Things No Amount of Learning Can Teach
Omni, 6:11, November 1983
QUESTION: Why do you believe that language behavior critically depends on the existence of a genetically preprogrammed language organ in the brain?
CHOMSKY: There's a lot of linguistic evidence to support this contention. But even in advance of detailed linguistic research, we should expect heredity to play a major role in language because there is really no other way to account for the fact that children learn to speak in the first place.
QUESTION: What do you mean?
CHOMSKY: Consider something that everybody agrees is due to heredity -- the fact that humans develop arms rather than wings. Why do we believe this? Well, since nothing in the fetal environments of the human or bird embryo can account for the differences between birds and men, we assume that heredity must be responsible. In fact, if someone came along and said that a bird embryo is somehow "trained" to grow wings, people would just laugh, even though embryologists lack anything like a detailed understanding of how genes regulate embryological development.
QUESTION: Is the role of heredity as important for language as it is for embryology?
CHOMSKY: I think so. You have to laugh at claims that heredity plays no significant role in language learning because exactly the same kind of genetic arguments hold for language learning as hold for embryological development. I'm very much interested in embryology but I've got just a layman's knowledge of it. I think that recent work, primarily in molecular biology, however, is seeking to discover the ways that genes regulate embryological development. The gene-control problem is conceptually similar to the problem of accounting for language growth. In fact, language development really ought to be called language growth because the language organ grows like any other body organ.
QUESTION: Is there a special place in the brain and a particular kind of neurological structure that comprises the language organ?
CHOMSKY: Little enough is known about cognitive systems and their neurological basis; so caution is necessary in making any direct claims. But it does seem that the representation and use of language involve specific neural structures, though their nature is not well understood.
QUESTION: But, clearly, environment plays some role in language development. What's the relationship between heredity and environment for human language?
CHOMSKY: The language organ interacts with early experience and matures into the grammar of the language that the child speaks. If a human being with this fixed endowment grows up in Philadelphia, as I did, his brain will encode knowledge of the Philadelphia dialect of English. If that brain had grown up in Tokyo, it would have encoded the Tokyo dialect of Japanese. The brain's different linguistic experience -- English versus Japanese -- would modify the language organ's structure.
Roughly the same thing goes on in animal experiments, showing that different kinds of early visual experience can modify the part of the brain that processes visual information. As you may know, cats, monkeys, and humans have hierarchically organized brain-cell networks connected to the retina in such a way that certain cells fire only when there is a horizontal line in the visual field; other hierarchies respond only to vertical lines. But early experience can apparently change the relative numbers of horizontal- and vertical-line detectors. MIT psychologists Richard Held and Alan Hein showed some time ago, for example, that a kitten raised in a cage with walls covered by bold, black vertical lines will display good sensitivity to vertical lines as an adult but poor horizontal-line sensitivity. Lack of stimulation apparently causes the horizontal-line detectors to atrophy.
An even closer analogy exists between language growth and the growth that appears in human beings after birth -- for example, the onset of puberty. If someone came along and said, "Kids are trained to undergo puberty because they see other people," once again everybody would laugh. Would we laugh because we know in great detail the gene mechanisms that determine puberty? As far as I can tell, no one knows much of anything about that. Yet we all assume that puberty is genetically determined.
QUESTION: Still, as your own example shows, environmental factors do play a major role in physiological growth.
CHOMSKY: And it goes without saying that the onset of puberty may well vary over quite a range depending on childhood diet and all kinds of other environmental influences. Nonetheless, everyone takes for granted that the fundamental processes controlling puberty are genetically programmed. This is probably true of death as well. You may be genetically programmed to die at roughly a certain point; it's a reasonable theory. Look, all through an organism's existence, from birth to death, it passes through a series of genetically programmed changes. Plainly, language growth is simply one of these predetermined changes. Language depends upon genetic endowment that's on par with the ones that specify the structure of our visual or circulatory systems, or determine that we have arms instead of wings.
QUESTION: What about the linguistic evidence? What have you learned from studying human languages to corroborate your biological viewpoint?
CHOMSKY: The best evidence involves those aspects of a language grammar that are so obvious, so intuitively self-evident to everyone, that they are quite rightly never mentioned in traditional grammars.
QUESTION: You mean that school grammars fill in the gaps left by heredity? They teach everything about French or Russian, for example, that can't be taken for granted by virtue of the fact that you're human?
CHOMSKY: That's right. It is precisely what seems self-evident that is most likely to be part of our hereditary baggage. Some of the oddities of English pronoun behavior illustrate what I mean. Take the sentence, "John believes he is intelligent." Okay, we all know that "he" can refer either to John or to someone else; so the sentence is ambiguous. It can mean either that John thinks he, John, is intelligent, or that someone else is intelligent. In contrast, consider the sentence, "John believes him to be intelligent." Here, the pronoun "him" can't refer to John; it can refer only to someone else.
Now, did anyone teach us this peculiarity about English pronouns when we were children? It would be hard to even imagine a training procedure that would convey such information to a person. Nevertheless, everybody knows it -- knows it without experience, without training, and at quite an early age. There are any number of other examples that show that we humans have explicit and highly articulate linguistic knowledge that simply has no basis in linguistic experience.
QUESTION: There's just no way that children can pick up this kind of information by listening to the grown-ups around them?
CHOMSKY: Precisely. But let me give you another example. English contains grammatical constructions that are called parasitic gaps. In these constructions, you can drop a pronoun and still understand the sentence in the same way as when the sentence contains a pronoun. Consider the sentence, "Which article did you file without reading it?" Notice that you can drop the pronoun "it" without changing meaning or grammaticality. You can say, "Which article did you file without reading?" But you can't say, "John was killed by a rock falling on," when you mean, "John was killed by a rock falling on him." This time, omitting the pronoun destroys both meaning and grammaticality.
Constructions of this type -- where you can or cannot drop the pronoun -- are very rare. In fact, they are so rare that it is quite likely that during the period a child masters his native language (the first five or six years of life), he never hears any of these constructions, or he hears them very sporadically. Nonetheless, every native speaker of English knows flawlessly when you can and can't drop pronouns in these kinds of sentences.
QUESTION: So we're faced with a mystery. How could anyone possibly learn enough about the English language to possess the rich and exotic grammatical knowledge that we all seem to possess by the time we are five or six years old?
CHOMSKY: There's an obvious answer to that: the knowledge is built in. You and I can learn English, as well as any other language, with all its richness because we are designed to learn languages based upon a common set of principles, which we may call universal grammar.
QUESTION: What is universal grammar?
CHOMSKY: It is the sum total of all the immutable principles that heredity builds into the language organ. These principles cover grammar, speech sounds, and meaning. Put differently, universal grammar is the inherited genetic endowment that makes it possible for us to speak and learn human languages.
QUESTION: Suppose that somewhere else in the universe intelligent life has evolved. Could we, with our specialized language organ, learn the aliens' language if we made contact with them?
CHOMSKY: Not if their language violated the principles of our universal grammar, which, given the myriad ways that languages can be organized, strikes me as highly unlikely.
QUESTION: Maybe we shouldn't call it "universal" then. But please explain what you mean.
CHOMSKY: The same structures that make it possible to learn a human language make it impossible for us to learn a language that violates the principles of universal grammar. If a Martian landed from outer space and spoke a language that violated universal grammar, we simply would not be able to learn that language the way that we learn a human language like English or Swahili. We should have to approach the alien's language slowly and laboriously -- the way that scientists study physics, where it takes generation after generation of labor to gain new understanding and to make significant progress. We're designed by nature for English, Chinese, and every other possible human language. But we're not designed to learn perfectly usable languages that violate universal grammar. These languages would simply not be within the range of our abilities.
QUESTION: How would you assess current research about universal grammar?
CHOMSKY: In the last three or four years, there's been a major conceptual change in the underlying theory. We now assume that universal grammar consists of a collection of preprogrammed subsystems that include, for example, one responsible for meaning, another responsible for stringing together phrases in a sentence, a third one that deals, among other things, with the kinds of relationships between nouns and pronouns that I discussed earlier. And there are a number of others.
These subsystems are not genetically preprogrammed down to the last detail. If they were, there would be only one human language. But heredity does set rather narrow limits on the possible ways that the rules governing each subsystem's function can vary. Languages like English and Italian, for example, differ in their choice of genetically permitted variations that exist as options in the universal grammar. You can think of these options as a kind of linguistic menu containing mutually exclusive grammatical possibilities.
For example, languages like Italian have chosen the "null subject" option from the universal grammar menu. In Italian, you can say "left" when you mean "He left" or "She left." English and French have passed up this option and chosen instead the rule that requires explicit mention of the subject.
QUESTION: What are some other grammatical options on the universal grammar menu?
CHOMSKY: In English, the most important element in every major grammatical category comes first in its phrase. In simple sentences, for example, we say "John hit Bill," not "John Bill hit." With adjectives, we say "proud of John," not "John of proud"; with nouns, we say "habit of drinking wine," not "drinking wine of habit"; and with prepositions, we say "to John," not "John to." Because heads of grammatical categories always come first English is what is called a head-initial language.
Japanese is a head-final language. In Japanese, you say "John Bill hit." And instead of prepositions, there are postpositions that follow nouns: "John to," rather than "to John." So here's another parameter the child's got to learn from experience: Is the language head-initial or head-final?
These grammatical parameters are interconnected. You can't pick them any more freely than, say, a wine fanatic who insists on white wine with fish and red wine with meat is free to choose any main dish once he's decided on his wine. But grammars are even more sensitive than this culinary example might suggest. A slight change in just one of the universal grammar's parameters can have enormous repercussions throughout the language. It can produce an entirely different language. Again, there's a close parallel to embryology, where a slight shift in the gene mechanisms regulating growth may be all that separates a fertilized egg from developing into a lion rather than a whale.
QUESTION: So what exactly would you say is the grammar of English?
CHOMSKY: The grammar of English is the collection of choices -- head-initial rather than head-final, and null subject forbidden, for example -- that define one of a limited number of genetically permitted selections from the universal grammar menu of grammatical options. And of course there are all the lexical facts. You just have to learn your language's vocabulary. The universal grammar doesn't tell you that "tree" means "tree" in English. But once you've learned the vocabulary items and fixed the grammatical parameters for English, the whole system is in place. And the general principles genetically programmed into the language organ just churn away to yield all the particular facts about English grammar.
QUESTION: It sounds as if your present goal is to reach the point where you can define every human language's grammar simply by specifying its choices from the universal grammar's menu of options.
CHOMSKY: That's the kind of work you would hope would soon be done: to take a theory of universal grammar, fix the parameters one way or another, and then deduce from these parameters the grammar of a real human language -- Japanese, Swahili, English, or whatnot. This goal is only on the horizon. But I think that it is within our conceptual grasp. Undoubtedly, the principles of universal grammar that we currently theorize are wrong. It would be a miracle if we were right this early along. But the principles are of the right type, and we can now begin to test our present system with complex examples to see what is wrong and to make changes that will improve our theory.
QUESTION: Judging from what you've said about language and heredity, it sounds as if you must be sympathetic to the aims of sociobiology. Is that a fair assumption?
CHOMSKY: Well, I think that in some respects the sociobiologists are on the right track. I think it's true that a good deal of our personal behavior, social behavior, reactions, and so on are the reflection of genetic programs, and I think that it's a worthwhile enterprise to discover what these programs are. But while I think the general idea behind sociobiology is right, I also think that sociobiologists should be extremely cautious about the specific conclusions they draw from their research. Unfortunately, they often draw conclusions that are remote from evidence or theory.
QUESTION: Many sociobiologists would dispute your note of caution. They claim that science has already gained enough information about the relationships between genes and behavior to permit some shrewd guesses about some of the ways heredity influences human social behavior. What do you say to these claims?
CHOMSKY: I'm very skeptical. I haven't really studied the newer research in enough detail to make any informed judgment. But as for the earlier work -- for example, E. O. Wilson's Sociobiology -- well, about 90% of the book was on nonprimates, and that looked interesting. There was a little bit on primates, which was more questionable. And there was a final chapter on humans that was completely empty. I don't think Wilson understood what he was talking about in that final chapter. There were real errors in what he did describe in any detail. I don't even understand why the chapter on humans was tacked on to the book. It didn't seem to belong.
QUESTION: What do you think about the claim made by Wilson and others that there's an innate incest taboo in human beings?
CHOMSKY: Sorting out what is and what is not genetically preprogrammed in human behavior is a very difficult task. As I said, I agree with the general approach of sociobiology. I think it's a reasonable approach. But it's important to be very cautious in making any claims about the role of heredity in human affairs -- especially claims that would have social consequences if they were true. Science is held in such awe in our culture that every scientist has a special responsibility to make clear to the lay audience where his expert knowledge actually yields scientifically verifiable results and where he is guessing, indulging in sheer speculation, or expressing his own personal hopes about the success of his research. This is an important task because the lay audience is in no position to make these distinctions.
QUESTION: Moving to another controversial area in the behavioral sciences, how do you think your views differ from B. F. Skinner's behaviorist theory of language, learning, and mind?
CHOMSKY: Skinner used to take a relatively extreme position. At one point, he held that, apart from the most rudimentary functions, essentially nothing of importance was genetically programmed in the human brain. Skinner agreed that humans were genetically programmed to see and hear, but that's about all. Accordingly, he argued that all human behavior was simply a reflection of training and experience. This view can't possibly be correct. And, in fact, Skinner's approach has led absolutely nowhere in this area. It has yielded no theoretical knowledge, nontrivial principles as far as I am aware -- thus far, at any rate.
QUESTION: Why is that?
CHOMSKY: Because Skinnerian behaviorism is off the wall. It's as hopeless a project as trying to explain that the onset of puberty results from social training. But I really don't know whether Skinner still maintains this extreme position.
QUESTION: What about the late Jean Piaget? Where do you stand on his theories of the child's mental development?
CHOMSKY: Piaget's position is different: it's more complex than Skinner's. Piaget held that the child passes through cognitive states. According to my understanding of the Piagetian literature, Piaget and his supporters were never really clear about what produced a new stage of cognitive development. What they could have said -- though they seemed to shy away from it -- is that cognitive development is a genetically determined maturational process like puberty, for example. That's what the Piagetians ought to say. They don't like this formulation but it seems right to me.
QUESTION: In other words, Piagetians place much more emphasis on the role of experience in cognitive development than you do. Are there other differences as well?
CHOMSKY: Yes. Piagetians maintain that the mind develops as a whole rather than as a modular structure with specific capacities developing in their own ways. This is a possible hypothesis but, in fact, it seems to be extremely wrong.
QUESTION: How do you mean?
CHOMSKY: Well, consider the properties that determine the reference of pronouns that we talked about earlier. Once you ferret out these rules for pronouns, they seem to have nothing in common with the logical operations that Piagetians single out as being typical of the early stages of the child's mental development.
QUESTION: In other words, a four-year-old who may not realize that the amount of water stays the same when you pour the contents of a low, wide glass into a tall, thin container nevertheless displays sophisticated logical abilities in his grasp of the complex rules of English grammar?
CHOMSKY: Yes. And these abilities are independent of the logical capacities measured by tests. There's just no resemblance between what a child does with blocks and the kind of knowledge he displays of English grammar at the same age. In fact, I think it's sort of quixotic to expect tight interconnections between language development and growth in other mental domains. By and large, body systems develop in their own ways at their own rates. They interact, but the circulatory system doesn't wait until the visual system reaches a certain stage of organization before proceeding to imitate the visual system's organizational complexity. Cognitive growth shouldn't be different in this respect either. As far as we know, it isn't.
QUESTION: What about the problem of free will? If genes play a crucial role in structuring the mind's abilities, is free will an illusion?
CHOMSKY: Well, that's interesting. Here, I think, I would tend to agree with Descartes. Free will is simply an obvious aspect of human experience. I know -- as much as I know that you're in front of me right now -- that I can take my watch and throw it out the window if I feel like it. I also know that I'm not going to do that, because I want the watch. But I could do it if I felt like it. I just know this.
Now, I don't think there's any scientific grasp, any hint of an idea, as to how to explain free will. Suppose somebody argues that free will is an illusion. Okay. This could be the case, but I don't believe that it's the case. It could be. You have to be open-minded about the possibility. But you're going to need a very powerful argument to convince me that something as evident as free will is an illusion. Nobody's offered such an argument or even pretended to offer such an argument.
So where does that leave us? We're faced with an overwhelmingly self-evident phenomenon that could be an illusion even though there's no reason to believe that it is an illusion. And we have a body of scientific knowledge that simply doesn't appear to connect with the problem of free will in any way.
QUESTION: Do you think that science will ever solve the problem of free will?
CHOMSKY: Personally, I don't think so. People have been trying to solve the problem of free will for thousands of years and they've made zero progress. They don't even have bad ideas about how to answer the question. My hunch -- and it's no more than a guess -- is that the answer to the riddle of free will lies in the domain of potential science that the human mind can never master because of the limitations of its genetic structure.
QUESTION: Can you spell out what you mean?
CHOMSKY: We can laugh at a rat that always fails a complicated maze. We can say, "The rat is always going to fail because it can't look at the maze in the right way. It's doomed to fail this test forever." Similarly, some other intelligence, organized along hereditary lines different from our own, could look at the human race and say, "Those humans are always formulating the problem of free will in the wrong way. And the reason they don't understand the problem has something to do with their biological nature." It could well turn out that free will is one maze we humans will never solve. We may be like the rat that simply is not designed to solve a certain type of maze and will never do so even if it works on it for ten million years. Look, in principle, there are almost certainly true scientific theories that our genetically determined brain structures will prevent us from ever understanding. Some of these theories may well be ones that we would like to know about.
QUESTION: That's a discouraging prospect.
CHOMSKY: I don't see it as much of a reason to despair. In fact, I kind of like the conclusion. I'm not sure that I want free will to be understood.
QUESTION: Do you think that any other human abilities fall into the same mysterious category as free will?
CHOMSKY: In my opinion, all of them do.
QUESTION: All of them?
CHOMSKY: Take, for example, the aesthetic sense. We like and understand Beethoven because we are humans, with a particular, genetically determined mental constitution. But that same human nature also means there are other conceivable forms of aesthetic expression that will be totally meaningless to us. The same thing is as true for art as it is for science: the fact that we can understand and appreciate certain kinds of art has a flip side. There must be all kinds of domains of artistic achievement that are beyond our mind's capacities to understand.
QUESTION: Do you think genetic barriers to further progress are becoming obvious in some areas of art and science?
CHOMSKY: You could give an argument that something like this has happened in quite a few fields. It was possible in the late nineteenth century for an intelligent person of much leisure and wealth to be about as much at home as he wanted to be in the arts and sciences. But forty years later that goal had become hopeless. Much of the new work in art and science since then is meaningless to the ordinary person. Take modern music -- post-Schnbergian music. Many artists say that if you don't understand modern music it's because you just haven't listened enough. But modern music wouldn't be accessible to me if I listened to it forever. Modern music is accessible to professionals, and maybe to people with a special bent, but it's not accessible to the ordinary person who doesn't have a particular quirk of mind that enables him to grasp modern music, let alone make him want to deal with it.
QUESTION: And you think that something similar has happened in some scientific fields?
CHOMSKY: I think it has happened in physics and mathematics, for example. There's this idea, which goes back to the French mathematicians known collectively as Bourbaki, that the development of mathematics was originally the exploration of everyday intuitions of space and number. That is probably somewhat true through the end of the nineteenth century. But I don't think it's true now. As for physics, in talking to students at MIT, I notice that many of the very brightest ones, who would have gone into physics twenty years ago, are now going into biology. I think part of the reason for this shift is that there are discoveries to be made in biology that are within the range of an intelligent human being. This may not be true in other areas.
QUESTION: You seem to be saying two things. First, that whatever defines our common human nature will turn out to be a shared set of intuitions that owe much of their strength and character to our common genetic heritage -- our species genotype. Second, that the exhaustion of these intuitions in many areas is producing a peculiar kind of artistic and scientific specialization. Further progress in music or mathematics, for example, requires a scientist or artist with an unusual heredity.
CHOMSKY: Well, it's a different mental constitution -- something like being a chess freak or a runner who can do a three-and-one-half minute mile. It's almost a matter of logic that this change is going to occur sooner or later. Has it happened already? That's a matter of judgment. It's a matter of looking at, say, the twentieth century and seeing whether there are signs of this change. Is it the case, for example, that contemporary work in the arts and sciences is no longer part of our common aesthetic and intellectual experience? Well, there are signs. But whether the signs are realistic or whether we are just going through a sort of sea change and something will develop, who knows? Maybe a thousand years from now we'll know.
QUESTION: Do these possibilities ever make you feel that you're living in a time of creative stagnation?
CHOMSKY: I don't really feel that. I think that there are too many possibilities. There's too much human potential that hasn't as yet been realized. And don't forget that the vast majority of the human race hasn't even entered into the world that we're claiming may be finished. Who knows what the Third World will contribute to mankind's store of science and art when it does catch up with the industrialized nations? We are well short of real stagnation or termination but that doesn't rule out the possibility that one might be able to perceive signs of such a change, or even be able to gain some insight into the ultimate limits of our intelligence by examining these signs.
QUESTION: How do these ideas fit into your choice of linguistics as a career?
CHOMSKY: My choice of linguistics was like most people's choice of work. It was an accident that depended on whom I met, where I was, and that sort of thing. Linguistics, however, was a fortunate choice for me because I think that linguistics is an area where it is possible to construct a very rich science.
QUESTION: How would you assess your own contributions to linguistics?
CHOMSKY: They seem sort of pre-Galilean.
QUESTION: Like physics before the scientific revolution in the seventeenth century?
CHOMSKY: Yes. In the pre-Galilean period, people were beginning to formulate problems in physics in the right way. The answers weren't there, but the problems were finally being framed in a way that in retrospect we can see was right.
QUESTION: How "pre-" do you mean? Are you saying that linguistics is about where physics was in the sixteenth century? Or are we going back still further, to Aristotle and to other Greek ideas about physics?
CHOMSKY: We don't know. It depends, you see, on when the breakthrough comes. But my feeling is that someday someone is going to come along and say, "Look, you guys, you're on the right track, but you went wrong here. It should have been done this way." Well, that will be it. Suddenly, things will fall into place.
QUESTION: And then we'll have a scientific revolution in linguistics?
CHOMSKY: I would think so, although to speak of scientific revolutions occurring outside a small core of the natural sciences is rather misleading. In fact, there was one major scientific revolution in the seventeenth century and there have been a lot of outgrowths from it since then, including biochemistry and molecular biology. But that's it. Nothing remotely resembling a scientific revolution has ever occurred in the social sciences.
QUESTION: How should a scientist exercise responsibility for the uses of his research?
CHOMSKY: The same way that any human does in any area of life.
QUESTION: Do you think that there are areas in science so potentially vulnerable to social misuse that they should not be pursued?
CHOMSKY: I think there are. For example, research on how to build more effective nuclear weapons. I don't think that should be pursued.
QUESTION: What about fundamental research -- say, basic research in molecular biology that might conceivably give the weapons makers of the next generation a new set of destructive tools?
CHOMSKY: There's no simple answer to that question. Human beings are responsible for the predictable consequences of their actions. I would stop doing what I was doing if I discovered that I was engaged in an area of scientific research that I thought, under existing social conditions, would lead to, say, oppression, destruction, and pain.
QUESTION: An anachronistic question then: If you were a physicist in 1929, would you have done basic work in nuclear physics even though there was already speculation about the possibility of someday building an atom bomb?
CHOMSKY: It's not an easy question. It's tempting to say, "Yes, because we have to understand the world." On the other hand, it could be that basic research in nuclear physics will lead to the extinction of the human race or to something close to that. So I don't think a glib answer is possible. Still, if you ask me specifically, I'm sure that my answer would have been yes. I would have done the work just out of interest and curiosity and with the hope that things would somehow work out. But whether that would have been the morally responsible path is not clear.