Conversations On The Leading Edge
Of Knowledge and Discovery
With Dr. Jeffrey Mishlove
LANGUAGE AND CONSCIOUSNESS
Part II: HOW WE UNDERSTAND LANGUAGE
with STEVEN PINKER, Ph.D.
JEFFREY MISHLOVE, Ph.D.: Hello and welcome. I'm Jeffrey Mishlove. This is the second in a four-part series on "Language and Consciousness." With me is Dr. Steven Pinker, a professor in the Department of Brain and Cognitive Sciences at MIT, and director of the Cognitive Neuroscience Center at MIT. Dr. Pinker is the author of numerous books, including Visual Cognition, and most recently, The Language Instinct. Welcome again, Steve.
STEVEN PINKER, Ph.D.: Thank you.
MISHLOVE: I suppose, as a starting point for looking at our topic now, which will be how it is that we actually produce language, and how it is that we come to understand language -- the miracle that language is for us, which distinguishes us pretty much from all other species -- a good starting point might be the young child or infant. It seems from your writings that children have an understanding of language structure and grammar and the use of language so vast that it would be impossible to write a textbook to explain it all.
PINKER: Yes. I think any parent will remember the experience of their child going for many months just saying a word at a time -- milk, doggy -- maybe combining them into tiny little microsentences, like "More milk," "Bye-bye, doggy." Then in a span of a few months the child is conversing in fluent sentences, often putting forth very articulate arguments about why they shouldn't go to bed at seven and they should go to bed at eight, and the parent often may not even remember that burst of language development that takes the child from a couple of words to almost all of English grammar. In fact we can even hear children as they struggle with figuring out English grammar, when they make errors, like "We holded the baby rabbits," or "Look at those two mouses." Now, we call them errors, but when you think about it they actually follow the logic of English pretty well. If you can have dog-dogs, cat-cats, why not mouse-mouses? Or if it's fold-folded, why not hold-holded? So those errors show that the child is constantly on the lookout for the grammatical rules of English, like, "To form a past tense add -ed; to form a plural, add -s." And the child is unconsciously applying them to words that the child has never heard in that form before.
MISHLOVE: I suppose one of the unusual things is that a child seems to be able to learn language much more rapidly and more perfectly than an adult can learn a new language.
PINKER: Yes, that's very clear to immigrants who will come to a new country, and will often take classes and buy books and listen to tapes and struggle, and their child, just going out into the playground, will come back speaking the new language so much better than the parents that the child will often make fun of the parents' grammatical errors and pronunciation patterns. There's a very clear difference.
MISHLOVE: How can we explain this?
PINKER: We don't really know why, but I think the best guess is that the brain of a child is literally different from the brain of an adult. We do know that at the age of two, which is when most language learning happens -- that's when this burst of language development tends to start -- that the brain of a two-year-old has fifty percent more connections between brain cells than an adult has. Basically from two it's all downhill in terms of the power of the brain to store new memories and form new associations, because it's the connections between the brain cells, we think, which is where the memory traces are laid down. And it could be that the parts of the brain that are designed to learn language are actually functioning better and have more storage capacity in the child than in the adult. The brain of a child is also literally working harder. A two-year-old's brain is burning fifty percent more sugar and consuming fifty percent more oxygen than an adult's brain, and that might be why they're so much better at it than we are.
MISHLOVE: So there's a very real sense in which a two-year-old is smarter than an adult.
PINKER: There is, and the paradox is that the two-year-old is figuring out language, as far as we can tell, pretty much effortlessly. I mean, the child doesn't have to struggle at it; we don't have to force them to practice the way we force them to practice at piano or at reading or at tying their shoes. It just kind of happens. On the other hand, of course, there are a lot of things that a two-year-old can't do. I mean, we won't let a two-year-old drive or vote; or a two-year-old can't read. So it suggests that the mind isn't just one thing. In fact smartness is even a misleading way to think about the human mind, as we can see in children. In some ways kids are smarter; in some ways they're not as smart.
MISHLOVE: Of course.
PINKER: It might be that the language centers of the brain gear up fairly early because language is the mechanism by which children learn what's important in their culture. That's probably why our species has such a long childhood compared to other cultures. And when you're building a child, so to speak, the first thing that you install is language, so that the three-year-old is all ready to pick up what's important in the culture, and the rest of the childhood can be spent learning it via language.
MISHLOVE: You used the term the language centers of the brain. That already is sort of a very theory-laden idea -- that language occurs in particular centers.
PINKER: Yes. It's certainly not as crisp as the diagram of a cow in the meat market, where the dotted lines show you where the flank steak is and where the sirloin is. The brain doesn't have those dotted lines, so there is no language center in that sense. But it's also not true that the brain is a meat loaf with no parts at all. We know this because of the effects, often, of disease and brain damage. If someone has a stroke and part of the brain dies because of lack of oxygen, depending on where the damage is they may have their language intact, or they may lose the ability to speak but be otherwise intelligent. And by looking at lots of patients who've had strokes, we can take a guess as to what part of the brain language is involved in, and it seems to be concentrated, at least in right-handed people, in the left hemisphere, and in a band of areas in more or less this part of the brain -- in between -- there's a big lump that you'll notice in any picture of the brain, called the temporal lobe, and a gulf that separates it from the rest of the brain, and on both sides of that gulf are where the language circuitry seems to be concentrated.
MISHLOVE: In each hemisphere.
PINKER: Well, more on the left, much more on the left, at least in right-handed people. In left-handed people it's also more likely to be on the left, but it's a little more likely to be on the right in left-handed people than in right-handed people.
MISHLOVE: So the use of language is considered a left-brain function.
PINKER: That's right, yes, in the vast majority of people.
MISHLOVE: I guess that's where we get the idea of linear left brain, because words come out one at a time, in a linear fashion.
PINKER: That's right. It's a bit misleading, because you also group the words into higher-order structures. I mean, that's why so many sentences are ambiguous. For example, the TV Guide once had an announcement that in this evening's program Dr. Ruth will discuss sex with Dick Cavett -- unintentionally humorous because of the two ways of grouping together the words. It's either discuss sex / with Dick Cavett -- that's what the person had in mind when they wrote it -- or discuss / sex with Dick Cavett. So language isn't really linear in that sense, and the popular depiction of anything in the left hemisphere being linear is a little inaccurate. In order to understand sentences you've got to chunk the words into phrases.
MISHLOVE: Well, one of the most interesting things that you've pointed out is that as we're speaking I'm not even pausing in between my words at all, so that somebody hearing this who isn't an English speaker will have a hard time understanding it, just as I would have a hard time making out what the words are if I'm in a foreign country.
PINKER: That's right. If you actually were to look at a spectogram -- that is, a visual picture of the sound wave when you speak -- there aren't any little pauses in between the words the way there are little white spaces in between the words on paper. It's one seamless continuum, and the brain actually hallucinates boundaries between the words. That's why, as you pointed out, when you hear another language it almost sounds like there are no words, that it's just one long ribbon of sound. It's also why in a lot of jokes and word play we can take advantage of that, like the song "Mairsey Doats and Doasey Doats and Little Lambsey Divey." The reason that that works is because in fact words don't have audible boundaries.
MISHLOVE: That's one of the most important features, I suppose, that would distinguish human interaction from, say, computer speech recognition programs.
PINKER: In fact that's one of the reasons why computer speech recognition has been such a hard problem. You know, you buy a computer, the state of the art, you know, the latest thing out of Silicon Valley, and you've got to sit there typing these things in and learning the computer language. Ideally you'd think you'd be able to speak into a microphone and have the computer take dictation. Now, the state of the art has been improving in the last couple of years, but still a computer's no match for a human being, for, say, a stenographer, and one of the reasons is the computer has a really hard time finding the boundary between words. In fact, even in the best systems -- if you go out and you buy the best system available for your personal computer, you - have - to - talk - to - it - like - this, because if you talk normally it won't know where one word ends and the next one begins.
MISHLOVE: And we're not even beginning to discuss the problem a computer would have in resolving the ambiguities that human beings seem to deal with all the time.
PINKER: Right. I've only talked about even hearing the words, let alone the content. One of the problems is that we don't realize how ambiguous language is. We unconsciously figure out what the other person must have meant, but the actual syntax and semantics of the words themselves often have lots and lots of meanings. Even something like "Mary had a little lamb" -- everyone knows what that means, right? But if a computer tries to figure out what it means using the rules of English grammar, it will give you a lot of different meanings, like "Mary ate a little lamb," or "Mary gave birth to a little lamb," or maybe "Mary had sex with a little lamb." These don't even occur to us, but they are in the words, and the common sense that we use to figure out what the other person means is something that no one has been able to program a computer to do yet.
MISHLOVE: Well, there is a sense in which intuition is at the root of language. Some people even go so far as to say that they don't know what they're thinking until they actually hear what they've had to say.
PINKER: Certainly the logic of language itself is something that we have an intuitive sense of but that we can't articulate. No one, unless they study linguistics, can tell you why some sentences sound bad and others sound good; why you can talk about an apartment as being mice-infested, but it sounds odd to talk about it as being rats-infested. You say it's rat-infested, even though when you think about it, you can't have an infestation with one rat.
MISHLOVE: Some things ring true.
PINKER: Some things ring true. I think there are very logical reasons behind these intuitions, but they're not available to consciousness. They're the kind of thing that a linguist or a psychologist, working through the logic, can figure out. And in the case of mice-infested versus rats-infested, I think we understand why there's that difference, but you can't feel it by turning your ear inward and trying to feel the logic of language inside you. It just happens; it's like digestion. It's a complicated process that goes on inside you, and you aren't aware of how it works.
MISHLOVE: Yes. Well, the great linguist Noam Chomsky has described the deep structure of language -- something that he believes is innate in the brain and is even common to all the languages that humans speak. It seems at this point in our discussion that it would be appropriate to explain to our audience what Chomsky meant, and where you would agree or disagree with these ideas.
PINKER: Yes. Chomsky proposed several things that I agree with that were truly stunning claims at the time at which he first made them in the late fifties. One of them is that to understand human behavior you've got to study the human mind, and at the time at which Chomsky wrote it was the era of behaviorism, where you studied rats pressing bars and laws of reward and punishment or dogs salivating to tones and the laws of conditioning, and Chomsky said that the only way you can make sense of human behavior is to figure out what's going on up here. And that was considered a revolutionary claim, believe it or not. He also said that the mystery of language that we have to solve is how children acquire it -- that a child doesn't know what language he's going to end up having to speak; the child hears parents use a certain number of sentences in a language, and within a short amount of time can then produce and understand an infinite number of new sentences. So children aren't memorizing, clearly; what they're doing is they're abstracting the logic underlying language. And he said that in order to figure out how children can do it, how a three-year-old can end up speaking English or Japanese or Swahili, the child has to be born with the basic logic of language wired in -- now of course, not the details of any particular language; it would be crazy to say that Swahili children are born with Swahili. We know that's not true, because if they're brought up in an English-speaking environment they end up with English, or vice versa. But the plan underlying language -- the phrases, the nouns and the verbs and the subjects and the objects -- that basic software is wired in. Otherwise language development would be impossible. And one of the reasons that he said that is that if you look at languages across the world, they are all based on a similar plan, which he called universal grammar. So even if you look at something like Japanese, which is not related to any other language on earth as far as we know, so there's no historical connection that we can trace back from English -- now, Japanese and English differ. English has subject-verb-object: "Dog bites man." Japanese has subject-object-verb: "Dog man bites." So they are different, but on the other hand they both have subjects, objects, and verbs to begin with, and there's just a minor difference of what comes first, the verb or the object. If you look at other systems -- at, say, computer languages, or languages of logic, or musical notation -- they don't even have subjects, verbs, and objects, so the most striking thing is that languages are built out of these same kinds of rules to begin with, and Chomsky's hypothesis is that kids can figure out how English or Japanese works because they don't have to start from scratch. They don't have to figure out what is a verb or a noun; they've got to find the verbs or nouns in their parents' speech.
MISHLOVE: So there's something, when you say it's wired in, it's part of our biology.
PINKER: It's part of our biology. That is the claim. And that was also considered shocking and revolutionary, because I think there's a widespread idea in our culture that the mind is a lump of wax that the culture shapes. So the claim that there's actually some structure in there to begin with, and that it's as complicated as knowing the basic design of grammar -- that was a very surprising claim, and there are many people now who disagree with Chomsky on that.
MISHLOVE: Well, the standard term in Western philosophy and psychology was tabula rasa, meaning a blank slate; the mind was a blank slate in infancy.
PINKER: That's right. The metaphor has changed, but it's the same idea. There's the lump of wax with Plato; the blank slate I think was Locke; and now it's the general-purpose computer. And all of those metaphors are things that Chomsky has argued against.
MISHLOVE: In any case none of them are particularly biological.
PINKER: That's right, as it turns out. Right.
MISHLOVE: You also point out, though, that Chomsky has been adulated and in many ways misinterpreted by people who have never read his works, which are highly technical.
PINKER: That's right. Chomsky has become something of a cult figure. He's one of the -- I think he's the most cited living person in all of the arts and the humanities, so he really has had an impact. But a lot of people don't get the details right, even the idea that there's, say, a deep structure that's common to the world's languages. Technically that isn't actually part of the theory. There's a universal grammar that says that the phrase structures of language are all built out of the same parts and organized according to the same plan, but it's sort of a picky detail; it doesn't happen to be deep structure. There's also been a lot of ridiculing of Chomsky because the claim that any part of the mind is innate is so shocking that people interpret it as saying that everything in the mind is innate -- that, gee, if you think language is innate, do you think then that the instinct to repair carburetors is innate, or matching ties with shirts is innate, riding bicycles is innate? You know, where do you stop? And people have actually thought that Chomsky has made claims that everything in the mind is innate, which obviously no sane person would make. But people can't imagine the middle ground. The middle ground comes from the analysis of language that finds things that are much more abstract, sort of the mathematical structure underlying language -- much more abstract than you'd find in traditional grammar books -- but that the day-to-day work of the linguist consists in finding. That is, what actually is common between English, Swahili, Japanese, and so on? That's the part that could plausibly be innate, and so this claim that there's abstract universals, and the claim that there's something innate about language, go together, because it's the abstract universal grammar that's the only thing that could be innate. English obviously can't.
MISHLOVE: And you mentioned that we do have systems of language, like music, a system of notation at least, that doesn't have the standard subject-object-verb, and here on this planet there are thousands of languages, and to your knowledge, to our scientific understanding, they all follow this universal grammar that Chomsky described.
PINKER: Yes. I wouldn't go so far as to say that they're all identical in terms of their use of subject, verb, and object, but they all do follow the same plan. If I were to go to a library and pick up a book on some language of Ongo Bongo, or any exotic place in the world, and I open it up, I don't think I would be surprised. It would say, well, in this particular language you've got four cases instead of three cases, but they'd all be recognizable, and the general stuff out of which language is made you find, I think, all over the planet. And the complexity of language -- there's no simple, rudimentary languages anywhere on earth.
MISHLOVE: Yes, and we're going to talk in a future program in this series about the evolution of language and the many different types and complexities of language. It's a fascinating topic. But I'd like to get back to the mystery of how we actually produce language. How much is understood about how I'm able to translate my thought into a spoken question?
PINKER: I think more is understood about understanding language than producing it, and that's because -- it's a problem of free will, when you come down to it. If I speak to you, I could at least control what I'm saying and see how you interpret it. But there's no way that I could cause you to say a particular thing at a particular time. So it's just easier to study comprehension in the lab than production. You can't force a person to say something unless you have him memorize it, in which case it's a different process.
MISHLOVE: Now let me ask you -- just for a moment I want to look, then, at language recognition. But you're saying as a scientist that language is a place where free will enters into the picture.
PINKER: In a way it does, in that when you speak you're not reacting involuntarily to what's around you. It's not like anything that pops into your field of vision you have to describe it. You convey your intentions, and that's obviously much harder to control. To the extent that there's something that we want to call free will, language is the place where it's manifested.
MISHLOVE: Well, I'm glad to hear that, because many branches of science don't even like to acknowledge the possibility of free will.
PINKER: Scientists studying language might just attribute it to causes that we can't quite control in our experiments. But yes, that doesn't mean that the problem goes away.
MISHLOVE: It's a real problem, and what you're saying, it even constrains our ability to do research on language production.
PINKER: I think there are probably ten studies on comprehension for every one study of production. It's not that we don't know anything about production. We do know things, for example, from slips of the tongue, like Spoonerisms. Everyone makes them, but the most famous person was the Reverend Spooner, who would say things like, "It is now kisstumary to cuss the bride. You've hissed all my mystery lectures and tasted the whole worm." But everyone does it once in a while. A friend of mine was asked about his new condominium. They said, "How do you like it?" and he said, "Oh, it seats my nudes." Now just trying to figure out where those errors come from tells you something about how we mentally plan the units of language before they come out of our mouths, and it suggests that we have a set of slots that we intend to fill with the sounds of language, the phonemes. And we sort of fire up the phonemes that we want to plug into those slots, but sometimes one phoneme will get plugged into two slots, or sometimes two phonemes, on their way to getting plugged into the slots, will have their positions switched. So we know something about the software, a little bit about the software that goes into planning a sentence, through that method.
MISHLOVE: And yet it's interesting, because the phonemes themselves don't carry any meaning.
PINKER: That's right. That's another interesting thing about the design of language -- that there are two levels of patterning in language. And this is true of all languages in the world, including sign languages -- that we arrange words into phrases, and the meaning of the phrase depends on the order of the words, so that "Man bites dog" has a different meaning from "Dog bites man." But then inside each word there's another level at which we arrange little beads on a string, but there the meaning has nothing to do with the particular units, the phonemes, and that's why doubting the existence of God isn't the same thing as doubting the existence of dog, to use an old joke. There it's the units of sound -- I think they're there so that we can generate an unlimited number of words when we need them. When we need to express a concept with a new word, we can pull out the alphabet and rearrange it in a different order, so there's a new word like crufty, meaning not in good shape, or to mung, from computer hackers, meaning to destroy or disable. We use old English letters in new orders, and there's almost an inexhaustible set of new words that we can come up with, but for each one the meaning has to be invented from scratch. You can't figure out what mung means from the meaning of mu-, the meaning of uh-, the meaning of ung. And that's again true of all human languages.
MISHLOVE: And the interesting thing, I suppose, in language recognition is that you can understand oftentimes a thought that has never been expressed before.
PINKER: In fact, except that's what you're always doing in understanding. I mean, unless you have a really clich‚d conversation, maybe picking up someone in a bar, those first few exchanges, or go according to an all too familiar script, except for those kinds of situations, virtually any sentence that you utter, it's never been uttered by anyone else before in the history of the planet, and you're therefore understanding thoughts that have never been expressed in exactly that form before. And that, I think, is the essence of language. Again, that gets back to Chomsky; he pointed that out as one of the key properties of language that we've got to account for.
MISHLOVE: And how do you account for it?
PINKER: Well, I think we have a set of rules -- not rules in the sense of the style manuals and the grammar lessons, but rules in the sense of software -- that order phrases together and order words inside the phrases according to a definite code that allows you to extract the meaning from the words and how they're arranged. So "Dog bites man; man bites dog," for example -- we know that we find subject; we look up in our mental dictionary, so to speak, what bites means. Bites says, "Subject of the sentence that I appear in is the one doing the biting," and therefore we know who bit whom. If you have a much more complicated set of rules like that, they can be applied to long strings of words to pick out what thoughts those words are expressing.
MISHLOVE: Steven, we're just about out of time, but I'd like to point out here that you've already explained that computers aren't able to do this. So these rules can't be just syllogisms; there must be something quite biological or artistic or impressionistic about language.
PINKER: Yes. The thing is that when we talk we're only giving hints as to what we mean. If I was actually to explain to you everything in my mind behind a particular sentence, I'd have to strap you down in that chair for half an hour to get one thought out. So what I do is I put out a few words, and I count on the fact that you and I are human beings, and you can guess why I'm saying what I'm saying, in order to figure out what I mean.
MISHLOVE: And I appreciate that, because we are out of time. Steven Pinker, it's been a pleasure being with you on the second of our four-part series on language and consciousness. Thanks so much.
PINKER: Thank you.
- END -