Noam Chomsky: Language and Other Cognitive Processes

[Applause] uh first question always is can you hear me in the back yeah okay um actually a week ago I thought Mother Nature was going to continue with her Antics and I'd have to come next spring but we made it the uh the title has a couple of presuppositions which I want to say something about uh one it's talking About other cognitive processes so it's assuming that uh the mind like the metal systems mind brain is like everything else in the body it's a modular system with subcomponents dedicated subcomponents that have their own particular properties

their own properties of growth and development and of functioning and that integrate with each other in the life of the organism and the modular approach to the Mind assumes that mental systems uh are the same I think that's pretty uncontroversial the other presupposition is controversial it's that language is one of those systems when I talk about language and other cognitive systems we assuming that language exists as a separate system not just some kind of accidental connection of interaction of other systems uh that's highly controversial in fact it's a that Language exists in in the sense

that can be an independent object of serious study uh is distinctly a minority position and in fact has been for uh a century uh the uh which is kind of strange when you think about it because on the surface it seems entirely obvious if you consider a a newborn in infant let's say uh the infant is faced with the famous blooming buzzing confusion of William James and somehow and how incidentally is not known uh some in Fact barely studied because it hasn't been recognized to be a problem though it is uh the infant somehow picks

out of this uh comp complex collection of unorganized data uh some sub parts which are language related that's not it's not easy to figure out how that happens as I say it's barely been studied just beginning to be studied now so it picks out the infant picks this out of the environment reflexively of course in fact we now Know that that's going on even in uterine to a certain extent and uh then continues again pretty much reflexively to U pick up the uh capacities that you and I are now using uh and does so with

very little evidence not only is the evidence scattered among other data but that for M much of what anyone knows there's almost no evidence at all that goes from the meaning of the simplest words when you look at them closely uh up to Complex uh structures their uh and and interpretation I'll give a couple of examples but it's a pretty clear furthermore it's a a humanly unique capacity so if an infant has a pet chimpanzee or song bird or kitten or whatever uh the animal can't doesn't even make the first step can't pick out of

the environment language related data that's not because of sensory motor defects or deficiencies or differences even a chimpanzee auditory system is Pretty much the same as a human auditory system of so much the same that it's recently been discovered it even picks out the distinctive features the kinds of sound features that play a role in language chimpanzee auditory system is somehow constructed so it specifically identifies those and the uh motor system really doesn't matter I mean the people learn language perfectly well children learn language actually normally if they're exposed only to sign uh not to

Sound and if they use only sound sign the course of development is virtually identical in quite remarkable ways uh and in fact the Lang the what's sometimes called the externalization of language the way it goes through the sensory motor system it seems to be modality independent uh and of course say Apes have you the same visual system essentially the same capacity for motor action even more developed in some ways than humans so that can't be the problem It's got to be something about the internal uh computational system that's uniquely human but despite all these facts

and there are glaringly obvious nevertheless it's there's kind of a Dogma that language just can't be a separate system uh it's got to be just a combination of other things common maybe to primates uh in fact it's if if you take a look across in comparative terms across other organisms about the only known animal That even has any of the rudimentary properties of human language are certain kinds of song birds and they are you know billions of years away in terms of evolutionary history so whatever similarities they are has to be convergence not a common

ancestry so so there's something very special about language uh also it's it's pretty recent in the human line there's really no evidence for its existence Beyond maybe 100,000 years ago which is you know Flick of an eye in evolutionary terms and many millions of years after the separation from the nearest species uh well despite all of this which again is glaringly obvious uh it has there is a kind of a Doctrine it uh pretty that it's if you took votes and among cognitive scientists today most almost all would agree with it uh that it can't

be true I think it's a interesting case of denial of the obvious for which maybe Psychologists might want to have something to say but uh you can easily determine it from the literature uh also it goes way back so if we go back to the 1950s when the kind of work that I'll be talking about what's sometimes called the uh generative Enterprise when it was sort of barely beginning actually a couple of graduate students at Harvard who were didn't believe any of the uh they were highly skeptical of the standard Behavioral Science concepts of time

at that point the same view was just rampant so the leading the leading philosopher most influential philosopher who uh had anything to say about these things was a very important philosopher at Harvard WV Quin and his conception which was highly influential and not just in the Cambridge area but everywhere else as well was that uh language anguage is just every language is just a fabric of Sentences uh associated with one another and with stimuli by the mechanism of conditioned response this was all within the framework of the radical behaviorism that was prevalent at the time

it's a view that's almost incoherent I mean how can you even tell whether some of the behaviors are sentences unless there's an independent concept of language but despite the coherence and the complete lack of evidence in fact masses of contrary evidence this Was a widely believed accepted view outside in the fields around Linguistics there was no cognitive science at the time was just barely beginning uh in linguistics itself a rather similar view is held so it was a a common standard view you can kind of read it in the major literature of the time was

what was called the bo iian view uh named after the famous uh Anthropologist linguist France boas whether he believed it or not is Another story but it was called The boasian View namely that languages can differ from one another in arbitrary ways and when you approach a new language to investigate you should do it without any preconceptions about what a language can be because they can all be entirely different actually that's the way I was taught in the late 1940s early 50s when I was a student the uh uh and of course that would mean

there is really no such thing as language it's Sort of anything that comes along uh the uh uh later on there were in in the philosophical tradition more modern philosophy another standard view that developed while these pretty much remain was not by not universally anymore by the 60s and the 70s it's beginning to change but uh a view that entered was and sort of was developed also tracing back to coin and in fact with deep roots in history of uh psychology uh is that uh any mental proc anything you can't Have a mental process at

all unless it's accessible to Consciousness may not be conscious but at least has to be accessible to Consciousness that's a very widely held view in fact if you look over the history of psychology it's pretty hard to find anyone who questioned it even Freud in fact talked about the unconscious but if you read Freud carefully he appears to have assumed that uh he's not a model of clarity but he seems to have assumed That whatever is unconscious can in principle be accessed by Consciousness in fact that's what underlies the practice of psychoanalysis trying to bring

things to Consciousness it's very hard you might check to find anyone who believed that the uh process the mental processes are processes of thinking planning interpreting and so on that they can in fact be inaccessible to Consciousness although I think again the evidence is Overwhelming that that's what they are almost entirely and in the case of language almost completely which is why you can't investigate a language by introspection uh you can look at the data but it's your of your own language but it's as faren to you as it is to anyone else what's inside

is well the uh the approaches to language at the time were what are called procedural you there were certain procedures of data analysis you applied them to a corpus of Material and you could rearrange the material and put into organized form which again that's what theoretical Linguistics was in fact the main theoretical book was called Methods in theoretic in structural Linguistics methods because that's all there is European Linguistics was approximately the same and that again has the same kind of framework of assumptions uh it it follows from this and was believed that there can't be

any Real problems about language can't be any puzzles because or doing is taking data and organizing it and unless you make a mistake or you have the wrong procedures it can't really be anything problematic uh it's uh uh there is a modern variant of this which as I mentioned is overwhelmingly dominates the cognitive Sciences I'll return to that well as I say it's kind of strange because of the uh not just for exotic reasons but because of the nature of What's obvious before our eyes but there's nothing novel about that that's the way the Sciences

have developed uh so for example for U one of the a great turning point in the Sciences was around the 17th century when scientists for the first time really allowed themselves to be puzzled about what looked like very simple things so for thousands of years the greatest scientists had a simple answer to some obvious fact obvious fact is That if say I pick up a cup of boiling water and hold my hand over it uh if I take my hand away the steam Rises and if I let go of the cup the cup Falls so

why does the cup fall and the steam rise well there's an answer given by Aristotle they're seeking their natural place and for thousands of years that was considered a satisfactory answer modern science actually begins when Galileo and later others uh agreed were were able to convince themselves There's something puzzling about that why does that happen and as soon as those questions began to be asked and investigated it was very quickly discovered that uh all of our intuitions about how these things work are just false intuition is about rate of fall and almost any other simple

thing and it's at that point that when it came to be discovered that what seemed to be obvious was in fact not at all Understood it's at that moment that modern science starts and that often has yet to happen in the human Sciences the things seem kind of obvious to us so we don't really question them but if you can bring yourself to be puzzled about them you find that we really don't understand anything uh well those are the two presuppositions I'm going to assume accept both namely that uh the mind is modular like everything

else in the biological world every other thing In or organisms and that language exists as a separate module uh this uh uh it shouldn't really be surprising that uh language and the mind should and the human intelligence should follow the biological Norm as I mentioned language has itself has very special properties I'll mention some of them and uh uh the things that should have seemed puzzling are very puzzling and still are when you look closely and uh it's a a very recent Development if you look at the evolutionary record uh uh of course we don't

have tape recordings but there's a lot of archaeological evidence and the archaeological evidence for the existence of language sort of implicates the existence of language is very modern uh maybe within 100,000 years or so and unique to humans it's not found in the other homed branches neander THS for example who lived up till maybe 30,000 years ago uh in fact we still having the Underfold change but uh it seems to be very specifically human millions of years after uh this the break from other organisms uh there is a huge literature by now just the last

20 years it's burent on what's called evolution of language which is itself kind of curious uh evolution of language is a very hard topic to study there is no direct evidence there's no comparative evidence uh nobody knows where to look for Neurological evidence uh nevertheless there there libraries full of books on it especially the last 20 years if you take much simpler topics there's almost no literature on them so take say the communication system of bees is quite complex uh some scattered papers uh practically nothing on it the reason is it's just understood to be

too hard it's too hard to study I mean Sciences can only study what's kind of right at the borders of understanding uh for bees a Study of uh evolution of the communication system of bees which is quite complex you all know about the waggle dance and that sort of thing uh is far more easy to study I mean there's about 500 species of bees they have plenty of comparative evidence they have different communication systems and some appear to have no communication system uh they get along about as as well as the others which raises some

questions about what the function of those Complicated systems is but anyway there's plenty of comparative evidence the brain is Tiny it's the size of a grass seed maybe 100,000 neurons uh minuscule as compared to the human brain uh very short gestation period couple of days it can breede and so on and so forth you can do any experiment you like you want to take them apart that's allowed you don't need consent forms or ethical issues or anything so it's a perfect organism to study as and humans Are impossible to study in all of those dimensions

and nevertheless there's almost nothing about it because of the recognition among biologists it is just too hard uh on the other hand the kind of tcid assumption that there can't really be much to language uh makes it a possible topic to study uh if you don't know anything about language and incidentally if you don't if you have quite uh confused Notions about Evolution there is a kind of a pop Biology which is very widespread that assumes that uh uh evolutionary change takes place in small steps and adds up and finally you get complex things happening

it was believed at one time not very long ago in fact uh nobody believes it anymore by now there's overwhelming evidence for uh what's called saltation you know sudden changes uh and it's even kind of understood why this has been understood in the biological sciences ever since at Least the 60s and the 70s when uh after the discovery of regulatory mechanisms cells that govern the uh action of the genes that govern the action of other genes if you change the regulatory circuits a little bit you get huge phenomenal differences uh and there's by now many

examples of very small changes genetic changes that lead to substantial uh differences in uh what the organism is uh famous discussion back in the 1970s by hob is One of the discoverers of these mechanisms Nobel laurate who argued that if you you could change his image was you could change an elephant to a fly by just uh changing the timing and organization of few regulatory circuits and that kind that that you can't prove yet but uh similar smaller results have definitely been shown so the idea that everything has to happen by small steps is out

the window I might say that back in the 1950s and 60s in biology too it was generally assumed that uh organisms can vary in arbitrary ways so the next organism you look at you have to approach it quite differently from others uh without preconceptions that is totally out the window uh so much out the window that by now there are even proposals which are taken seriously one by a molecular biologist at Boston University that U there may be a universal genome that is that all the Metazoa complex organisms that uh come from the Cambrian explosion

500 million years ago that they're all fundamentally the same they all have a basically the same genome with slight variations and that doesn't look exotic anymore it may not may be true maybe not but it's not an exotic proposal given what's known now about conservation of uh genes and genetic structures going all the way to bacteria up to humans and and uh uh Sim and deep homologies you know very deep Similarities in the way organisms are put together uh it's I think the same is true of language but it's much more controversial in this case

well um give you some illustrations from today I go back a couple of months there was a review article in the journal science Main joural scientific weekly American American Association for advancement of Science by a u well-known neuro cognitive neuroscientist at the MOX pla Institute in the Netherlands uh he was reviewing several books on evolution of language uh the uh and he kind of dismissed most of them because they were tainted by a a false assumption namely the assumption that language is that language actually exists that is that there are rule systems specific to language

which uh uh determine the character of sound systems other externalization systems uh determine the structures and meanings of Uh Expressions how they're interpreted and that can't be uh it's violates the Dogma uh and in fact the uh in order to sort of establish the point the editors added a photograph in the uh in in the along with the journal article which is a photograph of three infants uh properly multi-racial it's got to be politically correct uh so three infants who seem to be more or less paying attention to one another and the sub the title

says uh communication without syn Text without rules so that shows that you can have communication without rules uh they could have had a picture of three bacteria which would have made the same point they also communicate without syntax uh but underlying this is a fundamental Dogma which is almost never questioned though it's almost certainly false and that is that language is just a system of communication there's every reason to believe that that that's not true uh in Fact again obvious reasons if you just introspect uh almost all of your use of language like 99% of

it is internal you can't go for a minute without talking to yourself it takes a tremendous Act of will to stop talking to yourself well that's obviously not communication that's thinking uh and that's almost all the use of language in fact goes on all night unfortun Ely but uh uh and even interaction like you know interaction among people parts of it are Communication but the large Parts aren't they're just establishing social relations uh whatever nevertheless this Dogma is widely held and that leads to the uh assumption that you can understand something about language if you

look at communication systems of other languages that's what underlies the all of the work on evolution of language you take a look at this work you find it's not about evolution of language at all it's speculations Speculations of course about the evolution of communication to quite a different topic it's not language uh and in fact the language is used for communication but so is everything else we do style of dress whatever and the large parts of language aren't use of language is only peripherally for communication but but if you think of language as just being

a system of communication then you can look at animal communication systems and you can See if you can draw some analogies or whatever actually you can't very much but at least it's it's not ridiculous if you look at the properties of language is just instantly ridiculous even the ones that I simple ones that I mentioned uh well the uh uh with regard to uh uh with regard to um communic there Al the title of the article incidentally uh is also revealing the title of the article is without social context question mark and what it's the

point is To criticize the idea that you can study language without paying attention to the social context of its use well in fact if you're studying communication that's true uh it wouldn't make much sense to study communication without looking at the circumstances of its use uh it's close to topology on the other hand it's completely false of the study of the mechanisms that enter into communication or that enter into other aspects of life and in other areas of The cognitive sciences and biology that's just taken for granted so for example if you study the structure

of the digestive system you can do that in fact you do do that without asking what happens when you have a Big Mac or something in fact that you dismiss you try to abstract away from that uh if you're studying uh in the in mental psychological systems there's very good work on say object recognition by infant Elizabeth spelly Renee ban others they look at the try to discover the mechanisms by experiment uh it's without social context they're not interested in they don't look at it when two kids are playing maybe you can but that's peripheral

you try to abstract away from that and Carry Out experimental studies in a a refined situation in which there's no social context that's the way you study it automatically like anything else in the Sciences uh take one of the uh major uh Discoveries of uh study of perception um one of the most interesting principles that was discovered was uh Shimon alman's uh what's called rigidity principle it's a pretty surprising fact it turns out that if a person is presented with tcop images you know just dots in a screen a few dots maybe three or four

dots on a screen and you have several presentations of it what you see what you perceive is a rigid object in motion that's the structure that you Impose on the sequence of a few dots on a screen it's the rigidity principle which is a pretty remarkable Discovery but has absolutely nothing to do with social context again it's it's not even real objects it's looking at cystoscopic images like most of the study of vision uh so and in fact throughout the biological sciences just as in chemistry or physics all of this is taken for granted uh

but in the study of language and higher human mental processes this Considered something to be something wrong with this you just can't approach it in the manner of the Sciences uh the uh so take just take a concrete examples of this things in the way so I won't write in the Blackboard but it's simple enough so you can keep it in mind so take an actual sentence say of English uh take the sentence uh they Wonder uh if they wonder if the mechanics fix the cars okay and suppose You want to ask a question about

how many cars there were you can say how many cars did they wonder if the mechanics fixed suppose you want to ask qu same question about the mechanics well then it comes out uh how many mechanics do you wonder if fixed the cars well if plainly something wrong with that you can't say that uh has a perfectly good meaning it's fine thought but they just can't express it in language and that Generalizes very widely there's principles that underly it that extend to all sorts of other constructions but did you uh do you have to study

that in a social context or in a uh you know condition of communication could it possibly be learned I mean did you have evidence that you're not allowed to say that or you can't say it uh of course not I mean this is just something like the rigidity principle it's some property of the System which if you're willing to reach the stage of say Galileo if you want to enter into modern science you should be puzzled about uh and it's a typical example Myriad examples of that kind and any one of them illustrates the fact

that if you want to understand anything about the properties of language you're going to have to study it the way the Sciences are pursued even U the cognitive Sciences in other domains but that's contrary to the uh Dogmas well uh what's developed uh out of this is uh in modern cognitive science is a very strange concept of scientific inquiry it's U it's not true of say the study of VIs Vision like spelly by Aon Omen and others uh but it is true of the uh comp especially the computational cognitive Sciences which have the Merit that

the proposals are clear enough so you can actually investigate them they're not just hand Waving about you know everything happens like other primates uh they actually have models that you can investigate which is good and what you can show is they fail 100% totally and it doesn't make any difference because they succeed in terms of a new conception of science which dominates the computational cognitive Sciences uh the idea that science is a matter of uh developing models that M that more or less approximate Data usually basian methods of statistical analysis and basy methods have the

nice property that what whatever the data is you can find the basic analysis it will fit it you just have to fix the pick the right priors and so on and so forth not exactly fitted but kind of Come Close uh This Is Never Done In The Sciences so for example if someone is studying say B communication you know say the waggle dance of some species of bees they don't Proceed by taking videotapes massive videotapes of bees swarming and then do a statistical analysis of them you know basian analysis and from which you could get

a pretty good prediction of what's going to happen the next time be swarm but that's of absolutely no interest to be scientists and quite rightly if you apply for an NSF Grant to study it they'll laugh at you on the other hand if you apply for NSF Grant to do the same thing with the Corpus a collection Of spoken language you get get it right off that's considered very sophisticated or say physics uh if if trying to study Physics you don't take videotapes of what's happening outside the window you know leaves flowing around and so

on and do an analysis of them and get you know pretty it's massive collection you get a pretty good prediction of what's going to happen tomorrow when you look out the window a prediction that's way better than any that a physicist can give in Fact they can't give you any prediction at all and they don't care about uh but the Sciences just aren't done like that uh the only thing that is done like that is uh particular branches of the computational and other cognitive sciences that deal with higher mental faculties language in particular and there's

a Criterion of success the Criterion of success is you get a little better each time you have more data you know more complicated basian analysis More sophisticated methods and you get get a little closer to the data as I say in any other domain this would simply be ridiculed but it is the Criterion of success in the computational cognitive Sciences novel approach uh you can find it in all the major journals all the time and uh uh but only for language not even for something as close to language as say a numerical calculation so for

example suppose you want to study the rather interesting fact in fact very Surprising and puzzling fact that all humans have a capacity for a numerical computation they understand the number system that's incidentally a a a fact that did puzzle uh the scientists who created the modern theory of evolution Darwin and Wallace they were kind of struck by the fact that all humans that didn't experiment with it they just recognized it was true that all humans have uh comprehend the number system which is very surprising from their Point of view because it couldn't have been selected

uh it's barely been used in the whole history of humans a small group of humans mostly pretty modern and still small have you know made use of this system but for most of almost entire human history in fact most humans in their day it just they never never used it at all I they could handle small numbers but so can apes and they could handle what's called numerosity you know knowing that 100 things is more than 80 Things but not the particular characteristics of the digital uh number system you know the principles of multiplication or

addition and so on so where' it come from uh big puzzle for them if it wasn't selected how come it's there uh in fact Darwin and Wallace had a dispute over this co-founders of theory of evolution Wallace argued that there has to be some principle in evolution apart from natural selection uh to account for this mystery and Darwin didn't agree with that but he absolutely he had no alternative suggestion remained a mystery uh it has remained a mystery up to the present the likely answer which we can now begin to perceive is that if you

look at the actual mechanisms of language if there's any time I may get to them or at Le something about them but if you get if you take a look at them carefully turns out that uh uh if you simplify them radically they in fact yield the number System so it's very likely that uh the number system just kind of piggybacked off language if you have a language however that evolved uh you just get the number system as a free gift by simplifying the language actually reducing the Lexicon the collection of words to one element

then the basic computational principles do give you a model of the number system which may very well be where it comes from which would if true overcome the mystery well Um uh let's go on this I'll go back to Enfield again his useful collection of standard beliefs uh he argues that language doesn't exist because it is entirely grounded in a constellation of cognitive capacities that each taken separately has other functions as well so it exists in the sense that say today's weather exists I mean yeah it does exist but it's not a topic of scientific

study it's just a constellation of other factors that Operate independently and the argument is that that's true of language there isn't a particle of evidence for this but it's almost universally held in the cognitive Sciences not totally but very widely and in fact as soon as you look at the simplest thing you see it can't be true so for example infant acquisition of language uh or the one example that I mentioned U how is that a constellation how how can that how can What you know about that follow from the interaction of other cognitive processes

and if so how and if somebody makes this proposal in The Sciences at least they'd be expected to give some evidence for it look at some simple cases but not in the cognitive Sciences it's enough to wave your hands at it uh mention one well-known thesis in the study of acquisition of language and one of the main figures in the field is Michael Thomas C those of you in the Field are are familiar with this and his view is that language is just a structured inventory of linguistic constructions acquired by processes common to primates and

others which are kind of obscure so there's interrogative constructions and passive constructions and you just acquire them they could be any other way you know just kind of accidental this way and there's nothing more to say about them also notice that it would have to be finite obviously It's not finite language is unbounded so how does it become infinite well that's just more hand waving uh uh going back to Enfield talking about the evolution of the system he says there are well-developed gradualist evolutionary arguments to support the conclusion that there's no such thing as language

except as a complex of independent cognitive processes notice the notion gradualist that's a reflection of the Mythology about Evolution that's still maintained outside of biology so I'm in evolutionary biology everyone knows that's false but outside the field even as close as U Know cognitive Neuroscience it's still assumed that somehow everything has to be gradualist well known that that's not true but it's U kind of what you learn in high school and people believe it uh in this in the Sciences too he's particularly uh upset about what he calls the cist argument And it is cist

that the transition from finite to infinite was not gradualist so natural language is clearly infinite there's no bound on the number of Expressions that you can in principle produce and understand uh just as is no bound on the uh uh collection of triples of numbers XYZ where Z is the product of X and Y and in principle if you had enough time and energy and so on you could calculate it well language is the same so how do you get from finite to Infinite well the fact that that's saltation not gradualist is about as controversial

is the fact that 2 + 2 equal 4 there's no logically possible alternative it has to be exultation but the the fact that it is is incompatible with the gradualist conceptions now in a rational field when you have some conception and you have something that shows it's logically impossible you give up the conception but in a field that's dominated by Dogmas many of them ancient dogmas what you do is deny the log the the facts even if they're just logical truths and that's unfortunately pretty common well uh what do we actually know about evolution of

language not very much in fact but the little bit that is known uh is highly suggestive uh one thing we know which is kind of surprising if you're willing to be puzzled by Simple Things uh is that uh all humans have the same capacity for Language now all humans haven't been studied but a lot of linguistic groups have been studied and there are no known group differences there are individual differences but they're found in all groups so for example if you take a an infant from uh a tribe in say Papu New Guinea that hasn't

had contact with other humans for you know tens of thousands of years uh and you bring it to Boston it's child is raised here it'll be just like you same linguistic Capacities same cognitive of capacities go on to become a Quantum physicist and so on and conversely you take a infant from here put it in Pap guine same thing happen and in fact as far as it's known there are either no or very Marg at most very marginal differences among human groups in cognitive capacities Al alog together but in linguistics capacities they're so if they

exist they're so minute you can't find them by present methods so very marginal at most uh well Humans uh we all descend from fairly small group that uh left Africa maybe 50,000 years ago something like that and that what that means is for the past 50,000 years there's been zero Evolution none there's been change but change is not Evolution you know or the fact that we're different from hunter gatherers in our cultures and size and everything else that's uh those are signs of change but not Evolution if there is evolution it's very superficial Things like

say skin color and hair length things like that but not in fundamental human capacities as far as anyone can tell so that's 50,000 years with no Evolution at all which is striking that means that whatever for language whatever evolved was kind of rigid never changed again uh it you know of course languages differ but that's not Evolution that's just something else you know like the Norman Conquest uh changed English enormously it made it Kind of like French but that's not Evolution if languages change the way they do say by the effect of teenage jargon or

something which is a very big effect on language change it's not Evolution it's just changes take place within a fixed system and the system seems to be pretty rigid for 50 at least 50,000 years well if you go back about 50,000 years before that you just find no evidence that language even Existed uh the evidence and it's commonly accepted by paleoanthropologists that sometime in that narrow window there was something that sometimes Jared diamond called a Great Leap Forward there's a short period in which there's a sudden explosion of complex artifacts complex social structures uh uh

symbolic Behavior representational art Tak calculation of astronomical Events like you know phases of the moon and so on uh all of this happened at least in the archaeological record very brief period maybe roughly 75,000 years ago you can double the numbers if you like or triple them and nothing much changes it's because these numbers are all so small that from an evolutionary point of view they don't matter uh and again that's millions of years after the separation from uh other organ other surviving organisms uh the uh well that Seems pretty clear and it's highly suggestive

about the nature of language uh one critical property of language crucial property uh in fact the core property of language is that each language consists of at its root of some mechanism it's called a generative process a mechanism which constructs uh an infinite array of structured Expressions which have interpretations at uh in two systems two interfaces they're called sensory motor System you know for making noises or signs or whatever and thought systems for thinking planning interpretation so on each of these structured Expressions has to be mapped into those two systems I that and it's unbounded

as I said there's and it's also digital like like the numbers like there's a five-word sentence six-word sentence there's no five and a half word sentence and it goes on indefinitely as long as you have patience time and so on and virtually Everything that's produced is actually novel in your own experience maybe in the history of the language because the sca the size of these uh sets is so astronomical that you just keep producing new things well that's the most Elementary property of language now start around by around 1950 a generative processes like that were

understood from a mathematical point of view they really hadn't been in the past but you know algorithms they're Called or you know the kind of program that you can write for your laptop let's say these these things began to be understood and understood pretty well so it was possible to ask what kind of a generative process could have these properties well uh H how could such a uh notice that it it came apparent given the limited evolutionary evidence it must have happened pretty suddenly that's a very brief window so presumably what happened it's hard to

think of an Alternative is that in some small group these are Hunter gather societies remember like a 100 people or something in some small group uh wandering around Africa somewhere some individual uh underwent a slight mutation which led to a small rewiring of the brain which provided a generative procedure now mutations take place in individual uals not in groups so it would have been one person who was lucky Enough or maybe unlucky enough to undergo this mutation and that mutation provided a gener a generative procedure but remember it was one person hence there's no possibility

of communication uh if this generative procedure produced structures that were linked to pre-existing conceptual systems that that individual could think it could plan it could inter interpret and so on well that presumably yield some selectional Advantage uh a mutation can be transferred to offspring usually is partially at least uh and that means over time there might have been a enough people in this small group so that somebody would have gotten the idea that it's a it might be useful to externalize it to map this internal system into a uh something that can be perceived by

others maybe sound maybe sign you know maybe tactile whatever might be well that process of Externalization is a very tricky one the internal system that developed notice would have had no selectional pressures couldn't have it's just some something that happened internally so therefore it would have developed solely in terms of a natural law it would be something like a snowflake just takes on a perfect form because there's there's no external pressures on it so you'd expect an internal system to Develop which is kind of like a snowflake uh perfect in some sense and since these

are computational system systems what that would presumably mean is that it's it's computationally efficient it meets perfect conditions of computational complexity well a certain amount is known about computational complexity enough to set pretty strict conditions on what that might be like so what we would predict if if we're investigating language is that there's An internal system uh perfect in a sense that is as computationally efficient as a system of this kind can be uh which connects to the two interfaces one of them the interface the mapping to the thought thought systems again has no external

uh uh forces acting on it so that ought to be perfect too and also in variant uh wouldn't change because there's no pressures on the other hand the mapping To the sensory motor system is quite complex sensory motor system has been been around for hundreds of thousands of years pretty good evidence for that and it had nothing to do with the internal system so the finding the connection between the internal system and the data produced by the sensory motor system that's a complex cognitive problem and you can you can expect that it'll be solved in

many different ways and in fact it is that's the task fa faced by Every infant an infant gets some data from the sensory motor system has to figure out how to M map it to what may very well be a virtually invariant and maybe perfectly designed internal system of syntax and sematics and as far as we know about one of the pretty obvious you know overwhelmingly I me the facts about language are getting better and better established as research continues is that while the externalization systems vary enormously and are pretty Complex the internal system the

syntax and sematics they seem close to invariant uh which is to be expected because there's very little evidence for them they're very much like that one example that I gave you have almost no evidence for it so they must follow from rigid uh principles which are there because they're computationally efficient and may not have changed for let's say 75,000 years whatever they certainly haven't changed for 50,000 Years well when you study a language what do you study you study the external externalization you learn the sounds you learn the arbitrary word meaning Association not quite arbitrary

but somewhat uh you learn the irregular verbs you learn the order of words things like that those are all parts of externalization you don't learn the syntax and the sem semantics for one thing nobody can teach it do you because it's barely Understood uh for another thing you know it already it's part of your nature and the same is true of the child acquiring language that's roughly the these are the general facts that we observe uh they fit pretty well to this story this picture and U that looks like the direction in which uh linguistic

inquiry ought to go uh and it is internal to What's called the generative Enterprise but quite separate from uh beliefs and Doctrines in the cognitive Sciences uh especially the computational cognitive Sciences uh And I stress again it's a kind of a harsh judgment and I don't have time to talk about it that there are literally no results in the computational cognitive science with regard to language it's a huge literature but if you look closely the only results are in terms of this a novel concept of scientific inquiry that I mentioned and which has to do

with Matching data not known in any other field even in cognitive psychology well there's uh there's a lot to say about this but it's getting late so let me just make one final Point any computational system whatever it is has to have sort sort of atoms of computation you know minimal elements that are operated on by the computational principles say any system you program for your laptop uh any computational procedure so it has to be True true of language those are roughly wordlike first approximation not really so we can take word words first approximation to

be the minimal Atomic elements well where do they come from here's something else that one should be puzzled about uh how are words related to the external World a word like say cow or person or table or RI River or anyone you pick there's a Dogma about this as there is in most things connected with language the Dogma is What's called the referentialist Dogma a child learns the the word cow because it sees a cow and somebody says cow and an association is established between the cow and the word and that's the meaning of the

word again the simplest invest investigations show that this is totally false it's it's interesting it's pretty much true for it may be completely true for animal communication uh every animal communication system that's Known consists of atoms F finite number of them of course uh which like say which are tied to particular par particular mind independent events events that a physicist could describe so say a a monkey a vervent monkey may have say five calls uh one of them is we we call it a warning call it's connected to leaves moving in the trees verit see

Le leaves moving the trees it emits this coal and other verit run away uh or or could be I'm hungry you know some hormonal change which is identifiable and as far as far as is known that's what animal communication systems are like it's completely false for human language there are no sorry mind independent uh properties properties that can be identified by a physicist that correlate with the words and are constitute their meanings what they're referred to in fact the meanings of words really just give kind of Perspectives cognitive perspectives uh that you can employ in

uh discussing and referring to the world uh there's no time to give examples but uh it's literally true true of just about everything actually this was pretty well understood in the 17th and 18th century it's been interesting work on it it's mostly forgotten well that raises a very serious problem for one thing it's a problem of where these things come from how do they get into it must be that They're internal every child knows them you don't learn them you can't learn them very comp complex when you look at them they're invariant they're about if

they're different in various languages very minor differences uh and how did they evolve that's a total mystery in fact such such a deep mystery that it's very possible that there'll never be any insight into it well let me stop stop There [Music] well we would like to thank Professor chovsky very very much for this opportunity to come to speak for us this evening and we will open the floor for a few questions um so if you'd like to raise your hand um I can give you the microphone any questions over here thank you Professor uh

my name is Frank over here sorry right here how you doing uh you talked you were critical of uh what you call gradualism in evolution you were critical of what you called gradualism in evolution and how people think that uh it happened slowly over time and then you talked about how 50,000 years ago compared to now we've changed relatively little uh I was just curious how how that works they seem like contradictory ideas to me I was just curious what you meant by that by What by by by why nothing changed in 50,000 years well

you said we haven't really changed that much barely yeah but then you uh I feel like that is evidence of gradualism and I feel like you were critical of gradual against gradualism it goes against gradualism yeah if if uh you know there's just constant evolutionary change small changes why should it have stopped somewhere before 50,000 years ago on the other hand if what happened in human cognition is that Some rigid system were developed suddenly for reasons having to do with n natural law uh mutation of rewiring the brain computational complexity and so on but then

there's no reason to expect it to change I maybe it'll change someday with another mutation but but uh there's no reason to expect small changes so this doesn't prove that U evolution is not gradual is it's just consistent with the view that it isn't which we know anyway any other Question uh thank you Professor for your very generous talk um just had a um kind of a reflection I sort of um so I was interested in one comment you made about uh about kind of conscious and non-conscious processes and how the kind of the functional

aspect of your proposal is is more or less non-conscious whereas there's a kind of a conscious component which is like our everyday lived experience which is not kind of U going to be uh be able to Introspect upon this functional system so I was wondering um more on a philosophical not what how exactly we're supposed to think about phenomenal experience and kind of conscious experience in terms of its relationship to functional relationships like your um specifically your linguistic proposal but then also um other domain specific proposals like uh perception um there are those who don't

believe that perception is is you know Modu um modular or domain specific that it's embodied or something to that effect so I wonder if you had any propose um Reflections on this kind of um research now and uh how this relates to kind of our phenomenal lived experience well let's take visual perception uh which uh relatively uncontroversial and take the one example that I gave the rigidity principle okay when when you're looking At that sequence of tachistoscopic presentations you perceive a rigid object in MO in motion you can't prevent yourself from perceiving that that's what

you perceive period uh what are actually the stimuli that are actually reaching your your your eyes your retina are just a series of as few as three or four presentations on a toyos scope you know screen with lights uh of uh each of Which has several dots on it well you have no way of introspecting into that your phenomenal experience is a rigid object in motion but the stimuli that are hitting the retina are a very small number of uh pre presentations of a few dots you can't introspect into that uh the in fact it's

a that's why it's a discovery you know it's like the discovery of the chemical structure of you know coal Coal or something you Can't uh introspect into it uh so the phenomenal descriptions you know they're useful but they're useful as data they tell you very little about what's going on so it's kind of like everything else in the world where you have data but it doesn't tell you much about what's going on that requires that's why you have to take courses in physics and chemistry and so on you know because the data isn't transp isn't

transparent and it's hard for people to Deal with but the same is true of our conscious lives I we tend to take for granted kind of like you know it's just normal that that we can understand everything about our ourselves in fact we don't understand anything about ourselves it's just it's even harder than understanding uh how chemistry works because it's so much more complex and that's that's the crucial step that the human Sciences really have Not take taken yet for the most part uh that was taken in the Natural Sciences around the 17th century Ju

Just the willingness to be puzzled about things that look obvious like what could be more obvious than if I fact that if I let go of a cup it goes down instead of up well you know Galilea was willing to be puzzled about that that's why you take physics courses uh before that all you had to be told as things are going to their natural Place and uh it's a big U psychological step even in studying the uh external world even greater in studying ourselves because there we sort of feel that we know everything what's

conscious must be what there is and in the in philosophy modern philosophy it's it's kind of a principle it happened to be John surl who I was quoting but it's a widely accepted principle that nothing can be a mental process unless it's accessible to Consciousness and as I mentioned if you Look over the intellectual history it's very hard to find anyone who departed from this view okay we'll um take time for one more question cuz it's getting late and someone over here had raised a hand uh yes Professor um you mentioned the the problems behind

the uh referential Dogma um would you say that we are in a uh if we reject that are we in a situation that that Quin maybe puts us in with the sorry sorry are we in a Situation uh similar to quin's indeterminacy of translation would you would you say that's where we're left well Quin accepted the Dogma but that's not surprising because I everybody does today too uh yeah his uh his idea his doctrine of indeterminacy was in fact based on assuming this so if you take his framework that there's nothing but Association and conditioning

uh how do you how do you know to take his example Uh when you hear a word let's say rabbit this the case when a CH child hears the word rabbit how does it know that the word is referring to the uh that animal running around sorry and not to some part of the animal because if you see the animal running around its leg is also running around so how does the child know that it's not the leg and how does it not know in fact uh that the word doesn't refer to some disconnected object

like the leg of this One and of another one and so on that's a standard problem induction it's hume's problem of induction and as Hume understood but his successors don't seem to understand there's no way to solve that problem Hume himself contrary to what you may learn in a philosophy course was a rationalist not an empiricist he took for granted and me says that the only way you can solve the problem of induction is by what he calls an animal Instinct means there must be some internal structure in your mind that leads you to the

to a particular answer because there's no way to get it by induction and that I think is correct and I think the same could be true about the coin's Gava guy you know rabbit uh the child gets the right right answer because they're built to get the right answer and the right answer is not a uh in the case of say rabbit you can easily show That what a child understands to be a rabbit is not something physically identifiable uh kind of guessing your ages but say my grandchildren probably about your age when they were

kids U the story that they liked about a baby donkey named Sylvester and so somebody you know the baby donkey somehow is turned into a rock and for the rest of the story it's trying to conv convince its parents that It's not a rock it's their baby donkey and since children stories always end happily something happens and it it ends up and it's a baby donkey again and everybody's happy but the interesting fact is that every child understands that that thing that has all the physical properties of Iraq is in fact Sylvester uh and the

reason is and this was in fact noticed by John lock that uh people and in fact animals are individuated by properties Like Psy psychic continuity okay the same if you there's some psychic continuity no matter how you change physically I mean that's the standard fairy tale you know you know the evil witch turns the handsome prince into a frog and he has all the properties of a frog until till the beautiful princess comes and kisses the frog and he's a handsome witch handsome prince again well again every kid knows that it's always the Prince because

it has the property of Psy psychic continuity actually a lot of Science Fiction is based on this but it's even true of H the children's stories well that alone tells you right away that uh uh what individuates objects is not a a collection of physical properties but some complex mental structure that we impose on them like psychic continuity and uh hum didn't think about that lock in fact did but uh uh I should Say human recognized it was true that we don't identify words by physical properties but it runs across the whole collection of words

you can't find a a word so simple that it doesn't have those properties actually in the Natural Sciences what you try to do is to concoct you invent Concepts which do have those properties so when you invent the concept U say U you know electron or something you intend it to be physically identifiable you don't want it to have Its meaning change by what's in your head you know that's the whole point so it's a kind of a norm for the Sciences but it's just not true of human language and that goes back to quin's

example the solution to his problem is essent essentially hum's hum's solution to the problem of induction no solution except you're built to pick things to to structure the world in a certain way so that's the way to do it well thank you so much Professor for Giving us some insight into your area of expertise we'd like to present to you a small token of our appreciation um we'd love to have you back anytime you'd have us so ladies and gentlemen Professor Nom Chomsky thank you very much thank you thank you