Inside the Psychologist's Studio with Lila Gleitman

Just another WordPress site

Inside the Psychologist's Studio with Lila Gleitman

Susan Golding-Meadow: It’s my pleasure to be interviewing Lila Gleitman I first met Lila in spring of 1971. That’s 46 years ago, for those of you who are having trouble doing the math quickly, when I was applying for graduate school at the University of Pennsylvania. So at the time, it was not at all surprising to interview with faculty about a place that you wanted to go to. The surprising thing was that Lila wasn’t at Penn She was actually at Swarthmore. But it seemed clear to me that if you want to do something in language acquisition, which I did, and you were anywhere in the Philadelphia area, you had to go and interview with Lila Even in 1971, Lila was doing strikingly original work on language and learning that was changing the face of psychology and cognitive science So I came with my honors thesis from Smith College in hand, and I was doing great explaining all the control groups and the data which in itself was a little difficult since I brought this project back from the University of Geneva with Piaget. You know, Piaget really didn’t do control groups, but I was doing great I really thought I had it until I was stopped in my tracks when Lila asked me why I did the project So this was a highly embarrassing moment but, as you can see, it was an important one because I remembered it for over 40 years. That was the first time but not the last time that Lila taught me that doing a study elegantly and carefully is important, but then knowing why you did it is even more important. And you should be able to talk about what that study is about to anyone, preferably in terms that your grandmother can understand. So this is a fundamental lesson about science that I learned from Lila even before I started paying tuition So Lila’s research findings, which dates from the 60s, have energized the fields of psychology, linguistics, and cognitive science and have spun research programs that define current investigations of language and language learning And Lila is, if anything, more active and productive now as we just heard in her inspiring plenary address last Thursday Lila has received many well-deserved honors She’s a member of the National Academy of Sciences and the American Academy of Arts and Sciences. She has won the distinguished scientist award from APA, the John McGovern Award in Behavioral Sciences from AAAS, the Prix International from the Fyssen Foundation, and my personal favorite, the Mentor Award from APS In addition to doing research noted for its originality, clarity, and depth, Lila is an inspirational teacher and mentor, and her students currently populate psychology and linguistics departments all over the country So Lila’s goal is to unravel the mysteries of how children learn language. Her fellow graduate student when she was studying linguistics at Penn, Noam Chomsky, approached the language learning puzzle by documenting the principles underlying the linguistic systems that children learn. And because those principles are not transparent in the linguistic input that children receive, Chomsky assumed that they had to be innate in the world, [indiscernible] in the child But Lila’s approach to the problem of language learning was and is different. She conducts empirical studies of the language learning process itself. While not denying that children come to language learning predisposed to learn human language, indeed Lila is one of the most significant and influential proponents of this position. Lila bases her claims on careful and very clever empirical investigations So two themes run through Lila’s research The first is to discover how language learning is constrained, what endowments do children bring with them that narrow down the set of languages they can learn. The second is to understand how children learn the particular language to which they’re exposed, what processes do children use to extract units and meanings from the input they receive. In other words, how does it work? Well, before it was fashionable to be interdisciplinary, Lila was already forging the ties that now bind psychology, linguistics, and cognitive science. She has fostered the marriage of ideas through her own work, through the many students she’s nurtured, and the interdisciplinary cognitive science program she cofounded at Penn which had a really defining effect on

the field. We’ll hear about each of these influences in a moment. Psychological science will look very different were it not for Lila Gleitman. She’s one of those rare people who has altered the course of intellectual inquiry, and she’s continuing to do so As you will no doubt be able to tell for yourselves the minute Lila starts talking, she is from Brooklyn, not far from where Bernie Sanders grew up. So today’s interview is really the story of how a kid from Brooklyn found herself in the middle of a cognitive revolution in the 1950s and not only witnessed that revolution but was central in shaping its growth and direction. This revolution is now completely taken for granted. It�s the background for really everything that we do. But Lila was there at the start and she can tell us about it. So let’s start at the beginning in Brooklyn As I understand it, Lila, you started your career at James Madison High School, which surprisingly has spawned many well-known people Do you want to tell us a little bit about James Madison High School? Lila Gleitman: Well, James Madison High School was an urban school with a mostly secular Jewish clientele – what they called the student body. This was in the early 1940s I guess There was a war on. There was a war and that colored everything Susan Goldin-Meadow: You might want to tell them which war Lila Gleitman: I bet you can guess. The Second World War was then on and it colored our lives in some background way. But like high schools everywhere, it was mainly about ourselves as high schoolers. So it was an interesting school. A lot of interesting people came from there like Ruth Bader Ginsburg, and Judge Judy, and Chuck Schumer, later Bernie Sanders These people came later. So it was a relatively nice place to be, although myself, I was sort of an unfocused outsider for most of my high school career. I remember it mainly as a fog I don’t really remember too much about it Susan Goldin-Meadow: And so from that fog in New York, you ended up going to Antioch in Ohio. So how did you manage to get to Ohio from Brooklyn? Lila Gleitman: Antioch. You know what Antioch is? Antioch was a left-wing militant, left-wing lunatic French. How I got there, I mean it was a natural place for me to go actually But we did not have or at least I did not have any counseling, but I remember there was a very large desk and there were brochures from colleges all over it. I wasn’t too keen in this idea of going to college. It all seemed pretty alien to me at the time. But I happened to pick up this brochure for a college. We didn’t have to be in college all the time, only half the year. The other half of the year you would go out and work on co-op jobs, and that sounded great. I really didn’t have much of an idea of what the college was about, which was wonderful if you were political But that’s how it happened, I just found a brochure on the desk and that’s where I went Susan Goldin-Meadow: And so psychology didn’t really grab you at Antioch. What did you major in? Lila Gleitman: At Antioch? Susan Goldin-Meadow: Yeah Lila Gleitman: English Literature Susan Goldin-Meadow: Because everybody majored in English Literature Lila Gleitman: It was wonderful. I always loved to read novels and so forth and so on But even in college I wasn’t particularly focused. Just vagueing [sounds like] around Susan Goldin-Meadow: And so then from there you went back to New York, right? Lila Gleitman: Yes Susan Goldin-Meadow: And you looked for a job as an editor Lila Gleitman: Yes. Actually I was the editor of our little college literary magazine, which was called Idiom. And I went to New York with

a friend, a very interesting friend who’s name was Eliot Fremont-Smith. In the left-wing proletariat ways of Antioch College, he came to be called Hyphen. So Hyphen and I went to New York. I was the editor and he had been the assistant editor So this is the one woman’s story. I’ll tell you what it was like being a woman in those days. So we went to New York, and I don’t know how but morning came and we went out looking for jobs. We came back that night and he had gotten a job as an entering junior editor at Doubleday. That was wonderful. And he went on to a very nice career. He became the editor of the Village Voice. Eventually he followed that trajectory. I came back and I had gotten a job as a gal Friday for the Journal of the American Woodworkers Association So you notice the trend [sounds like]? But we’re talking, what, 1948 or something like that. No, it was later than that. 1952 But it doesn’t matter. It was the bad old days, probably it still is. The disparity and what employment we were able to get was not even noticeable for us. We had a glass of wine. We toasted each other for the wonderful jobs. But you know, my job was I got the coffee and I type the envelopes. And I didn’t stay there. I mean it was sort of — I mean I’m not saying no woman succeeded, but it was kind of a dead end. The fact that women ended up as gal Fridays, I mean that’s such an undignified word in retrospect. So that’s the way it was Susan Goldin-Meadow: So there you were, being a gal Friday. And now how did you get to psychology? Lila Gleitman: Well, again, life is a series of accidents. I went and visited some friends at Dartmouth, I think, that summer and there was a guy there who had just gotten his first job as an assistant professor of psychology at the University of Pennsylvania. That was my first husband. He wasn’t my first husband yet. It took a little longer than that. So we rapidly hit it off. We got married, and that was really great. And that’s how I got to the University of Pennsylvania Susan Goldin-Meadow: So who was this first husband? Because he’s not insignificant in the cognitive science revolution Lila Gleitman: No. No. He is a very interesting, very smart man named Eugene Galanter. Some of you oldtimers will remember the name. There was sort of a dawning change going on in psychology and everywhere, as we’ll soon see, in linguistics, the emerging computational sciences. Anyhow, he notably wrote with George Miller – Miller, Galanter, and Pribram. It was called Plans and the Structure of Behavior, and it was very controversial at the time. Structure of behavior, so sort of an oxymoron from the point of view of the preceding associationist generation. So there were jokes around, disparaging jokes about the book which — how did the expression go? Miller devised it, Galanter wrote it, and Pribram believed it. Have you heard that joke? So it was quite controversial, but it had an effect in this dawning change of perspective Susan Goldin-Meadow: So Galanter was really significant in writing this book, but the other way in which Galanter was significant is he got you to Penn

Lila Gleitman: Yes Susan Goldin-Meadow: And that changed your life Lila Gleitman: Yes Susan Goldin-Meadow: So tell us how that changed your life? Lila Gleitman: Well, again, I was sort of an unfocused person intellectually. But when I married him, he was an assistant professor It turned out women could take courses free If you were a faculty wife, you could take courses free. And that’s what I liked to do So I took more courses. I decided I’m going to learn Greek and I’m going to do some of those things which as an English major, which was wonderful and which continued to occupy my interest for life, find out something about the — it was the 20th century at that time So I thought I’ll also take an undergraduate sequence in mathematics. I had done nothing like that in college. I was very devoted to Greek culture. So as it panned out, the mathematics [indiscernible] along in the background. But the study of Greek became my first real intellectual focus, something that kept me up night and day translating sentences Susan Goldin-Meadow: Well, yes, you figured out you had no intuitions for mathematics and you were great Lila Gleitman: Very little. Very little Susan Goldin-Meadow: But you also, I think, discovered at that time that it really wasn’t the culture of Greek that you were interested in. It was truly the language, right? Lila Gleitman: It was truly the language Susan Goldin-Meadow: And it was Henry Hoenigswald who helped you discover that Lila Gleitman: Yes. Yes Susan Goldin-Meadow: So why don’t you tell us a little bit about Henry Hoenigswald Lila Gleitman: Henry Hoenigswald was a brilliant historical linguist, and maybe for those of you who know linguistics, maybe the major figure in developing the comparative method by which you could not completely quantify but close to quantify the relationships among how languages change and grow and diverge and come together over historical time and place. A wonderful brilliant Indo-Europeanist as well. But, of course, he was a teacher in a big university, so he had to teach introductory Greek among other things. But he was very patient, and that was fine. And introductory Greek. But you know, if you were a normal person and you teach Greek, as you’re going through the text, you talk about what’s going on I recall that was a fateful day. We were translating sentences and there was something about Alcibiades who was a big dandy of Athens. And he was saying, oh, in that year 430 BC or something, everybody in Athens was wearing Alcibiades� sandals. And I remember saying in some surly Brooklyn way, okay, crush the next sentence So we went on, and he buttonholed me at the end of the class. He said, you know, you’re in the wrong department. I know my face fell because I was determined to be this great Greek scholar. And he said, “That’s okay, it happened to me too. I moved to the linguistics department. You should move to the linguistics department.” Susan Goldin-Meadow: And he is the one who graciously gave you up as a student, as I understand it, and introduced you to Zellig Harris Lila Gleitman: Yes. Yes Susan Goldin-Meadow: Which is another pivotal point in your life Lila Gleitman: But he said to me that day, “And you won’t work with me. You will work with –” And his face was kind of shining “You will work with Zellig Harris.” All of Penn and much of the linguistic world was devoted to Zellig Harris for a very good reason Susan Goldin-Meadow: So you came in as a linguistic student in Zellig Harris’ lab Lila Gleitman: Yes Susan Goldin-Meadow: And the first thing he did was tell you you had to go find a job Lila Gleitman: The first thing he said is that he forgot to apply for the scholarships yet, so I go get a job Susan Goldin-Meadow: All right. So tell us a little bit — Lila Gleitman: So I had to get a job for a year, yeah Susan Goldin-Meadow: And tell us a little bit about this job because this job is pretty interesting Lila Gleitman: Well, I remember I had connections with psychology. So I got a job at Eastern

Pennsylvania Psychiatric Institute. There was a psychiatrist, and this psychiatrist was something of a — I don’t know what to call him. But he said, oh yeah, I’m going to hire you as a structural linguist. I had been in linguistics for, what, six weeks or three weeks or something like that. You know, you put on the door the name of the person and underneath it was Structural — Lila Galanter, Structural Linguist. So he hired me for his own purposes and not too much happened. But something did happen Susan Goldin-Meadow: Something did happen So tell us about that Lila Gleitman: He had gotten the contract to write the psychological entries for the next edition of Webster’s Dictionary, and that was quite interesting. So he decided that I should do it, and it was wonderful You get 100 cards. I mean, I think they have troops of monkeys that type 100 cards on which your mystery word has occurred somewhere So you have a little paragraph or text and you get about 100 of them. You’re supposed to riffle through those and figure out a paraphrase presumably that will be the dictionary entry for that word And a lot of funny things happened there of which there was another person working as a secretary there. A woman. I loved this story She came and she said, “What are you doing with those cards all the time?” I could imagine how proud I was. I said, “I am writing the definitions for Webster’s Dictionary.” And she said, “How do you do that?” So I thought about it for a second and said, “Well, I make them up.” She said, “You?” And then she said – I always tell this to my undergraduates – “You make them up? I will never look up a word in the dictionary again.” It was so wonderful because people think the dictionary came from God. It wasn’t God. It’s an unemployed first year linguistics student somewhere Susan Goldin-Meadow: All right. So tell us about that famous word that you got into the dictionary Lila Gleitman: Oh yeah. So yes. So this was also wonderful because, believe it or not, among the psychological words was the word “fuck.” Actually it was quite interesting We won’t talk about that, but those of you who thought about symmetrical predicates can realize that “fuck” is quite an interesting word. Words like “marry” or “equal” or so forth, it behaves in very interesting ways. So that was my first approach to that. But the interesting thing I have taken it as my chief accomplishment in life, that I’m the gal who put the word “fuck” in the dictionary Susan Goldin-Meadow: That may not be your most Lila Gleitman: Not bad Susan Goldin-Meadow: We’ll vote at the end of the interview to see. All right, so let’s go back to Harris a little bit because he did a little bit more than just get you underemployed Lila Gleitman: Yes Susan Goldin-Meadow: He had a very important theory, and I think it would be good to tell the group a little bit more about Harris’ theory because I’m not sure that everybody understands what Harris was about at that time Lila Gleitman: Well, Harris was the pinnacle of structural linguistics which had been developing mainly in America for 50 or 70 years and was parallel to the behaviorist revolution going on in the next departments in psychology Basically, it was a procedure for building the whole structure of the English sentence starting at the very bottom with the lowest level items, and everything then grouped together by class inclusion. So you had the phones and the phonemes, morphemes and the words and so forth. And what the theory was, was how you get there. So we can’t talk about that theory But to get an idea of how this all worked,

you remember the game Hangman? Okay. So as each person chooses a letter, the actions for what can come next and be a word of English get narrower and narrower and narrower. So when it gets down to only one, you lose in the game. So that was via discovery procedure I read a very famous article called From Phoneme to Morpheme about the distribution of items at the level below. Then from morpheme to phrases, from phrases to sentences and so forth and so on. That’s the story Susan Goldin-Meadow: But at some point you broke with Harris because his theory wasn’t working for everything that you were thinking about, and that was a pivotal moment too Lila Gleitman: Right. So to put that in a nutshell, the idea of transformations was Harris’ idea, although in Chomsky’s hands it became a very different thing. So he was trying to look at the relationships among sentences. He was trying to go up by level by level to a discourse for that matter. But at any rate, in order to analyze sentences, the idea was they all had to be reduced to one normal form. One normal form. So you had to see the relationship between the active and the passive, for example, because the passive is just a distortion of some — I wouldn’t say underlying because he would turn over in his grave. But for every sentence of this kind, there’s a sentence of the other kind So here’s the shock, and here we go with symmetrical predicates again. I was doing conjunction to get sentences like John and Bill walked, and the source sentences that John walked – and what did I say – and Mary walked. Very good, you could write a little algorithm for that. But I ran into a set of funny cases. Like John met Mary. Isn’t John met and Mary met? And you can go through the fuck example for yourself again as often as you’d like. Two and two is four. It wasn’t a combination of two is four and two is four That just won’t fly. There is something wrong with the — what that pointed to. Of course, it’s just particular points to a place with the theory It cannot handle a particular kind of data and should make you look around Susan Goldin-Meadow: But Harris wasn’t so flexible because at that point, the fact that this idea didn’t fit with his theory was enough for him to say out, go Lila Gleitman: Yes Susan Goldin-Meadow: And you never finished your dissertation with him, right? Lila Gleitman: No. As a matter of fact my daughter, Claire, and I had dinner with Noam Chomsky the other night here before the convention started. He had also been — and Chomsky said, you know, Harris thought after his book came out that was the end. It was all done. Linguistics was finished and there was nothing more to do except some details. He was not really good at confronting these kinds of challenges So never mind me, there was a separation between him and Harris. Chomsky had quite different ideas about the forms of the grammar and what the grammar was telling you about the human mind Susan Goldin-Meadow: Right. Okay. So before we just go to Chomsky, tell us — because Harris, although you broke with him, he did set the stage for some of the theories that you developed that are really important. Do you want to tell us a little bit about that idea of reverse engineering that you’ve talked about? Lila Gleitman: Well, this is hard. Okay. But remember there was this distributional analysis of — relative distribution. This was not frequency distribution. The relative distribution of items in a linguistic stream. So it was

sort of a discovery procedure. If you started out, I mean if you could imagine a child starting out, your mother is saying blah-blah-blah-blah-blah-blah and you do something like Hangman and discover the morphemes. Then you do something like higher level Hangman. That’s sort of the idea, to build over it. There’s something to that I’ll tell you what’s to that. Everything, every time you look up something in Google, you’re sort of making use of Harrysian ideas – Harris’ original ideas – which have become extremely popular because you can go do a great deal of work So if you try to think about meaning, that’s very, very hard. But if you try just to think about distribution, it turns out that all the words that mean the same thing occur in the same place in the same structures. So the structure of the language imposes a course semantic partition [sounds like] or as fine as you like depending on the units you choose, it imposes a semantic partitioning of the words and phrases and so forth. It does that quite well if you ignore the — I mean there some warts [sounds like], as I tried to say before, but you can do a lot with that Susan Goldin-Meadow: It does it well enough So one of your theories is that children make use of that structure in their language learning Lila Gleitman: Oh, yes, children definitely So back to the child, remembering what the child is hearing is blah-blah-blah-blah-blah-blah-blah, okay? Some time or another, the child has to connect these structures at various levels with what’s going on in the world presumably I mean that’s the first thing you would think So this isn’t just like the next mathematical theory. It connects you to the world. You have to discover that John saw Mary means John saw Mary, and it doesn’t mean Mary saw John and so forth. Well, all of that semantics, the structure arises from that semantics That’s true. But it arises as a very complex organization of that semantics Susan Goldin-Meadow: That the child can take advantage of Lila Gleitman: The child has got to unravel it. I mean how is the child supposed to know that John saw Mary and Mary was seen by John mean the same thing? Even worse, how would you find out even what the words meant? So everybody starts from this intuition that this is easy. Pretty easy. So when a tiger walks by, that’s sort of an occasion in which you’d likely hear the word tiger. So here’s how the child learns the language. You see, it notices that the sound tiger occurs most systematically in the presence of tigers in the world and the least systematically in their absence. That’s wonderful, okay? So now all you need is a theory on each side But it turns out the problem is this. There’s maybe a lot of tigers in the world, but not everything is like tiger. You also have to learn words like think, okay? So let’s consider think for a second. I suppose that people at the center are thinking and the people over there are not thinking. Supposed that’s really true. So I say, “Susan, these guys are thinking. Those guys are not thinking.” She can look earnestly around Susan Goldin-Meadow: Oh, it’s clear they’re not thinking. It’s so clear Lila Gleitman: Yeah. You really can’t tell, okay? So think doesn’t reveal itself in any simple way from observing the world, and it gets worse Susan Goldin-Meadow: So you need the structure of language really to understand which is really — Lila Gleitman: Yeah Susan Goldin-Meadow: So this is an idea that also came out of perhaps what Chomsky taught you. He was your fellow graduate student He also broke with Harris, and he had a big impact on your thinking Lila Gleitman: Well, not only in my thinking He had a big impact on all of your thinking Susan Goldin-Meadow: I know. But you’re the lady of the hour today, so I —

Lila Gleitman: Okay. Yes Susan Goldin-Meadow: So how did Chomsky change the way you started to think about things? In a way he got you — he’s a linguist, but he got you to psychology. So tell us about that Lila Gleitman: Well, something worth reading is his 1965 book Aspects of the Theory of Syntax in which the opening chapter, as many linguistics books, the opening chapter is about psychology. It’s about how people get to use language. So if there are any linguists here, if you read Leonard Bloomfield talking in the structuralist framework, he’ll say the sorts of things that we just vaguely went over. If you only knew exactly the circumstances under which a person heard sentences, you would be able to predict exactly what you’re saying. Later on you could read this in Skinner’s famous, infamous book, Verbal Behavior predicting from the world what you would say. And Chomsky said this is not going to work Susan Goldin-Meadow: So what you introduced — then you came back. You did your dissertation But in fact you sort of did a dissertation in the psychology world, and you did it with Henry Gleitman Lila Gleitman: With Henry Gleitman. Well, yeah Susan Goldin-Meadow: He wasn’t your advisor He was your spouse. So that didn’t really count Lila Gleitman: Yeah Susan Goldin-Meadow: But you did it with I think Henry Heish [phonetic], right? Is he your advisor? Lila Gleitman: I didn’t really have an advisor at that time. I had a real break with my advisor This is not a wonderful position for a graduate student to be in. My mentor, my actual mentors — again, Harris and I weren’t even speaking at this point. So somehow or rather, through the intercession of the saintly Henry [indiscernible], I managed to reinstate myself enough to get a degree. So really he wasn’t a big advisor Susan Goldin-Meadow: Right. And this project was really pretty based in language behavior, right? Lila Gleitman: It was based in language behavior because – Susan Goldin-Meadow: Which was different from Chomsky’s, I think. It’s really different Lila Gleitman: Like any normal American semi-intellectual, I found what Chomsky said in his book, I found one thing after another completely ridiculous So Chomsky was saying, for example — this is where my dissertation came in. He was saying all human beings come to native level knowledge of their language in the natural course of informal exposure and they’re all the same After all, I came from a background in English literature which I continue to admire, and I thought, well, really Shakespeare and I are not of equal competence in English so there are systematic differences here. So in my usual way, and I think many of us would recognize this, I thought I’ll fix you. I will show you on equal competence in the human population, and I sort of devised an experiment in which people were asked to paraphrase sentences like black house bird and black bird house and house bird black and so forth and so on And indeed they varied very extensively in how well they were able to do so Susan Goldin-Meadow: Which was a surprise given — Lila Gleitman: It wasn’t a surprise to me [Cross-talking] Susan Goldin-Meadow: To you, but it was a surprise theoretically Lila Gleitman: — and we’re writing this book, writing this book. Then Henry Gleitman, who was a very smart, educated, thoughtful man kept who saying to me, “You know, you can’t shoot him, this guy Chomsky, with a pop gun,” that there was something wrong with something not thoughtful enough about the way — by the time we got to the end of the book, the important part of it was not that people differ but that you needed the same theory to describe

all of that so that in a deeper sense indeed they’re all the same. Of course, there are always superficial differences. How big is your vocabulary, and blah-blah-blah-blah-blah, and how well can you paraphrase components But what you can do, what a whole population can do such that they can read the ads, so when tan is in this year, right? Susan Goldin-Meadow: So you took advantage of data from adults, but you also started to look at data from children. And you did that with your colleague, with Shipley. And you also simply took advantage of the fact that you had children who were teaching you a little bit about language and learning Lila Gleitman: Yes Susan Goldin-Meadow: So that was another, I think, change in your mind Lila Gleitman: Of course another big thing which settled the rest of my life — because one can enjoy drawing trees, syntactic trees on the blackboard for the rest of your life But then you see a kid learning a language, it’s a miracle. And again, I always maintain this sort of no nothing, very behaviorist orientation. I mean, the world kept saying no every time I tried to show this, but I kept on saying it. So as far as language acquisition is concerned, I don’t think I really believed in it Susan Goldin-Meadow: You didn’t believe that they acquired language. Is that –? Lila Gleitman: That’s right. I mean how could that be possible? They couldn’t do it. So I had this kind of a transformative experience one day. This was in the old days before there were seatbelts and before you made your little kids sit in the back of the car. I had with me my three-year-old. This one is sitting over here actually. We were in the car and we were coming to a curve, and I said to her, “Hold on tight.” We go around the curve and we come out the other side of the curve and this little voice says, “Isn’t it tightly?” I said I can’t believe I’m hearing this Susan Goldin-Meadow: So really, I think your insight was that kids know a lot more that we think they know Lila Gleitman: They know a lot more than they say, yeah Susan Goldin-Meadow: So just tell us a little bit about that study that you did with Liz which was really pretty important. With Liz Shipley Lila Gleitman: Yes. Well, as soon as you get anybody started all about linguistics into whatever the psychology of language acquisition, they will say language learning in young children, you look at the psychological literature at that time. And we’re talking now of the early 60s. And it says boring things, like at age such and such children have a vocabulary this size. Then you look later and it’s bigger and they have graphs, smooth and so forth. Utterly uninformative and in many ways beside the point. So any linguist, anybody with the slightest linguistic training walk into that situation and say yeah. But does what they say exhaust or even properly describe what they know? So linguistics precedes [indiscernible] writing down the sentences that you utter, at least post-Harris it wasn’t just writing down the sentences that you utter but the sentences you judged to be grammatical Susan Goldin-Meadow: But you can’t ask a little kid to judge Lila Gleitman: Well, maybe you can Susan Goldin-Meadow: So what did you do? You did something clever Lila Gleitman: Implicitly you can ask a child, okay? So he has a sentence, Johnny ball throw And then just watch what the kid does. What the kids did, I mean it was astonishing. It was to fall to the ground. There was a great deal of keening. There were kids putting their hands over their ears and running out of the room. The mother was the one saying the stimulus So when we said these weird sentences, we had the mother set a script to offer these sentences grammatical and then ungrammatical And we’re talking about kids who said things like ball throw, okay? I mean they couldn’t say more than two words at a time Susan Goldin-Meadow: But they knew Lila Gleitman: But they would not accept those same two-word sentences from their mother

They show a great deal of rejection, emotionality, and they would not carry out your command But if you said throw me the ball, they just threw you the ball and that was that Susan Goldin-Meadow: So this focus on data is something that is carried. It certainly would influence you. But then you brought that idea over to your students. So let’s talk a little bit about the many people you have influenced and that whole program of looking at language acquisition by coming up with clever studies Lila Gleitman: Yes. Well, I’m not so sure that you and I even today agree to this extent The linguists were doing clever enough studies As Obama said to Hillary, you’re clever enough if you ask people for judgments of grammaticality Because the data aren’t your problem for a long period of time. The data are very easy for people to generate. So if you say to somebody house the is red, they say, no, the house is red. So people have this, to some extent, an ability not only to talk and understand but to comment. They have some access to those tracks. So I think of linguists as having been experimental so long Susan Goldin-Meadow: Right. I’m not arguing, actually, that there is no data in linguistics I’m really not. What I’m arguing, or suggesting, is that what you brought in was a different kind of data that was really helpful in fleshing out these theories. So the work that you did with Barbara Landau and with Elissa Newport, and what we did together was all sort of subsequent to this, and it was generated in part by the need for more different kinds of data Lila Gleitman: Yes, because if you start as we did, as I think any sensible person would, by saying it’s obvious that language is learned on the basis of very specific input data That’s why all the kids in England learn English and all kids in France learn French, so what you learn about language is a function of the input. At that gross level, that’s certainly true. So how refined can you really make a theory in which you can relate as Bloomfield claims if you read the first chapters, that if you can find out exactly what the input was, you can predict what will get learned in the long run So what we set out to do, our group – Elissa Newport, you, Barbara Landau – is to ask ourselves if you change the input to the child, if you look at different kinds of input, can you predict exactly how language acquisition will grow in that population. And over and over again, as we did this in several studies, God kept saying, no, you’re wrong again. Okay? Because when they changed the input first with Newport, just looking at people in the normal range, the mothers in the normal range, mothers and fathers and the growth of the children in some very clever kinds of analyses, we couldn’t predict much. A little bit but negligible about the growth of the language in that particular child by properties that we could identify in the input. So we wrote a paper which was based on an advertisement at the time. The paper was called, Mother, I’d Rather Do It Myself, which was an Excedrin ad at that time Susan Goldin-Meadow: So all of these studies really speak to that first question that you’re very interested in, which is what constraints the child Lila Gleitman: Right Susan Goldin-Meadow: But then at Penn, you were at Penn at this point, you hired John Truswell which allowed you to get into another whole set of studies and to start thinking about how the child takes those constraints and learns a language like English. So can you talk to us a little bit about that? Lila Gleitman: You skipped over 30 years because — Susan Goldin-Meadow: I did, I know. But that’s because we don’t have that much time left Lila Gleitman: Okay. But these input studies

became more extreme, right? So now we can refer to really interesting, serious people, like David Hume. So if you were blind, you certainly couldn’t understand words like look and see. So Barbara Landau and I changed the circumstances of learning and you can predict what the child said in just one sentence The astonishing finding was the first verb in the blind children’s vocabulary was see, s-e-e, with look not far behind. I mean this begins to have an effect on your behaviorist leanings here, on your idea that there’s something simple about the input and output. And, of course, with Susan Goldin-Meadow and Heidi Feldman, we then began to look at, well, what if you didn’t hear the sentence but you could interpret the word So by varying these inputs and outputs — so we were looking for Harris’ discover procedure, but instead we found much to support a kind of Chomsky or should I say platonic – to be fair – outcome that kept confronting us. It was that the child rose above the limitations of the input and they all construct a grammar and a semantics of essentially the same kind if they’re humans of normal mentality Susan Goldin-Meadow: So these are great ways of discovering the constraints that children bring to language learning, but of course a child has to learn the language to which that child is exposed. And that was another whole set of studies that you really made major contributions to Lila Gleitman: I don’t know if major, but I fell in with a young psycholinguistic. I was getting old by that time. A young psycholinguist whose name was John Truswell who had been very involved with sentence processing and so forth, how you understand the sentence in the moment and produce issues of production and comprehension which are foremost in many psychologists’ mind largely separate from the issue of acquisition. So we were in different places. We weren’t theoretical opponents or anything, but we seemed to be interested in different things But actually, after all these structures, semantics and relative indifference to input comes about, you do learn the meaning of the word “tiger” by noticing that there’s a tiger walking by. And that, simplest of all the problems, for 30 years I had never thought about. And Truswell and I forged a collaboration which I think was very good because it drew us both back into this nexus of all of these problems in how the child moves from an initial state to knowledge of the language. So that’s what we’re still doing We’ve been working together for 20 years now It’s another class of problems Susan Goldin-Meadow: Right. So you were there at the very start of this revolution, this cognitive revolution which sort of made it It got its place when you and others at the University of Pennsylvania created IRCS. That really put the cognitive revolutions sort of on the map. So tell us a little bit about the state that the field was in at that point Lila Gleitman: Okay. Well, again, you want to be careful because unlike Gore was it, I didn’t invent the cognitive revolution Susan Goldin-Meadow: You didn’t? Lila Gleitman: No Susan Goldin-Meadow: I’m disappointed Lila Gleitman: But actually when I arrived at Penn, it was a very wonderful time. First of all, there was this visionary and computational person, Zellig Harris. And Penn people, Eckert

and Mauchly, had been material people in making the first viable computers. They were stolen of course immediately by the Rand Corporation which gave Penn ENIAC and then UNIVAC. And Harris foresaw computational linguistics, so how you could formalize and implement a theory of language of the kind that he had So I walked in on this and I had the background, a background I deeply respect but very different of a person who read novels and a little poetry here and there. And he said, well, so here you are and you’re going to implement something about English on a computer. And I said, what’s a computer? I never heard of a computer. In common with most people at the time, I had never heard of it. We’re really talking 1960 And he said, “Don’t worry, there’s a young engineer in the engineering school at Penn and he has agreed to teach you how to use a computer.” The guy’s name was Aravind Joshi. I hope you know him. Maybe he’s the dean of computational linguistics then and now, but at the time he didn’t know anything about linguistics. This guy was supposed to teach somebody how to use a computer, and what the computer was to do is to analyze English sentences. I looked at the computer and said I never saw anything like that. He looked at me and he said the computer wasn’t designed to do English sentences; it’s designed to do numbers. But we were very young people and we began to see it right away and, in another sense, we worked together for the rest of our lives Susan Goldin-Meadow: And the two you of you formed — ? Lila Gleitman: Formed the Institute for Research and Cognitive Science which began to bring together in our shop — and again I say this isn’t the only place it happened. It happened in maybe a dozen or 20 universities where people from different departments were realizing that they had been working on the same issues all the time and they were first getting together So when we got together with people in robotics and people in philosophy and linguist psychologists and so forth for a couple of years, eventually this institute started with big pushes from first the Sloan Foundation and then the National Science Foundation Susan Goldin-Meadow: And that’s cognitive science. That’s today’s cognitive science Lila Gleitman: And that’s cognitive science, yeah Susan Goldin-Meadow: Do you want to say a word about where you think cognitive science is now? Lila Gleitman: Well, it’s certainly a lot more enlightened now I supposed but it always looks — everything looks rosy about today Where is cognitive science? Well, I don’t know. You know the era I’m talking about was a very optimistic era when all these fields were coming together in very fruitful ways, and I think very fruitful things happened I mean think about like Folger. I mean they were coming from all kinds of backgrounds And maybe the apogee here was this book, Aspects of the Theory of Syntax by Chomsky, where linguistics was really making big progress in this interdisciplinary way But you know what happens in science? You look at the details and the theories never work so they begin to fall apart together And I think we’re in sort of a powerless period now. We have those big data people. I say those. I should say these to be a little more neutral. You have big data people who were saying, you know, we could beat this problem to death. If we only give the kid let’s say 500,000 form to meaning pairings for tiger, I bet the kid will learn tiger. I say but

the kid learned it in one trial, so what does big data got to do with it? I mean I’m sure there were really three huge things. Neuroscience says, oh, we ought to do this stuff by inferencing. Or just open up the brain and look inside and you see the real theory. So this is neuroscience and it’s developing in its way, and those big data which is developing theories about the same thing but quite in a different way, and then linguistic psychology which is a tradition which continues where it was before. These things aren’t measuring yet at the moment, but of course they must mesh and they will But this is not a great movement. IRCS by the way, Institute for Research in Cognitive Science, closed last year. Finished Susan Goldin-Meadow: So you were there. You began it. It is not over even though IRCS in some ways is cognitive science. That revolution is taken for granted now. It’s sort of maybe it doesn’t need to be an institute because it’s everywhere. And maybe not. But do you have, after looking back at this whole wonderful career that you’ve had, any advice for young people just starting out? Lila Gleitman: Yes. Don’t do what your teacher did. Everybody is wrong. The words will set you free. It took me a long time to learn and a lot of rules [sounds like] and misdirection But what really doesn’t pay is to do the next experiment that’s on your lovely mentor’s agenda. That usually won’t do much Susan Goldin-Meadow: Okay. Well, that’s excellent advice for all the young people here, to defy their advisors. That seems like just what we all — Lila Gleitman: At least graduate school Susan Goldin-Meadow: Just what we advisors wanted to hear. Thank you, Lila. That has been a wonderful tour through your life Lila Gleitman: Thank you