"What is Language and Why Does It Matter" - Noam Chomsky

I'm delighted to introduce Noam Chomsky widely regarded as one of the leading thinkers in human history the chomskyan revolution in place language study in a psychobiological setting and played a critical role in fueling the modern reemergence of the cognitive sciences Chomsky Scientific Revolution resurrected 17th century rationalist principles and then implemented them by analyzing human cognition in terms of a formally explicit computational representational theory of mind the idea being to study the properties of human beings including human cognitive capacities as part of the physical world using the methods of normal science Chomsky's latest incarnation of the generative Enterprise the minimalist program further advanced linguistic theory as normal science by continued endorsement of Galilean experimental method and by articulating goals fundamental in the natural sciences including the central goal of seeking explanation through theoretical simplicity and by possible unification with related fields Chomsky's understanding of and commitments to normal science rationality abstraction and explanation continue to this day to pioneer human understanding of human nature of course who we are as vitally important humanitarian implications regarding how humans should be treated with repercussions in philosophical ethical social political psychological and economic arenas fields to which Noam has also made profound contributions for well over a half a century while continued to do so to this very day the title of his talk tonight is what language and why does it matter Chomsky well the title that was supposed to be announced on our that was added a personal perspective and yes it's a personal perspective because not only is there no consensus about what I'm going to talk about but actually it's a distinctly minority view and the whole range of fields that kind of converge around this topic there is consensus on one manner from the second half of the 20th century there's been a huge explosion of inquiry into language by any measure you take skill character depth of anything far more penetrating work is going on into a vastly greater array of theoretical issues conceptual issues and just psychologically varied languages far beyond anything before many new topics have been opened the questions that students are working on the day could not even be formulated or even imagined a half a century ago or that matter much more recently and new problems and puzzles are coming into view more rapidly than old ones or being at least partially resolved now these are all very positive signs of a lively exciting array of entertaining disciplines I've got my own reservations about things with Joe balloon as I continue I think it's fair to trace this sharp and unmistakable change in large part to new options that became available in the mid 20th century options for considering more seriously the most fundamental question about language and they leave what is it by that time advances in the formal Sciences had enabled a clear formulation understanding of what we ought to recognize to be the most basic property of line so refer to it from now on it's just the basic property namely that each language provides an unbounded array of hierarchically structured expressions that receive interpretations of both internally and externally externally through the sensorimotor system internally as some rather obscure system thought that we know is there but not much about conceptual intentional interface sometimes called for mental processes that allows the ability to formulate this for the first time really allows a substantive formulation of classical dictum goes back at least to Aristotle that language is sound with meaning before that you could say exactly nor demand especially the word with was obscure was it mean sound with mean work of recent years as I'm sure shows that sounds much dinero but all keep do it for simplicity and I think there's a pretty good reason all returned to it to suggest that the classic formulation is misleading and important ways and ought to be revised well just given that much follows that each language incorporates a computational procedure which satisfies the basic property when I just mentioned and therefore by definition a theory of each language is what's called a generative grammar and each language is what's been called an I like I here stands for internal an individual internal individual and intentional intentional here with an S means you're interested in discovering the actual computational procedure not some set of objects that it enumerates technically that would be what it strongly generates there was another there are other notions like weak Jenner capacity and what some people not may call a language whatever that is these are derivative notion have to apologize for that term it's actually my term but it's not used the way I would ever use do not understand how it's used frankly it seems to me something like something like that but that's not a coherent notion I don't know if a generative capacity in the e language or whatever it is or even definable for natural language these are questions that were discussed quite a lot about in the 1950s so laate seems to have been forgotten on judging by the literature well correspondingly every approach to the language no matter what it is such a linguistic whatever should recognize at least that one perfectly obvious fact each language is a property of an individual it's internal to that individual mostly to the mind brain that's what sometimes now called the wild linguistic framework should be a truism and whatever this core concept must be understood to be a prerequisite to any further enquiries there plenty of further engrams they doing all kinds of topics of acquisition use origin of languages like Josiah the internal mechanisms that implement the systems that means both the system knowledge competences and the various uses the performance too distinct but related tasks well evidently investigation of these further topics that relies on guidelines may be there at least tacit that they have to be there somewhere which are provided by the answer to the question of what language is that shouldn't be considered controversial so for example no biologist would dream of proposing an account for the development or the evolution of the eye without telling for something pretty definite about what an eye is and the same truism truism should whole of enquiries into language regrettably but it should in earlier years before this shift became possible the basic property did resist clear formulation in fact it was also even true in mathematics even the notion of proof basic notion wasn't really closely understood until a little over a century ago the quest accede morsel so you take some of the classical literature for example the serve centenary coming up right now for him language in the relevant sense of language is a storehouse of words and images in the brains of a collectivity of individuals founded on what he called a sort of contract for Bluefield language and quoting is an array of habits to respond to situations with conventional speech sounds and to respond to these sounds with actions I think it's another difficult definition his postulates for the science of language 1920s here are languages the totality of utterances made in a speech community that's something like William Dwight Whitney's traditional conception of language as the sum of words and phrases by which any man expresses his thought that's what he called audible thinking that's a slightly different conception and ways to which I'll return Edward Sapir defines language as a purely human and non distinctive method of communication communicating ideas emotions and desires by means of the system of voluntarily produced symbols those are the classics there's others from less famous figures which are similar and with such conceptions it's perfectly natural to follow what the Martin Joe's called the bow as Ian tradition referring to Franz boas holding that languages can differ arbitrarily and that every new language should be studied without any preconceptions obviously can't be literally true but something like that seemed natural so accordingly linguistic theory can consist of nothing more than collection of modes of analysis procedures of analytic procedures to reduce corpus to some organized form basically procedures of segmentation classification worked out considerable detail both in European and American structure linguistics well the shift of perspective to generative grammar within the bio linguistic framework opened the way to much more far-reaching inquiry into language and language related topics and also greatly enriched the variety of evidence that bears on the study of each individual language so it's not just a matter of organizing the date of that language but she's studying say Japanese you can study acquisition neuroscience dissociations of linguistic and other cognitive capacities much else and you can also tree studying Japanese have learned from what's discovered in the study of you know well any other language that's all on the basis of a pretty obvious and quite well confirmed assumption that the capacity for language relies on shared biological properties as far as we know virtually totally share in group differences that's the topic of ug universal grammar contemporary adaptation of a traditional phrase doesn't mean what it meant traditionally well in earlier years it was understandable that the question what is language it received only such indefinite answers as the ones I've just alluded to ignoring completely the basic property it's however I think kind of surprising to find that similar answers remain current in contemporary cognitive science not untypical as a study in a current journal frontiers of psychology study of evolution of language at characterizes language only the despite two well-known people skip the names characterizes language as the full suite of abilities to map sound to meaning that's language that's basically a reiteration of Aristotle's dictum it's much too empty to ground any further inquiry again no biologist would study evolution of the visual system assuming no more about the phenotype then that it provides paraphrase the full suite of suite of abilities to map stimuli to percept couldn't get off the ground that way or in this case well there are also broader reasons to be concerned with the question what is language is a fairly clear indication of these and some interesting comments by one of the leading scientists who studies human evolution recent book by Ian Tattersall it's a review of currently available scientific evidence he observes that it was once believed that the evolutionary record would yield early harbingers of our later selves the reality however is otherwise for it's becoming increasingly clear that the acquisition of the uniquely modern human sensibility was instead an abrupt and recent event he actually dates it in the very narrow window of about fifty two hundred thousand years ago and goes on to say that the expression of this new sensibility was almost certainly crucially abetted by the invention of what is perhaps the single most remarkable thing about our modern selves namely language therefore an answer to the question what is language matters are greatly to anyone concerned with understanding our modern selves Homo sapiens the founders of modern biology of course they lacked the evidence of current science that Tattersall is reviewing but they adopted kind of a similar view so Darwin for example wrote that man differs from animals solely in his almost infinitely larger power of associating the most diversified sounds and ideas that's an infinite version of Aristotle's dictum of course the phrase almost infinite that's a traditional phrase but we should interpret it as meaning infinite because there's no sense to almost infinite you can't and it also similarly makes no sense I'm sorry if this offends lovers of big data but it makes no sense to contemplate huge finite non extendable lists that's close to meaningless something else that should be kept in mind unfortunately well even earlier than Darwin at the origins of modern science Galileo it was entranced by what he called a marvelous invention that provides the means to construct from 25 or 30 sounds the infinity of expressions that enable us to reveal everything we think and all the movements of our soul of our mental acts we would say that's audible thought and Whitney's phrase but Galileo went beyond by recognizing the unbounded character of each language it's unusual it's an obvious point in fact if you look over the whole history of 2500 years of history of inquiry and the language extremely hard I've been able to find four or five cases where anything actually pointed this out explicitly maybe they knew it well the same recognition as Galileo's and a much deeper concern for the creative character of language use normal use of language now that pretty soon became a core element of cartesian science what we call philosophy there's no reason today to doubt the fundamental insight of Descartes that use of language has creative character it's unbounded typically innovative no limits it's it's appropriate to situations but not caused by them that's quite a crucial distinction and it can engender thoughts and others that they could have they recognized they could have expressed themselves without limits it's a critical insight crucial for the history of philosophy but for modern should be for modern cognitive science and linguistics today we should also bear in mind that a an aphorism of Humboldt that's often quoted these days neighbor that language he said that language involves infinite use of finite means he's talking about use there's been a lot of progress in understanding the finite means that are that make possible infinite use but the latter notion infinite and appropriate use that remains as much of making mystery as it's ever been though there has been some progress in understanding conventions that guide appropriate use well a century ago Otto Jespersen raised the question of how the elements of language and quoting him come into existence in the mind of a speaker on the basis of finite experience yielding a notion of structure that's definite enough to guide him in framing sentences of his own crucially he said free expressions typically new to speaker and hearer so again he's alluding to the infinite unbounded character of language so the task of the linguist then is to discover those making how they arise in the mind go beyond that to unearth what yes person called the great principles underlying the grammars of all languages Eugene are terms and by unearthing these great principles to gain a deeper insight into the innermost nature of human language and of human thought those are ideas that sound much less strange today than they did during the structuralist behavioral science europe that came to dominate much of the field marginalized Jefferson's insights well reformulating yes persons program today the basic task is to investigate the true nature of the interfaces and the generative procedures that relate them to determine how they arise in the mind how they're used if we can ever get that far the primary focus of concern of course being free expressions well as soon as the earliest attempts were made to construct explicit generative grammars roughly 60 years ago immediately many very puzzling phenomena were discovered they never really been noticed as long as the basic property was not clearly formulated and addressed and cinta there was syntax of course but it was considered just use of words determined by convention and analogy which gets you nowhere actually this is you go back sixty years ago it's kind of reminiscent of the very earliest days of modern science around 1600 for millennia scientists had been the greatest scientists had been satisfied with simple explanations for familiar phenomenon so for example rocks fall and steam rises because they're seeking their natural place objects interact because of what we're called sympathies and we perceive a triangle because its shape floats through the air literally and implants itself in our brain and so on those were the received answers in the sciences as soon as Galileo others allowed themselves to be puzzled about these facts modern science began and it was quickly discovered that our beliefs are all senseless and our intuitions are mostly wrong the willingness to be puzzled is a very valuable trait to cultivate it's from early education to advanced inquiry unfortunately that's much too little recognized in the human sciences in the physical sciences by now it's routine but it wasn't not very long ago well one puzzle that came to light about sixty years ago and remains alive and I think highly significant has to do with very simple but curious fact so consider the sentence Eagles that fly swim simple enough so enough to write it on the non-existent blackboard over there and then put a word in front of it say the word instinctively instinctively Eagles that fly swim or the word can-can Eagles and fly swim well there's a evident fact about that the words instinctively and can they link to a verb but they link to swim not to fly so the thought takes sentence can Eagles the fly swim there's a thought that it can't express with can being associated with fly and that's pretty difficult to formulate it's a fine thought but try to formulate it even with circumlocution very hard and it impedes communication one of many such cases but it's somehow part of the design of language well what's puzzling about this is that the Association of the clause initial element instinctively or kin to the verb is remote and based on structural properties not proximal and based on linear properties linear procedures are far easier to compute so language makes use of a property of minimal structural distance it never uses the much simpler operation of minimal linear distance that's sometimes called structure dependence of rules and the puzzle is why it should be so not just for English but for every language not for these constructions but for every construction and even where data for the child's learning the language is ludicrously small or in the case of instinctively non-existent totally never any errors no no alternatives this is just reflexive so why well there is a very simple explanation for it namely the child take the child reflexively knows the right answer in these cases in all the cases because even though the evidence is slim or non-existent and the reason is that linear order is simply not available to the language learner who's confronted with such examples the language learners is guided by a principle of ug one of those great general principles that yes person was alluding to didn't think of this one a principle of ug that restricts search to structural distance minimal structural distance that's a plausible explanation and as far as I know it's unique I don't have any other proposed explanation I mean there's proposals with the quickly shot down it's somehow resistant it's the explanation is resistant in fact dismissed which i think is a sign of the immaturity of the field I think it's kind of like the sciences in the 14th century or something you have a fine explanation but for ideological reasons you can't can't accept it the prince the general principle of minimal distance is used all over the place in language design it's presumably one instance of a much more general principle that enters into design and acquisition of language and elsewhere to call it minimal computation computation tries to be as as as a efficient as possible and the evidence shows strikingly that human language invariably makes use of minimal structural distance rather than linear distance in every relevant cases no non exception now despite the far greater simplicity of linear distance which is a puzzle but I think the answer is what I stated there is some supporting evidence in this case from the neurosciences there's a research group in Milan Andre Mauro many of you know as the linguist involved they studied brain activity of subjects who are presented with two types of stimuli these are all invented languages some of the invented languages satisfied ug minimal structural distance and other ug principles others were designed so as to violate ug principles so for example a rule of negation that places the negative element after the third word it's a linear distance much very simple computation much simpler than the ones in natural language well what they found is that in the case of Conformity to ug there's normal activation in the length and the standard language areas but when linear orders used it's just diffuse activation over the large parts of the brain and that task case the test is apparently being treated it can be solved but treated as a non linguistic puzzle there's analogous work by Neil Smith and Jung theme Mariya simply is colleague who work because maybe you know with a cognitively impaired but linguistically fluent subject and similar experiments reached the same conclusion actually there's a small industry and computational cognitive science trying to show that these properties of language can be learned by statistical analysis of massive data large number of papers on this every attempt that's clear enough to investigate has been shown to fail and irremediably can't can't do anything with them which is but it really doesn't matter because the efforts are beside the point in first place suppose they were to succeed which happens to be a virtual impossibility but suppose they were that would leave entirely untouched the only question namely why does language invariably use the complex computational property of minimal structural distance and why does it never employ the far simpler option of minimal linear distance actually that's it that question somehow can't be seen that's I think a good illustration of the unwillingness to be puzzled that I mentioned earlier the first step in serious scientific inquiry was recognized in the hard sciences since Galileo and until it penetrates the cognitive sciences they're never going to get off the ground I think that's pretty evident well a broader thesis still is that linear order is never available for computation not just this kind of case but never at least in the parts of language that involve syntax and semantics so-called semantics the core parts so why do you have linear order well there's obvious reason for it the sensorimotor system requires it you can't talk in parallel so you talk linearly but so that requires somehow that whatever's going on the mind be sent through a pass some kind of a filter that makes it come out with linear order that's the sensorimotor system which is not specifically adapt at the language bender the parts that are essential for external authorization of language and perception they appear to have been in place hundreds of thousands of years before language emerged in fact the chimpanzees have apparently pretty close to the same auditory system that humans do they even pick out pretty much the same phonological features but it's just noise as far as they're concerned well the matter isn't settled but - there's very considerable evidence that this broader thesis may in fact be correct so if so the basic property is not the way I formulated it before and the way it's formulated in the technical literature my papers - rather the basic property should be the generation of an unbounded array of hierarchically structured expressions that map to the conceptual intentional interface to the mental system system of thought and the rest is kind of ancillary tacked on well if that's correct and it seems to be there's a good reason to return to a traditional conception of language as an instrument of thought it's to revise Aristotle's dictum accordingly language is not sound with meaning but meaning occasionally with sound more generally some form of externalization typically sound though by now pretty clear that it's modality independent and in fact externalization is rarely used if you think about it by far overwhelmingly the most use of language is never externalized it's what sometimes called internal dialogue is very limited research into that so it could be studying I think the reasons there's limited researchers again kind of ideological not intellectual it's interesting topic but if so most of the research is basically introspection yours is as good as anyone elses at least my introspection you can think about it yourself is that what reaches consciousness in internal dialogue walking long thinking or something what reaches consciousness is just fragments tiny fragments and then since the fragments come a fully formed expression can be formed in your mind usually is can do it complicated fully-formed expression and it's instant it's far too quick for articulate errs to be involved or probably even instructions to articulate errs and so often not produced even internally so it's all going on somewhere inaccessible to consciousness that's an interesting topic and it could be studied could be explored and thick waste doing it but basically hasn't been little but not much well issues like that aside investigation of the design of language which is the starting point for any further inquiry gives pretty good reasons to take seriously the intuitions of Galileo and others that language is essentially an instrument of thought an externalization then would be just an ancillary process tacked on now an end and there's plenty of further investigation that supports that conclusion won't go through it but if it's established at least to me it looks pretty sound it follows that particular uses of expose language that depend on external ization are even more peripheral aspects of language one of them is communication that's actually contrary to a virtual dogma in all the related fields that has no support that I know of but that just pervasive languages commonly describe this print somehow in essence a means of communication it seems to be anything but that would also follow that much of the extensive speculation about language origins about evolution of language is just on the wrong track to begin with it's treating it almost always as something about speculations about the evolution of communication is a totally different topic well the matter this matter hasn't been studied either though it could be but I suspect that the modern doctrine the that communication is somehow the essential form of language function of language it probably derives from the powerful influence of association estándares assumptions they retain a very strong grip even when people who publicly reject them along with that there can easily find highly oversimplified and quite untenable interpretations of modern evolutionary biology it's an interesting topic but on put it aside well let's return to the basic property now we reformulate it the computational system of I language yields an unbounded array of hierarchically structured expressions mapping to the mental system the conceptual intentional interface there are ancillary processes that may or may not externalize them in some sensory modality naturally we seek the simplest theory of the basic property of the theory with the fewest arbitrary stipulations any such stipulation of part being from being unwanted justum normal methodological grounds is also a barrier to some eventual account of origin of language and I stress eventual because we're not even close well this is standard scientific method we ask how much can we how far will that carry us the simplest there's simple simplest of all computational operations it's embedded in some manner in every relevant computational procedure is an operation that takes objects column X and y that have already been constructed and forms a new object call it Z that's that's the operation it's now sometimes called and the question is what is it what's the operation that does this well the principle of minimal computation entails that neither X nor Y should be modified in this this this process that's minimal computation and that they should be on order ordering would add further complication and as I just mentioned that's a conclusion that's quite strongly supported on other grounds well what that means is that merge is just set formation so merge of x and y just gives you the set containing x and y now of course that doesn't mean that the brain contains says I reason I mentioned that is that there misinterpretations and the current literature which dwell on this topic fulminate about it in fact what it means is that fundamental whatever is going on in the brain which we don't know has properties that can be characterized in these terms this is again commonplace in Sciences like chemists that don't expect to find the calculated diagram for benzene and a test to know they're not confused about that only linguists are confused about such things suppose x and y are merged and neither as part of the other so as in combining say read and that book to form the syntactic object wherever it is the corresponds to read that book that's called external merge I suppose that one is part of the other so say Y is part of X so you combine say which book and John read which book and you form which book John read which book that surfaces is which book - John breed by further operations all come back to them that's an example of the ubiquitous phenomenon and natural language displacement phrases are heard in one place but they're understood both in that place and somewhere else so the sentence which book the John read is understood something like for which book X John read the book X now in this case the result of merges again the set X Y but now there are two copies of Y one is the original one remaining inside wha X the others the copy that's merged with X ok that operation is called internal merge notice incidentally contrary to some confusions about this that there is no operation of copy formation and no operation of reemerge that exists it's just plain urge simplest possible ways set formation internal merge and external merge are the only possible cases of binary merge so if we assume the merge is a binary operation these are the two options there's nothing else and notice that both of them come free it would take a arbitrary stipulation to bar either one of them and that's a pretty important fact it's important as importance has only been gradually sinking in for some years since it was noticed in the late 90s by he's a Keith ihara first which I just discovered for many years it was assumed by me in particular that displacement is a kind of an imperfection of language some strange thing that has to be explained away by some more complex devices and assumptions of ug but that turns out to be incorrect displacement is what you expect on the simplest assumptions it would be a problem if it didn't appear that would be a affection it can in fact be plausibly argued that internal merge is actually simpler than external emerge if you think it through it requires much less memory but essentially they're both they're free and it's a problem with the language doesn't have either one of them there's another important fact about a internal merge in its simplest form that is satisfying the principle of minimal computation it yields the structures that are appropriate for semantic interpretation in a quite a broad variety of cases it's illustrated in the simple case of which book did John read as I said it really means which book for which book X John didn't read the book X which is what you get for an internal merge however of course that's the wrong structure for sensorimotor system the sensorimotor system drops the copy universally in language only the structurally most prominent copy is pronounced the lower copies deleted acted as a revealing class of exceptions at which in fact support the general thesis but put that to the side why do you have deletion of copies well that follows from another application of the same overriding principle of minimal computation namely essentially pronounced as little as possible got to pronounce one of them there's no evidence that the operation took place would do as little as possible there's a result the result is that the articulated sentences have gaps and the hearer has to figure out where the missing element is well it's well known in the study of perception in parsing that that yields quite difficult problems of interpretation fact these are some of the most standard parsing problems filler gap problems are cool so in this quite broad class of cases language design favors computation and disregards complication in the use of language right so it said language designed to be computationally perfect but no good for communication that fits the other things that I say notice that any linguistic theory that replaces internal merge by other mechanisms has a double burden of proof to bear first it has to explain why it's tip why does it have the stipulation barring internal merge and second it has to give a justification for whatever no mechanisms are yielded are in design to yield the displacement phenomenon in fact displacement with coffee is noticed because that's generally the right forms for semantic interpretation well this holds interestingly for much more complex cases but I'll skip them that would require a blackboard but just as the but they're quite an interesting class of complex cases that work exactly like this but and just as in the simpler cases like say instinctively eagles fly swim it's absolutely inconceivable that any form of data processing yields these outcomes relevant data simply are not available to the language learner and the results therefore must derive from what David Hume called the original hand of nature our terms genetic endowment specifically ug universal grammar and in ways like these we can derive quite far-reaching and firm conclusions about the nature of ug side comment on the literature linguistics philosophy psychology very common claims and current literature technical literature that there are no genuine linguistic universals the no ug the reference is not the ug just confusion the reference is the descriptive generalizations so for example Joseph greenberg famous universities which are quite interesting but they're generalizations and generalizations are quite likely to have exceptions that's the nature of generalizations so for example the generalization I mentioned about the leading copies it has some exceptions quite interesting ones which strengthen the principle behind it and that's all over the sciences in the standard sciences that's understood so for example 19th century there were take a famous case there was discovery of perturbations in the orbit of Uranus if that had been linguistics not astronomy would have led to the conclusion ok let's throw out physics because there's a generalization and they here there's a problem about the perturbations that shouldn't be there well since this is science and not linguistics it scientists went on to try to figure out why and sooner or later they found that tune explained the perturbations and exceptions to largely valid generalizations played a similar role all over the place in the sciences and repeatedly in the study of language too but there is a strange curious pre scientific belief that if you find exceptions to generally valid generalizations of a means they've got to throw out everything you can find plenty of that in the literature well putting those I think perversions aside you can conclude I think that if language is optimally designed it's going to provide structures that are appropriate for semantic interpretation but that yield difficulties for perception hence for communication and there are many other kinds of examples that illustrate the same conclusion structural ambiguities for example or garden path sentences particularly interesting cases islands there aren't too well understood partially understood so take say we're called ECP constructions so take the sentence they asked if the mechanics fix the cars you can ask how many cars and you can ask how many mechanics the sentences are how many cars did the mechanics fix how many mechanics how many mechanic does it say how many cars did they ask if the mechanics fixed how many cars did they ask how many mechanics today as they fix the cars they're different strikingly different that's an e CP violation and the one that you can't say is a fine thought but you have to express it through some kind of circumlocution again that impedes communication all islands are like that plenty of similar cases they're partially understood it's a big task to study them completely but insofar as they're understood these structures follow from the simplest free application of the simplest rules yields difficulties for perception and quite generally to repeat where there are conflicts between communicative efficiency and computational efficiency in every known case communicative efficiency is simply disregarded sacrificed and that lends further support to the revision of the common sense Aristotelian dictum and support for the view of for wreckage action of that and support for the alternative traditional view of language as an instrument of thought with communication and other uses being side properties ancillary properties actually that conclusion fits pretty well with the very limited evidence we have about the emergence of language this Tattersall pointed out its apparently quite sudden and very recent than the evolutionary time skill that's a fair guess just looking at that and we know quite confidently that there's been no change no evolutionary change no detectable change ever since humans began to scatter around the earth that's 5060 thousand years ago few left Africa and quickly we're all over the place there's no detectable change since then so what you seem to have is no detectable change in maybe 50 60 70 thousand years and nothing around maybe twenty or thirty thousand years before that you can change the numbers a little if you like it doesn't matter much that's a very narrow window in evolutionary time and a pretty fair guess and that's about all we know about evolution of language I should say huge literature based on absolutely nothing these are the only things that are known the rest is fantasy it's an interesting kind of pathology in the field well I think a fair guess is that some slight rewiring of the brain in an individual of course yielded merge unbounded merge that provided the basis for unbounded and creative thought sometimes called the great leap forward which archaeologists find in the archaeological record and they very remarkable differences that separate humans from their predecessors and from the rest of the animal kingdom as far as we know it is just as the Cartesians recognized at extremely sharp divide nothing mystical about it that's things like that happen in biology no no well these remarks all stop here only scratch the surface what I hope they can do is to illustrate why the answer to the question what is language matters quite a lot and close attention to this fundamental question can yield conclusions that have many ramifications for the study of what kind of creatures humans are okay so we have about 40 minutes for Question and Answer and there's microphones two microphones one there and one there so if you want to step up and ask a question please feel free to do so hi I'd like to ask about how do you see how would you like to see the surrounding fields interact with linguistics given that linguistics formal linguistics has become quite technical a lot of philosophy looks very exotic to a lot of linguists a lot of people are content to work on their own very specific technical work how would you like to see philosophy psychology other cognitive sciences linked in with linguistics well all of these are philosophical issues in fact they go back centuries they Carter it's called the philosopher humans golda philosophy a lot was a philosopher they dealt with these problems in interesting ways and modern philosophy may not deal with them but if so that's a comment about modern philosophy not about philosophy there are obviously psychological questions as far as I can see linguistics just is part of psychology so can't ask what the relation is it's like asking how is perception related to psychology so I don't see a question these are psychological problems by definition and throughout the whole history of philosophy until pretty modern times there's been core philosophical problems they still are to some extent but not the way they were in the throughout the tradition I think the tradition is worth saving in this case I first ah thank you very much for your talk at this evening and I my question is about the third factor principles you proposed in recent papers and in those papers you hold that the third factor principles include principles of efficient computing computational efficiency and principles of natural law so I like to hear your thoughts on principle organism independent principles of natural law that are likely to so what is position law would apply the language yeah well I mentioned one which I suspect as a general principle of international of wrong topic some of it that's why I have two sides of the brain so the principle of minimal computation that probably is all over the least organic world maybe the entire world with different kinds of application and I assume there are others after all weird you know this is the language is just you know it's kind of like an organ of the body it's a subsystem of the organism which is develops out of the normal ways and is subject to whatever laws hold of organisms not not too much is known even in general biology about this general biology is pretty much a descriptive field it's just it's not a field that has a lot of theory or laws because it's too complicated but a principles like this probably do apply maybe others you can find something that's great yeah thank you I should say that even simple ones like this carry it pretty far you can go quite far with just pursuing the few things that we sort of partially understand try to give some indication I think it goes far enough to show that an awful lot of what goes on in the field is just seriously misguided hi I have a two-part question so the first question is if I understood your talk correctly this evening the idea is that the computation system is meant to feed the CI interface and then sometimes this comes out as an utterance but not always so this might be a thought for example one that doesn't map to the sensorimotor interface is that correct so if there's a mapping it's just a generation of something at CI maybe it's a mapping a narrow synth ecstasy or maybe some other thing but it ends up at that interface then it's a thought okay so then does that mean that on this in fact an interesting case is whether this exhausts thoughts I mean has been suppose that that exhausts thoughts so he reads say from Humboldt again he argues you know suggests that that is the totality of thoughts the things that could in principle be are too expressed in language that's an interesting question okay so then part two would be then so if we're trying to account for ungrammatical ungrammatical sentences that speakers are Derwood we have to then say that the source of ungrammaticality would have to be at the sm interface where something is gone wrong maybe the wrong copy has been deleted or something like this and this is where we would then find the locus of ungrammaticality what does this have to say about ungrammaticality yes the first of all and romantic ality is kind of a funny notion the whatever is in your head assigns some kind of interpretation even the word salad okay that that means that that's generated we can call it ungrammatical if we like but that's a kind of a theory internal notion among the various kinds of expressions they have many different dimensions some loosely are called more or less dramatical others are called more or less appropriate other many dimensions but it's you know there isn't a split between grammatical and ungrammatical that's incidentally one of the reasons for those of you who know the technical literature why the work on weak weak generative capacity is mostly meaningless because it assumes a sharp break between what's grammatical and what's ungrammatical and it's simply not the way natural language works in fact what are called and grammatical expressions are used all the time perfectly naturally perfectly appropriately in fact every metaphor is an ungrammatical expression if you say say misery loves company or something it's an ungrammatical expression but it's certainly meaningful literature uses on grammatical expressions all the time purposely because they're evocative they force the hearer to construct you know something in their own minds to kind of fill out what's missing so they're perfectly meaningful the part of language they're determined just as much as everything else is so again this if you go back to work in the 50s this was discussed fair amount I mean no real answers but it was discussed and should be good okay well I guess I might more and more of a technical sense of like a derivation crashing for example that maybe we would we would we would think that this might happen at the SM interface since we wouldn't tend to think we would have word salad thoughts for example I guess we could could can you have word gel at thought we'll try it I guess I can I mean try reading to say Finnegan's Wake or e Cummings it's pretty close it's not of course it's not word salad it's contrived and constructed but the first time you read it it's word so you think about it some more something comes through maybe but it certainly means something and in fact even with less exotic examples literature just uses it normally and so does normal speech often not word salad but things that are designed to violate principles and have their own interpretations they take a look sometime at a book like a books of literary theory like William Empson seven types of ambiguity about fifty years ago what he points out is that that's the essence of poetry it's to try to put things so narrowly and so economically that the reader is forced to contrive a world of interpretation that's perfectly decent use of language and normal speeches like that too it's often highly elliptical for example the hearers was just filthy thank you Thank You professor Chomsky it's an honor to have you here in as a student of linguistics which I am I love linguistics I enjoy studying it very much and I think I believe strongly that scientific inquiry is one of the best things that we humans do I also have no good answers for the for the question about to ask which is given all of the misery in our world all of the social economic and other problems how do we justify pursuing something so comparatively esoteric I think I'll tell you a story battlin was that a linguistic Institute I think maybe the first time I went to a Summer Institute of linguistics was probably the early 60s maybe 1962-63 was in the Indiana I had just come back from civil rights demonstrations in the South one of Jackson Mississippi I don't know how much of you know about this stuff but they were pretty violent I mean the State Police just went berserk you know they were beating everyone bloody people were fleeing to the steps of the federal courthouse and the marshals who were sent from Washington to protect the demonstrators were throwing the demonstrators back into the crowd so the state police could smash of the pieces in the evenings people would gather in black churches try to get their courage back up to go out the next day and this went on for day after day anyway I was there for a while I I came back and went to the linguistic Institute at at Indiana and I just happened to run into a kid there who had come from the same demonstration we we'd met down there and we were walking across campus together much and he suddenly turned to me and he said how can they be so interested in phonemes it's a good question it stuck with me ever since I think that's your question I think it can be I mean I think you can both be concerned with the problems of the world and the problems of intellectual importance which tell us in this case tell us something about something I'm not directly related to problems of the world but about as close as you can come in the sciences what's the nature of human beings okay I don't think that's an answer to your questions kind of question you have to answer for yourself thank you I was afraid about so thank you very much for being here this evening so it's fairly clear that linguistic derivations rely on some notion of economy and one of the culminations of economy constraints is phase based derivation and a common critique to that type of thinking is that the capacity for human memory has been largely underestimated what would your reply to be DS 2 - a critique like that and we know if if you're talking about you memory first of all with distinguish between short-term memory and long-term memory okay so short-term memory is pretty limited it's pretty much like other animals that that there's research like for example a lot of interest these days that there shouldn't be but there is an embedded sentences sentences with embedded it's mostly misunderstood but it was worked on 50 years ago and it was found that in normal speech you almost never have embedding very limited embedding it's just too much computations they can't do it it doesn't mean you it's not there and it's kind of like saving arithmetic if you look at people's use of arithmetic in their heads it's extremely limited if numbers get big can't add them that doesn't mean you don't know arithmetic you know arithmetic just have to have external memory it's kind of like a computer's name the same is true of embedding so if you listen to speech you're not going to find much it's mostly paratactic more or less because too much memory limits that's short-term memory but what about long-term memory well that's pretty large like a normal person may know say 50,000 words but doesn't do you any good George Miller who worked on memory lock during the back in the old days he wants just did a calculation of how many pretty much grammatical sentences there were of the length of the reader's digest reader's digest is eighth grade reading level okay so how many grammatical sentences would you have of eighth grade reading level it was greater than number of particles in the universe I mean you can't talk about memory it's nowhere in the right dimension so yeah you can memorize a lot of words and you can recognize a lot of people more or less they were specialized kinds of memory but in this domain it does nothing for you thank you very much for this wonderful lecture actually throughout your lecture I feel that you somehow completely dismissed the research program that is based on characterizing the language language based on some usage based or some mainly communication efficiency and I find it quite unfair because it I think nobody nowadays denies that there is some genetic basis for language and even if we accept the hypothesis that language is optimized for computation rather than communication not all the grammars that are allowed by ug for example are actually implemented so there could be some limitation or some constrain that could be imposed by the necessity of the communication so I'd like to have your thoughts on this yeah so I've dismissed communication you basis studies if you want to carry them out it's okay this wisdom I think it's extremely unlikely than anything will be learned from just as if say physicists started say taking videotapes of things happening outside my office window leaves flying around and stuff like that they could get a lot of data in fact they could get a pretty good prediction of what's going to happen next in fact they do way better in the physics department does but try to get a thesis for that well in the sciences you can't because they're not interested in something in some kind of rough generalization you can pick up by looking at a lot of data they're interested in understanding things okay so even if that non-existent videotape experiment couldn't get a better prediction than the physics department could which it certainly would that nobody would care and I don't understand why it should be any different in the case of language I mean if anything can be discovered from a data analysis fine let's see it it doesn't look very likely it hasn't happened in other fields but if so okay as far as communication is concerned you know it's an activity language is one languages sometimes used for communication it's only one of the many means of communication I mean everything we do is a means of communication how you comb your hair you know what clothes you wear just about everything you do is some kind of presentation yourself it's saying something and language is one of the ways of communicating it's not the main thing in language but it's part of it and you can study communication in fact you can study evolution of communication so for example Mark Howser has a book called evolution of communication he actually has a chapter on language at the beginning and a chapter and which at the end but that's kind of to sell the book it's really about you know bats and location and stuff like that and every organism from bacteria on up has means of communication humans do to lotsa so you can study you can't find out anything much because evolutions are hard topic but you can at least look at I think the reason why people like to look at it in the case of evolutions communication you can kind of imagine some kind of continuity from bacteria to humans and that makes if you kind of like this idea of you know evolution taking place in tiny pieces doesn't happen that way but if you like that idea it can make you feel good but that's because every organism has communication systems but only one organism has language and like Tattersall said and if you're interested in language it just doesn't tell anything except gosh or like everything else this is one of the means of communication I mean I'm kind of exaggerating there are some things you can get so for example there's interesting work on this a technical question we're called neo greicy and conventions it's quite interesting work on that you know what are the proper conventions for discourse and there is quite interesting work that goes on on that that's not what people are studying when they do massive data of analysis and try to find something out about communication but it bears on communication thank you starting many years ago and often on your authority or with an appeal to your authority I was told that competence not performance was the key problem of linguistics now listening to your talk today I hear you putting far more emphasis on performance I wonder if you could clarify well I didn't performance in confidence yes if I did I wasn't aware of it I didn't intend to I mean this is all about this is all about competence I mean there is of course you know we use that this is considered to be a controversial distinction I haven't the slightest idea why it's a very simple conceptual distinction there's a distinction between what you know and what you do okay nothing controversial about that like you know arithmetic we could study what is it that you know you know maybe what you know in your head somewhere is Panos axioms or something like that who knows we can study what you do with it isn't it like what happens if you're given you know you're given a you're asked to multiply two big numbers how do you do it we could ask about that the second question is not a very interesting one it because too many factors enter into it like say short-term memory too many things to separate out an interesting question is what's your knowledge of arithmetic and we could ask the same question of any other cognitive system in the case of language that's competence performance but everything I've talked about today at least designer standard is basically about what you know about competence how's language design there are other questions about how you use the language and as I just mentioned here what you use is a very small part of what you know okay look for example takes imbedded sentences you can embed sentences indefinitely if you have enough memory time space and so on you want indefinitely easy to show that so it's infinite but what you use is very narrow because you just have too little memory I heard I heard you referring a great deal during your talk today about what at least I interpreted as performance but I may have been I may have misinterpreted well well my question I'm afraid that my question might sound a little bit too elementary but I like to see why well the minimalist framework assumed that a sentence is constructed from bottom to top right through the cyclic merge but at the same time we pronounce and understand the sentence from top to bottom because the minimalist framework assumed that all human language so more or less or right branching nature right so I'd like to see why humor angry shows such a mirror pattern you know how it's not I mean how we process is very much debated for example there are analysis or synthesis models that argue that you process top-down we of course process linearly because that's why we hear it you're stuck with that you're just as you're forced to produce through the articulatory apparatus whatever is going on in your head like for example it their peers it's trying to argue I think that just don't have linear order in the competent system but of course you perform linearly you have to and you're perceived linearly so how you use the system is determined by a lot of cognitive systems that we have a memory the structure of the auditory system the sensory motor system that these are the use of the internal knowledge got to be processed through those things it just as a user a rhythm ticket and if there are and there are incompatibilities you're correct in fact I tried to emphasize one of the most striking ones one of the most striking cases of incompatibility No is the sharp conflict between computational efficiency and communicative efficiency language is just badly designed for communication but well designed it to be efficient it seems I mean that's a there's a kind of a phrase that sometimes used for this it drives people crazy language is beautiful but unusable it's a kind of true you know even if people don't like it and there's a reason for it I think probably the reason that I mentioned let's do the weight evolved and the kind of creature we are it's a system that developed in a way which satisfies apparently pretty narrow constraints on what a well-designed system can be but happens to be used by people who have problems with this because we have some sensory motor system memory system which doesn't work for it so you struggle through and if there are incompatibilities like top-down and bottom-up that's an element thank you whatever you say we're going to wrap it up thank everybody for coming don't ask don't ask you to sing okay could you hear what he's saying so wait reading a book you kind of process thoughts in your head that allow you to see what's happening in the book and I was wondering everyone has a different perspective of the characters or the scene everyone has a slightly different perspective of it and I was wondering why you you and a lot of other people with that it's an interesting question and there's some like in most of the sciences there are little pieces of it you can study one interesting case maybe some of you know something about this is autism it turns out that with autistic children commonly they may be expanding no cases where an autistic kid will be mesmerised by children's cartoons you know watch them over and over again same cartoon and not understand anything literally not understand because the kid can't understand why the characters are acting the way they are like why is this character running away I'm not afraid you know why is that character afraid the ability to it's called theory of mind you know nobody knows what that means but the ability to a gain the perspective of someone else which in normal children is around three or four words when kids start showing signs of theory of mind so cool but the autistic children often don't have it they cats why they some people who are autistic sometimes seem to be oblivious to the way you think to feel you know they don't know when you want to talk to them or when you don't want to talk to them they can't interpret your experiences as being different from theirs and you see it in things like what you're describing like inability to watch cartoon or read a book because when we do it normal people do it you're you're just imposing a lot of rich knowledge and structure on the little bits and pieces that you're seeing the bits and pieces that you're seeing are kind of hints you know what you read is a kind of a hint and you add a kind of a rich interpretation and array of knowledge to it that's why you can read the same book over and over again to get more a richer experience each time or a movie or anything else and the extreme example is what I quoted from absent on poetry the idea is to make you impose quite a lot so it's an interesting question and this you know AB it's only barely understood bits and pieces or research but you'll find out you

Channel: Linguistic Society of America

Views: 56,248

Rating: 4.8674846 out of 5

Keywords: Language (Quotation Subject), Noam Chomsky (Author), Linguistics (Film Genre)

Id: -72JNZZBoVw

Length: 82min 45sec (4965 seconds)

Published: Fri Sep 06 2013