Hello, and welcome to our expert interview for the third module. Today we're honored to have with us maybe a linguist who needs no introduction, Professor Noam Chomsky. Thanks for scheduling us in. >> Glad to talk to you. >> So, we'll just start right away if that's okay? >> Sure thing, of course, yeah. >> So first of all, and I'm sure you've answered this question many times before, but I think it's something that really interests our, the participants of this course, which is the reason that you got into linguistics in the first place. [LAUGH] >> The real reasons? >> Well. [LAUGH] >> I was a 16-year-old undergraduate at the University of Pennsylvania, and I had gone to the university with great expectations. The catalog looked really exciting. >> [LAUGH] >> High school had been very boring, but this looked like a way out, but practically every course I took was just more of boring high school material and I lost interest. There was one course that I was quite interested in, Arabic. There was a fine professor. Fine scholar, an Italian anti-Fascist immigrant, this was 1945. Sure got a way with the but I was ready to practically ready to drop out of college I was so bored with it. When I happened to meet Zellig Harris through separate connections. Political, common political interests. And he was a very impressive person, and I always enjoyed talking to him. And he suggested to me that I proof read his, a draft of his, the book of his that was coming out. It came out later as 'Methods in structural linguistics.' It was my introduction to the field. Then he suggested that I start taking his graduate courses, which I did, and then he suggested other graduate courses with other faculty members. Philosophy, mathematics, and I started doing that. Ended up with a very weird undergraduate education, which consisted of a scattering of graduate courses. [LAUGH] With no professional credentials, but I just kind of got more and more interested in the field. Had some, there was some background involved, my father was a Hebrew scholar. >> Yeah. >> His doctoral dissertation had been on Kimhi, the medieval Hebrew grammarian And when I was maybe 12 or 13, I'd read his doctoral dissertation, and some work on the history of semitic. I was interested in that, and all of that fed into these common interests. >> Yeah. So it all just came together at some point? >> Yeah. >> It just felt- >> True credentials and linguistics or any other field would never be admitted to any respectable department. >> [LAUGH]. >> [LAUGH] Okay, so well you've been studying linguistics all of your life I think and what is about language that you think is so fascinating that you've kept on studying it? >> Well, language has been recognized for centuries by philosophers, by scientists; Decartes, Darwin the whole list, >> Yeah. >> as the essential unique human endowment, the striking faculty that distinguishes humans from other organisms. That's the traditional view, and I think there's considerable evidence for it. So, and it is kind of the human essence in a way. It's the source of their creativity or originality or ability to think and what-not, plan in ways that are unique in the animal world. And when we, it also has some very strange properties when you look at it carefully. So language is a, what's called a digitally infinite system. It's infinite. There's no limit to the number of expressions that you can produce and understand. Which is already interesting because how do you gain an infinite capacity from finite data? >> Yep. >> And it's also digital, it's not continuous, like say the communication system of bees is a continuous system. >> Yeah. >> The language is a digital system; five word sentences, six word sentences, but no five and a half word sentences. And digital infinity is an interesting property. It's, by the mid 20th century it had become quite well understood. The theory of computation the mathematical theory of computation and so on. And it became possible, then, really for the first time to try to capture accurately what the nature of this property is. And it's there are many interesting, it's quite unusual in the biological world. You don't find such systems. And traditionally it had been pretty hard to study because the concept of digital infinity computability, was not really clearly understood theoretically until pretty much early mid 20th century. Within mathematics and then it was-. >> So-. >> Yeah-. >> So what I be correct it summarizing when I say that it's both the mathematical elements of the system and the way in which, by studying language, you can study the way that we are. >> Yeah, the essence of human beings >> Yes, the essence of humanity. >> Kind of creature we are. This is the kind we are. Not the whole story, but it's a. But that combination of accessibility to formal inquiry, careful formal inquiry, and human significance is, I've found, a kind of irresistible combination. >> Yeah, yeah. Well we, of course wholeheartedly agree. >> Yeah. [LAUGH] >> As you know, budding linguists. Yeah, another question we have is what is the most surprising thing that you know now, but that you didn't know when you started doing linguistics? >> Well, at the very beginning. First of all I should say in connection with what you said before, that this is not the way language was language was looked at, at that time. >> Yeah. >> In fact the general view of language was that languages can vary arbitrarily, and each one is different from every other one. You shouldn't approach any particular one with any preconceptions and that language simply... A good illustration of how language was conceived was actually Harris' book the 'Methods in structural linguistics. It was a collection of procedures, that a linguist can use in the field to organize the materials that he collects from an informant. That's what linguistics was. There was nothing much to say theoretically, and maybe some things about, you know, the structure of phonology, which kind of features there are. But beyond that, not very much. The, and that was because of the inability, at the time, it's not really a criticism, to capture what a generative procedure is. What is an infinite generative procedure? >> Mm-hm. >> But the most, it was pretty clear at the very beginning, say around 1950, that whatever the, that each language has a kind of a basic property which ought to be common to all languages. That is that a generative procedure which yields an infinite array of hierarchically structured expressions. >> Mm-hm. >> Each of which has a determinate fixed interpretation at two, at the interface with two other biological systems. The sensory motor system for externalization and the internal, conceptual, planning thought systems that sometimes go to conceptional, intentional system. So there's these two external conditions that the generative procedure must satisfy. And this system had to evolve somehow. There was a time when it wasn't there. Now it's there. It's apparently uniform among humans. There's no known differences in linguistic capacity in the human species. So for example if you take someone from an Amazonian tribe which hasn't had contact with other humans for 20,000 years and bring them up in say, Boston. They'll be like my grandchildren. And conversely, there's no known linguistic or other cognitive differences. >> Yeah. >> So somehow something arose, which is fixed. And it seems to be pretty recent, but we didn't know that at the time. But by now it's pretty clear, reasonably clear that it hasn't evolved for at least 50 or 60,000 years, since humans left Africa. And that if you go back, not long before that in evolutionary time, there's no evidence that existed that this was not anachronistic. What I'm saying is now reasonably well known, it was then not known. What was known was that it somehow evolved, and that it seems to be the same for all people. Now that raises a very serious question. It suggests that when you try to write a linguistic, a grammar, a grammar of a particular language, it looks extremely complex. >> [LAUGH] >> Each one looks very different from the last one you tried. And therefore, the mechanisms that you have to propose for this basic principle appear to be extremely complicated. If you look at early generative grammars, 1950s, 1960s, the theoretical framework allowed for very complex operations, interdirections of operations. And that seemed to be necessary for just descriptive adequacy, to try to capture the data. But it was obvious that it couldn't be correct. For one thing, there's no way in which they could have been learned. And for another reason, they couldn't, it couldn't have evolved, the system, this complex. So the right answer, somehow has to be that, there's an extremely simple system which somehow yields this diversity and complexity just for the interaction of extremely simple principles and general laws of nature, which are probably laws of computation. But that looked like an idle dream at the time. >> [LAUGH] >> The most interesting discovery is that it may not be an idle dream. >> Hm. >> There's no reason today, I think, to take seriously a thesis which can't prove, but I think we're kind of approaching, that language is an almost perfect system. Meaning a perfect system would be one that satisfies universal and linguistic conditions of computational efficiency. >> Hm. >> These are essentially laws of nature that uses the simplest possible generative operation, combinatorial operation. We know what that is. And then by the way, just the way it works yields the apparent diversity and complexity of languages through small modifications here and there. >> Hm. >> That sometimes builds a strong, minimalist thesis. And it would have looked crazy 20 years ago hopelessly impossible 50 years ago and now, more or less plausible. >> So, what you're saying is that actually, it's all surprisingly simple. Or it's surprisingly orderly, maybe? >> It's almost, it's a possible thesis, which maybe your generation will prove, is that language is close to a perfect system. >> [LAUGH] >> It's close to the, at its core, it's close to the optimal way of satisfying the interface conditions. There's more. By now, I think, this is, I'm not speaking of a consensus in the field by any means. This is my personal opinion, and a few other people. I think by now, there's mounting evidence that the relation to the interfaces is as measured. That is that the, in particular the relation to the sensory motor system to externalization is ancillary. It's a secondary property. The core system of synthetics and semantics, and thought and so on, is independent of the properties of externalization. So for, in particular, for example, it's independent of linear order of words. We have to speak. The words have to come out in linear order. >> Yeah. >> But I think that's just a property of the articulatory system. But it isn't a language, it's a filter through which language has to pass. And in fact if you look at other systems of externalization, like sign, we now know, it wasn't known 40 or 50 years ago, but its now known that sign is essentially the same as spoken language. >> Right. >> Right. >> It has remarkably similar properties but the externalization is different because it's a different modality. So you can use simultaneity in sign or spatial orientation in ways that you, say anaphora for a reference and so on, that you can't use in the linear, spoken language. And if you had other modalities it, different conditions. But if this is correct, and I think there's good reason to believe it is, then virtually all actual linguistic work is on a peripheral system. >> It's on a system of externalization. And that makes practical sense. And so for example, if somebody wants to learn, say Dutch, they don't have to learn, they can't learn the fundamental principles. Nobody knows them to teach them. What you learn is the pronunciation, the vocabulary, the facts about word order or things like that, which seem to be extremely superficial. >> Yeah. >> Very complicated, but extremely superficial. And don't, apparently don't feed don't yield consequences for the core syntax and semantics, which it's independent of. And so for example, say the relation between a verb and an object, a transitive verb and an object is the same semantically whether the verb precedes or follows the object, it doesn't care, you know. And that seems to generalize quite, quite widely to some interesting properties of language. >> So if this is correct, as I suspect it is, then the core principle, the basic principle that I mentioned at the beginning, generating an infinite array of hierarchic structures mapping to the interfaces should really be mapping to one interface - the thought >> Yeah, conceptual >> interface. Then there's another system, which is, has to do with externalization, putting in public what's going on internal to your mind. Which is probably mostly unconscious. Now there's another possibility, I think, which looks to me increasingly plausible is that much of the mental operations that are going on when you interpret, understand, create, produce expressions is not only unconscious, but beyond the level of consciousness. >> Yeah. Actually we were, this sort of ties in with another question we had. because you're currently also involved in bio-linguistics. And well, what is the focus of that field, we would like to know. And could you maybe explain something about it? About bio-linguistics? >> Well this again sets us back about 65 years. Around 1950 when this kind of work began. There were a few of, there was kind of a party line if you like. A broadly accepted >> [LAUGH] >> consensus in the field, it was, it's essentially behavioral science. You study behavior, organizational behavior. In linguistics you study the arrangement and organization, and the patterns of data that you pick up. And things, language acquired by training, by habit, by this evolution wasn't, maybe natural selection, or something, but, nothing to say. But there were a few of us who had a different view, actually three graduate students. One of them is in the adjacent office, Morris Halle, another is Eric Lenneberg, who went on to found Biology of Language. The three of us were graduate students at Harvard around early 50s. And we just disagreed sharply with the consensus view. A part of the disagreement was the recognition of what, seems to me a truism that is widely contested. And that is that language, your language, is an internal property of you, that is inside you, it's not in the world. It's inside you're brain represented somehow. Nowadays something's wrote in pie language and internal language. >> Yeah. >> If language is an internal property of a person. Then, it's a biological system. And you ask yourself what kind of a biological system is it? That's bio-linguistics, and at that time there wasn't a lot to say about it. But over the years its become a richer topic. Eric Lenneberg's book 'Foundations of biology of language', or something like that. Came out in 1967, and in many ways it's still one of the best. I think he has the best discussion of evolution of the language, that one of the best. Til this, but all of these topics were, kind of, thought about, but there wasn't much to say about them. So let's say evolution of language, these are topics that were discussed right away. You had to recognize that somehow the system evolved, and it has to be simple enough so that it could have evolved. It's a condition a kind of over-arching condition on construction of linguistic theories. And it showed, that every one of them's just got to be wrong. Because they're much too complex to have possibly evolved. That's actually the motivation for searching for something, like what I call the strong minimalist thesis. >> Yeah. >> The, it could have evolved, the question now is is this possible. But there wasn't much literature about it. There were lots of meetings, conferences, symposium courses, I taught joint courses with a laureate in Biology at MIT for years, but very little came out because there wasn't much to say. Linguistic theories were too complex for a sensible approach, either to acquisition or to evolvability. In the 1980s, 1990s that began to change, there were things you could say. One change that was significant was around 1980 when the, what's called the principles and parameters approach crystallized. That actually opened up lots of possibilities. The earlier approach to linguistic theory within this general framework was that linguistic theory provided a kind of a format for grammars. Each grammar of each particular language had to fit that format. And you picked the grammar on the basis of some evaluation measure given data. So the idea is a child has data, there's a fixed format in its head, which is genetically determined. And then it picks the simplest grammar in terms of the fixed technique of measurement, given the data. That does, in principle, yield an answer to how language can be acquired, but it doesn't work, because it requires astronomical calculations. So it's, it's totally unfeasible. If understood it's in principle. Possible, but can't be right. The, principles and parameters approach, which was, pretty much crystalized in the early 80's, offered a different way of looking at it. The principles are fixed, they're part of the genetic endowment, and we hope to make them simple enough. So that they could have evolved, that's the, what became called the minimalist program it's just the continuation of the effort since the early 50s, heading hopefully to the strong minimalist thesis or something like as the answer to be reached. The parameters are the options available to the trials to fix given data and there's quite interesting work on parameter setting and the choices and arrangements of parameters and so on. There is incidentally an interesting evolutionary problem on the side. Where'd the parameters come from? And, there are some interesting ideas about that. One possibility which is being explored is that there really aren't any parameters. It's just that the principles are Underspecified in certain respects. >> And that specification of what is underspecified gives the various options. >> If that turns out to be true, it'll solve the evolvability problem. But this is really hard work. To try to show in detail that the strikings, the superficial diversity and complexity that you see, is actually misleading. And that it's reducible to some fundamental, simple principles. And this is pretty familiar in the sciences. And I think any science you look at the data just looks hopelessly complex. In physics say that was true in the, when Galileo and his associates tried to find simple principles of motion, they were pretty much ridiculed because it's obvious that it's hopelessly complex. Just look at the leaves blowing in the wind and so on and so forth. >> Yeah. >> And it took a long time for the understanding to be reached that you can account for diverse and complex phenomena in terms of simple principles if you can discover what they are and how they function. >> The same in biology. Now, in fact, at the time, when it was generally assumed in linguistics that languages are arbitrarily different from one another and no limit on diversity, then pretty much the same is assumed about biological organisms. That the diversity and complexity of organisms was so great, that each one had to be studied on its own. Over the years, it's been found that that's not true. That there's striking uniformities conservation of fundamental forms limited possibilities for organisms and something. In fact, it's gotten so far that there's even a proposal which is taken seriously, though not accepted, that there's a universal genome, one genome for everything. Just minor variations of it. It's not an outlandish proposal by now, it's just like the strong minimalist thesis. >> Yeah. So, I'm afraid we're, I suppose that we're running out of time soon. And I'm sure that you have, many other obligations today. So, [LAUGH] just as a, to wrap it up, of course, there's people watching this video who are already interested in linguistics, they're taking this introductory online course. So, after this course they will know a little bit about linguistics. What would you recommend for them? Would they, should they read your, your oeuvre? >> [LAUGH]. >> Or is that maybe one step too far at this point? >> They should, everything I've been saying is a minority opinion. I'm not giving a consensus in the field, and that should be understood. >> Yeah. >> All my life, in fact I've been part of a small minority of linguists that go kind of against the stream. So that has to be taken into account. >> Yeah. >> It's the way it looks to me, it's not the way it looks to the field. But that's for you know, students have to figure that out for themselves. >> Yeah. >> [LAUGH]. >> There's a diversity of opinion which are the right ways to think about these things >> But that's-. >> But it's >> That's a good thing, I suppose. Right? >> Sorry, to interrupt. That's a good thing, I mean-. Differences of opinion lead to interesting new-. >> Exactly, that's what makes the field exciting. >> Yeah. >> And, and this one is a very exciting one, both because of the significance of the study of language, as we talked, it's kind of the core of human nature. >> Yeah. >> The apparent conflict, and I think it's apparent between the diversity and complexity of the data. And the recognition that there's got to be a core fundamental simple explanation for it. >> Because of the conditions of learnability- >> Vulnerability, and so on. And that, kind of, superficial contradiction, makes an extremely interesting field. >> Yeah. >> Good. Yeah. Thanks very much.
