) I asked what these labeled items look like and I never got a satisfactory answer. It is impossible to say which concept corresponds to a word like, say, "Schubert", as the associated information differs from person to person. Actual concepts are not found behind a label but are constructed in a context. Since possible contexts cannot be specified, it cannot be stated once and for all what concepts can correspond to a given "label". Take a sentence like: "Schubert can be downloaded on request". The most likely referent of Schubert here (and therefore the concept involved) depends on technologies that have only existed for one or two decades. This is what traditional linguists used to call 'polysemy' or 'flexible symbolization' and it was understood, in rudimentary form, by Aristotle in his Poetics.
If I remember well, Chomsky's reply made a distinction between "technical vocabulary" and "elementary vocabulary". The extreme, preformationist view was supposed to hold only for the elementary vocabulary, such as in the child's acquisition of the meaning of the word "water." I see no way to formulate a clear dividing line between the two vocabularies. But apart from that, even a child's conceptualizations are open and evolving, allowing the child to integrate ever more contexts.
Thanks for this. If you are trying to convince me of the implausibility of Chomsky's internalism you are preaching to the choir. But in an effort to evaluate his position fairly i thought i had to make at least sure he holds really what i thought he holds based on his writing. If something like 'sexual harassment' is part of the innate lexicon [labelled or otherwise] this lexicon must be massive. He did not even offer a 'combinatorial' proposal [like it being composed of the simpler 'sexual' and harassment' because this would not have challenged the feminist's claim that they invented 'sexual harassment' English lacked such a term].
DeleteI have not the foggiest idea how one might even begin to account for such a rich innate lexicon in biological terms. I am not aware of any account Chomsky gives but gather it may fall under one of the mysteries possibly always beyond our cognitive grasp he has been talking about frequently at least since 2009.
Like you i believe there must be a not insignificant cultural component to [knowledge of] language and that we must allow for the possibility that a child's conceptualizations are open and evolving.
Maybe Norbert can clarify this. I heard Fodor and Chomsky say in class that ALL concepts must be innate, if we want to avoid the conclusion that humans are angels (sometimes with reference to the allegedly analogous case of the immune system that "foresees" all antigens). So, the argument seems to be that it must be true because the brain is a finite physical system. I must be missing something enormous, because I find the idea absurd (really, I have been trying very hard to be open to a conversion experience!). It's like the old preformationist idea in biology that the sperm of Adam contained all future humankind. In any case, conceptual predestination is metaphysics, not a testable biological claim. It seems to me that without this questionable metaphysics the 'internalist' view of language collapses.
ReplyDeleteTwo small comments on Jan's remarks:
ReplyDeleteFirst, I was addressing the question of how the study of FL could be assimilated to a version of the Marr program. This was intended to assuage your skepticism that this was appropriate. I did this by noting that I don't believe that the HCF discussion (or any other re Minimalism) effectively alters the logic that has been with us since Aspects: we try to figure out the properties of FL and a Marr like break up of the problem seems apposite. Or, at least I don't see why it is particularly problematic.
This said, there is a second question: what is idiosyncratically linguistic in FL? I think that Chomsky has moved between various positions on this, sometimes thinking that there is nothing specifically linguistic (this fits with your view (Cedric has a version of this as well) and sometimes thinking there is. My earlier post on birdsong and what it means fort merge tried to discuss this a bit. My own view, is that there are proprietary linguistic operations (I discuss them in my book) though there are not many. If one takes a look at what's in a standard minimalist proposal nowadays, it seems to me that there are lots of operations and relations without cognitive analogues in other domains (Long distance Agree for example). At any rate, I also think that this is at right angles to whether some kind of 'miracle' occured 100k years or so ago. Something happened that had not happened heretofore. To my mind whether it was a "mutation" or a redeployment of an exaptation is all the same to me. Either way some kind of rewiring happened and whether this is due to some mutation or something else the idea would be that it is pretty minimal, i.e. that the rewiring was not extensive and so most of what is in FL is pretty cognitively general.
So, we should not run together, I believe, the minimalist question as to what kinds of parochial features FL has with the other question of whether there is an FL. Even a minimalist can assume that there is one in humans and this makes the Marr approach to figuring out its structure perfectly apposite.
Thanks for this clarification. There are undoubtedly some parallels between vision and language. So to a degree the analogy is helpful. But it seems eventually we need to get past the parallells and focus also on the differences.
DeleteI am a bit concerned about the largely scaled down evolutionary miracle, especially if you're right and not even a mutation is needed anymore. The reason is this: if it is so simple to turn a language-less ape brain into a linguistic human brain, then is seems very strange that this kind of minimal change occurred just once in the entire animal kingdom. If it is so 'easy' to install language, should we not expect to find language in other apes as well?
If on the other hand the miracle could only occur in a 'language ready' brain that was in important aspects different from say a chimp brain, then these differences ought to be part of the evolutionary explanation. I am not suggesting there could be no explanation for such differences but think that the 'minimal change' story does not really explain what is in need of an explanation.
Yes. I favor the mutation story because it seems to me that it goes some ways to explaining why this only happened once. However, Chomsky conjectured that it did happen before. Birds. That was the topic of my post on minimal birds. Note that there is likely more than one miracle anyhow, we know next to nothing about lexicalizarion, another thing humans do orders of magnitude better than other mammals.
DeleteA second point as I was running out of room. YOu ask about the Chomsky-Fodor point. It is discussed in chapter 6 of the Royaumont volume. Fodor observes that if there is to be inductive learning then there must be a way of representing what is to be learned. This is the crux of his claim that all concepts must be innate (c.f. 146ff). Here's his main argument:
ReplyDeleteThe standard inductive theory "assumes as "given" the "criterial attributes" which form the hypotheses that are "fixed" in the experimental situation. In consequence, a thoery of how our beliefs are determined by our experiences is not a theory of the source of our inductive hypotheses. On the contrary, IT PRESUPPOSES THE AVAILABILITY OF SUCH HYPOTHESES...and tells us only how the likelihood that one or another hypothesis will be accepted by an organism varies with one or another aspect of the organism's experience of the environment. I AM TELLING YOU THAT AN INDUCTIVE LOGIC...CAN'T TELL YOU HOW THE CONCEPT MIV IS ACQUIRED BECAUSE IT PRESUPPOSES THE AVIALIABILITY OF THAT CONCEPT WEHN IT ASSUMES THAT MIV OCCURS IN THE CONFIRMED INDUCTIVE HYPOTHESIS" (my emphasis here).
That's his argument. It is based on the logic of induction, nothing else. There is a second part, as you know; viz. that there is no lexical decomposition. Given that the number of concepts is roughly equal to the number of possible lexemes and so...
Frankly, there is not a lot of room to go after this argument as it is based on the logic of induction: you can't gather data for a hypothesis you cannot represent. In this sense, the concepts that get "fixed" by experience, must in some sense (what sense exactly is not clear) be there already. That's the argument. Good luck.
So I'd take the basic facts about insect classification in English vs Arrernte (Central Australia) as a straightforward embarassment to Fodor's position. So for example our 'grasshopper' corresponds roughly to their 'inteltye' (don't even dream of trying to pronounce this stuff unless you've been there), except that green grasshoppers fall under a different term 'nwekepeltherre', which also covers our 'katydids'. So 'inteltye' appear to be bugs with vaguely fingerlike bodies and big hind legs for jumping, plus four short front ones, unless they are green, in which case they are nwekepelhtherre. The round vs blade-like distinction that we register with 'katydid' doesn't seem to matter in their terminology. Pushing on with the finger-like body crowd, their 'iltywiltwye' covers both praying mantisses and stick insects, whose common characteristic is at least 4 long walking legs; the difference between the mantises and the stick insects being that the former have robust front 'arms' for grabbing things, while the stick insects just have 2 more walking legs.
DeleteThe point of this ramble being that if you look at a corner of the vocabulary in detail, and compare across languages, it's often easy to see how the various concepts are related to each other by what are basically definitional differences; if Fodor wanted to say that all of the six distinct insect concepts I enumerated above are innate, with different ones activated in different cultures, he can, but a definitional account seems much more plausible, to me at least, supposing that shapes can be done either Marr's way or Anna Wierzbicka's way, and the color aspects of it probably by neural networks (Wierzbicka is very wierd on colors).
I agree that this is a problem for internalism/innatism.
DeleteAlso note, that if Chomsky is right that the main function of language is 'expressing my thought'/'enabling inner dialogue', and language is no more important for 'communication' than haircut or anything else we do [as he claims in Science of Language], we also have a coordination problem. To express my thought it does not matter which of the plethora of grasshopper terms [fascinating info BTW!] I use; I know which one I mean. Same for you. So why would we use the SAME term?
Input can't explain it because of the massive poverty of stimulus. Communication can't explain it because language is used in communication just like haircut [what exact haircut would communicate "Hey I'm cool" - well have a look at a bunch of teenagers...]. Even innatism can no longer help you because even IF we have all exactly the same innate lexicon which never changes we still have the labeling problem: why would you and I use the same label for a concept if communication is not what keeps pressure on us doing just that?
I think if you take internalism/innatism seriously there are a lot of bullets to bite. Chomsky or Fodor might be willing to bite them all but I'd bow out very early...
PoS is a problem that doesn't arise in this instance, for one of at least two reasons. One possibility is the old no-overlap principle, saying that terms by default don't overlap in reference. The other is Bayesian fit vs complexity; you see a few things called grasshoppers and find the best balance between complexity of description and the fact that various other things, such as butterflies and beetles, aren't called grasshoppers.
DeleteBayesian is better, I think, because it implies that you could a few bugs were, say, iltywiltywe, without leaping to the conclusion that butterflies were also that. But, as many people would correctly point out, you need to have a definite description language for bug body shapes (and perhaps other features, such as locomotory styles) to flesh this out properly - Marr's ideas seem like an excellent starting point (and Anna's ideas about defining shapes seem to be working better than I expected them to when she started doing it).
Note that your first paragraph presupposes externalism and that words refer to external objects. This is the view Chomsky explicitly rejects for reasons he seems to find very convincing (some of them he discusses in Science of Language). Again, that is not a position I agree with but one Chomsky has consistently defended over a long time (Norbert may know exactly for how long)
DeleteWhat Chomsky says is a bit different, at least as far as I understand him. He says that while animals refer to external objects or internal physical states (”I’m hungry”), language (or most of it) doesn’t. It’s true that most of language elaborates on internal conceptualizations of the external world. Indeed, if our language strictly referred only to external objects, there could hardly be, for example, any Platonist.
DeleteThink of 'reference' as shorthand for "a propensity to classify items that is induced by the speakers' internal representations of the shape, color etc that distinguish the various kinds of creatures." So, my first paragraph says that people are disinclined to accept classification systems that put things into two classes at once (in the absence of clear evidence that it's necessary, as with the 'pet'/'wild animal' divide that overlaps the taxonomic divisions, superordinate taxons, etc. (so the Arrernte have 'arne' for 'tree/bush' (no difference), alongside of species names for all the individual species (of trees and other plants; not of insects, as far as I could make out).
DeleteIn that part of the world, under the original conditions, if your classifications based on appearance don't provide good predictions of dispositions, you don't stay alive very long, so I guess that's a kind of 'externalism', but not one I'd be inclined to dismiss.
Michael Ramscar & his coworkers also have an approach to learning classifications like this based on 'Feature-Label-Order' which also manages PoS problems by maximizing predictive value, & they also have experiments.
Delete"He says that while animals refer to external objects or internal physical states (”I’m hungry”), language (or most of it) doesn’t."
DeleteIt is true that in language we do not have the stimulus response correlation attributed to animal [communication?] here (I say 'here' because there is now some evidence that it is not even the case for all animal communication systems). This is not a new insight but was forcefully put forward in Chomsky's criticism of Skinner and behaviourism. Virtually everyone accepts that there are important differences between language and animal communication so I am not sure why Chomsky keeps repeating this uncontroversial point as if he had to battle a hostile army of Skinnerians.
putting this aside: only because not everything in human language can be accounted for by the kind of interaction with the external world Avery aludes to does not mean NOTHING can. I am not aware of any philosophers who currently hold a pure referentialist account of meaning and most of the more sophisticated accounts have a way to deal with the fact that we cannot get all meaning from 'the outside world'. Given the names Chomsky lists in his attacks against 'referentialist accounts' I have to assume that he rejects all of them. And I have to go by what he says about his own account not by what I think he 'should have said' or what he could possibly have meant. Many people have attacked his view on this issue [again Norbert may know more details than me] and Chomsky has defended himself.
@Avery. It's not really an embarrassment to Fodor's account that some concepts appear to have definitions, because a key part of his analysis in the original Concepts book was that people could "lock on" to concepts via definitions. So, different groups of people can lock onto different subsets of the animal kingdom via different definitions. They key point is that the definitions aren't constitutive of the content of the relevant concepts. So the belief that an inteltye is in front of you is not the same thing as the belief that a non-green bug with a vaguely fingerlike body and big hind legs for jumping plus four short front ones is in front of you.
DeleteFodor's point regarding induction applies just as strongly to your examples as to the standard examples, as far as I can see. You can't test the hypothesis that an inteltye has a fingerlike body unless you already have the concepts FINGER, *-LIKE, FINGER-LIKE, BODY and FINGER-LIKE BODY.
I think you make an excellent case for why work on early acquisition is so important: do we really 'come equipped' [so to speak] with the kinds of concepts you mention or do we learn them from the input?
DeleteAlso how many of the distinctions we as linguists or philosophers can think up are actually meaningful for the average person [e.g. how many people will just think 'yucky thing' when they see an inteltye? Is it really true that we all have the exact same concepts for even 'river'? Or is there just enough overlap that we can use them as tools?
We probably have by now heard most of the theoretical arguments for both sides but in the end, as Chomsky says, 'these are empirical questions'
@alex, yes there need to be some undefined concepts, but that some concepts are undefined does not mean that they all are. So the Wierzbickian story about the bugs would be that the definitions (called 'explications' in that approach) start of with the info 'a kind of living thing', which deals with the standard problems about robot grasshoppers, and probably twin earth ones as well, given some additional fiddling (plus, we don't know how indigenous Australians would react to twin earth scenarios).
DeleteA further point is that the concept of 'definition' developed by philosophers and that segment of the publishing industry that produces dictionaries might well need some serious adjustment to work as a theory of how brains connect words to the world as they encounter it, such that a new term is called for, but it would surely be sensible to stick with the familiar terminology until we have some idea of what the needed adjustments are. Does *anybody* who works on lexical semantics find Fodor's line even remotely useful?
Chomsky is better; his 'tea story' I think is useful. Along the lines that there is a large but limited number of processes that take water as input, and produce something that is mostly composed of H20 as output, and some of them have standard names (tea, whiskey), but if they don't, they're just water (the stuff that comes out of the tap in Cambridge MA).
@christina I don't think we can tell what is innate vs learned from the input at this point, but we can remove a lot of stuff from the list of 'plausibly innate'. Color and related concepts (all the unimodal sensory concepts would be my contention, vigorously rejected by AW) I think can be taken off this list due to the existence of algorithms for training neural networks for recognizing them. For the grasshoppers, some of the plausibly innate ones would be:
Deletething, living, living thing (AW has the last as a composite of the first two, I wouldn't bet a whole lot of $$$ on that)
kind of
move (part of jumping and walking)
touch (part of 'grab')
body
part of (legs etc are part of body)
To be removed from the plausiby innate list, a concept much either be a) defined in terms of things that are on it b) get a creditable procedure for being learned from the input.
Folk vs scientific terminology is another issue, and the gaps between rural people who can distinguish more species from urban who don't etc. is another problem. I suspect that the Arrernte don't distinguish the mantises from the stick insects because they're not interested in either - they think they are poisonous and will give you a headeache if you touch them (the big ones can deliver a painful bite or pinch). The Aussie students in my semantics class last week mostly seemed to know the differences between grasshoppers, katydids, mantises and stick insects.
@vkodytek: the words (or 'word types') in language arguably don't 'refer to things' in any simple way ('kinds of things' is a different matter, but there could be debates about the existence of kinds, I suppose). But expressions composed of those words (ie instances of combinations of words) surely do refer, if the word 'refer' means anything at all. But reference is produced by words in a situation.
Delete@Avery, I don't think anyone suggested that all concepts must lack definitions because some do, so I'm not sure what you're getting at there. Fodor's position is that concepts don't have associated content-constitutive descriptions (i.e., definitions). As he explains over and over again in Concepts, this position is perfectly consistent with my acquiring the concept INTELTYE via the description "non-green bug with a vaguely fingerlike body and big hind legs for jumping plus four short front ones". The caveat is that the relation between this concept and this description is not a definitional one, so that my process of acquiring the concept can't be understood in terms of learning via hypothesis testing. But from a sociological point of view, it's no surprise on Fodor's account that different groups of people have different bug concepts which bear more-or-less transparent definitional relations to each other.
DeleteI wasn't quite sure who was the target of your second paragraph, but I assume you're not suggesting that Fodor himself should stick with the term 'definition'. This seems like a bad idea, since a definition (by definition!) is content-constitutive, and Fodor is denying precisely that there are content-constitutive descriptions associated with (most) concepts. Maybe someone will come up with a new definitionist model of concept acquisition based on some new notion of definition, but vague hopes of future success don't cut much ice against presently-existing proposals.
"so that my process of acquiring the concept can't be understood in terms of learning via hypothesis testing." - I guess I am still failing to make heads or tails of Fodor, since hypothesis-testing seems to me to be exactly a way in which concepts of this nature can be acquired.
DeleteOr is this hingeing on something such as the difference between dictionary definitions, composed of actual overt word-forms, and 'cognitive definitions', composed of 'concepts', frequently associated with but not identical to overt word-forms. For better or worse, I don't have time to re-read Concepts ATM.
Continuing with this, I don't see any reason why calling lexical decompositions 'definitions' is any worse than calling the things that generative grammarians produce 'grammars'; arguably the same confusion of a branch of the publishing industry with something that a putatively scientific activity finds to be an aspect of certain organisms, but not one worth making a fuss over in our present level of limited understanding. Nobody but the outer fringe of methodological moaners has any problem with understanding the difference between publisher's grammars and generative grammars, and nobody gets into any substantive difficulty by applying the same term to both.
DeleteOK, thanks Norbert, this is helpful. Hermann Paul, in his *Prinzipien der Sprachgeschichte* (1882), observed that what I called 'polysemy' makes it possible to express many more concepts than we have words. So, it seems, the idea that words simply label concepts is still wrong, even if all concepts are innate.
ReplyDeleteSusan Carey's recentish book "The Origin of Concepts" is worth reading if anyone here still believes in Fodor-ian radical concept nativism.
ReplyDeleteI am a little surprised that it is still taken seriously -- most people (apart from Fodor) take it more as a reductio than as an actual empirical claim that humans have an innate concept of 'carburetor'.
According to Science of Language [and also 'Of Minds and Language' Chomsky takes it seriously. And according to a couple of talks I went to that left his audience open-mouthed in disbelief McGilvray takes it seriously. Have a look at his appendix XII in SoL and some of the work of his own he references...
DeleteI just finished 'Science of Language' and Chomsky makes it quite clear that he does NOT agree with Fodor on this. In fact he has made it quite clear over the years in various different places. So I wonder why the other poster would claim that he does. Quite strange.
ReplyDeleteI assume you mean the short exchange on p. 34. What is at issue there is whether concepts are atomic. Chomsky does indeed disagree with Fodor on that. What Alex and i talked about was whether concepts are innate and on that point Chomsky and Fodor agree.
DeleteYes, Alex made the point about carburetor. And Chomsky doesn't believe in that kind of "radical concept nativism". He says it would be along the lines that Fodor suggested but "not as extreme as his position". I would think that Chomsky's position is more nuanced.
Delete