George Lakoff

Richard and Rhoda Goldman Distinguished Professor of Cognitive Science and Linguistics at the University of California at Berkeley, the American linguist George Lakoff (New Jersey, 1941) is a very busy scholar, constantly sought by graduate students, and other colleagues in the University of Berkeley, where he has been teaching since 1972. Professor Lakoff’s current research covers many areas of conceptual analysis within cognitive linguistics, such as the nature of human conceptual systems, particularly metaphor systems for concepts such as time, events, causation, emotions, morality, the self, and politics.

His eight square-meter office in Berkeley is not big or luxurious. The walls are surrounded by high shelves all fully packed with books and papers. As I walk in, Professor Lakoff invites me to sit in a cozy rocking chair in front of his desk. In his classes, at his office or in a colloquium meeting, he always brings a coffee mug from which he constantly sips water. Instead of a structured interview with the questions agreed in advance, Professor Lakoff insisted on having a spontaneous conversation flowing freely between different topics from the past, present and future.

I would like to ask you about some of your work(s). There is an article called «Humanistic linguistics» (1974). I was struck by the word humanistic. Does this have something to do with a sort of «intellectual illumination» on your part? Why did you call it «humanistic»?
Well, in first place because of my background. I was in MIT and I studied both mathematics and English literature, especially poetry. The way I got into linguistics was that my last year at MIT, which was 1961-62, was the first year of Noam Chomsky’s Department, and in honor of the founding of the Department, Morris Halle had Roman Jakobson invited to teach a course. Jakobson was in Harvard, and he decided to teach a course in poetics. So my advisor in the English Department said: «Roman Jakobson is coming to teach poetics, you’re interested in poetry, you should take this course, but if you’re going to do it, you should know all your linguistics, so also take Morris Halle’s Introduction to Linguistics». So I wound up in the same class with all of Chomsky’s first year graduate students, but unlike them I was taking Jakobson’s poetics course, none of them took Jakobson’s poetics course. And at the same time, I learned his technique of analysis and I was learning phonology. I studied phonological and grammatical analysis of poetry, and William B. Yeats, who was his favorite English language poet.

So how was your first experience as a teaching professor?
My first lecture in the university was substituting for Roman Jakobson in his course, because I did a pretty extended Jakobsonian analysis of Yeats’s Byzantium, and at one time he said: «I will not be in class next week, I must go to Cornell to lecture. Would you like to lecture in my class on your analysis of Yeats?» Ok, so I was twenty years old and I was lecturing to all advanced graduate students from Harvard, and that was fun, but the point is that it was always humanistic. What Jakobson taught – remember that he was the person who taught Levi Strauss, so my teacher and Levi Strauss’s teacher was the same person, surprisingly enough – should be applied to poetry and to anthropology…, that was the way I saw it. Generative linguistics wasn’t like that at all, right? Then, in 1974 I was already at Berkeley, and it was clear to me that Generative Linguistics wasn’t going to work, and one had to look at humanistic discourses. I was teaching stuff on poetry and pragmatics. So this was not unusual at all, and not new.

Photo: Esther Enjuto Castellanos.

I understand that you then started to work on the semantics of grammar. I suppose this work was first published in Metaphors we live by (1980). Is this perhaps your first seminal work on cognitive linguistics?
There was one before, in 1977, in the Chicago Linguistics Society, a volume, and it was a paper that outlined all these same things, but this was before Metaphors, it was talking about frames, it was talking about what grammar had to be, that grammar had to take experience into account, how it was embodied… but Metaphors we live by was really the first major work on that. However, the earlier work on embodiment came in 1975, not from me but from two other people. So, there were lectures here in 1975 by Paul Kay on the neural account theory of color, where color depended on the physiology of the eye and the brain, as well as the world. Then there was Len Talmy’s work on image schemas and there was Eleanor Rosch on basic level categories, where they were defined by mental imagery, motor programs and Gestalt perception, and that was all embodied. Then Charles Fillmore’s frame semantics was following up what he had done on case grammar, and there in case grammar and later on in frame semantics, he said that grammar is based on basic experiences that you have.

Was all this work essential for your decisive departing from Chomskian theories?
Yes, what I learned about that, I had to give up my previous work on logic, as we were trying to make logic work for linguistics or for generative grammar, and it was just clear then it was not going to work, and I had a great crisis. There were a couple of years when I wrote nothing. I used to write a dozen papers a year and then I just could not write anything, and then, I wrote that 1977 paper, and then when I met Mark Johnson, we started working on all the stuff about embodiment and embodied metaphors.

I see that you have been working with other European cognitive linguists, such as Zoltán Kövecses, in «The role of metaphor and analogy in representing knowledge of presupposed worlds. The cognitive model of anger inherent in American English» (1987).
The work with Kövecses was interesting. When he was doing an English idiom dictionary, right after he read Metaphors We Lived By, he was under the letter «a» and he hit «anger». He collected four hundred idioms for «anger» and he said: «This must be something to do with metaphor», and he applied for a grant, he came here, and he walked into my office, just as you have, and he stood there. He was six foot three and two hundred and thirty pounds, all muscle, and he says: «I want to study anger», and I say: «OK, ha, ha, ha… let’s go». And it was true that he had a lot of examples, but he did not know much about linguistics, and he did not know much about metaphor but he was here to learn, and he learned how to do it, and being a world-class athlete, he was pretty disciplined. So we worked from nine in the morning right through, until we wrote that paper. But the big thing about it was that, since it was all about embodiment, and I knew Paul Ekman, who was the world expert on embodiment of emotions, we started talking to him, and we noticed that the metaphors were all based on the physiology of anger. And then it took us nine months to figure out how to do four hundred metaphors with this, and it was very interesting to do the work. So a lot of the later metaphor theory, but also the work on frames and categories, was the same.

«Now people used to call us armchair linguists, but we were gathering vast amounts of data, huge numbers»

I assume that your work More than cool reason (1989) has to do with literary metaphor. Is there any difference between these poetic metaphors and the ones we just spoke about? Are you still interested in researching literary metaphors? 
Of course and I work on it, I give talks on it, and I work with graduate students on it, but I’ve been working on political versions of this, which are like literary metaphors in many senses. But let me tell you about that book. The first graduate student in Cognitive Science was a joined honors graduate student in the English and Cognitive Science Departments, named Mark Turner. Now Mark Turner was an interesting person, and he is still a very interesting person. Before he came here he had studied, as an undergraduate, Latin, Greek and classical literature. Then he came to Berkeley and he kept on studying classical languages and poetry, but he also studied mathematical biology, and then from mathematical biology he got a Masters degree in Computer Science while still studying classical literature.

Photo: Esther Enjuto Castellanos.

So how was your collaboration with Mark Turner?
One day Mark Turner came into my office with a lot of boxes, and the boxes are filled with papers with lists of things, and he says: «Well you know I am interested in metaphors but I have this computational part of me, and I have this list of cases where I took all words of the family: brother, mother, father, sister, cousin, uncle, everything I could find, and I went through the major authors, Milton and Shakespeare, and they are all in these boxes. I want to do a dissertation on this». I replied: «What are you doing with all this?», and he says: «That is the problem, no one in the English Department understands this». So I agreed to be his unofficial thesis advisor and I worked with Turner, and he wrote seven drafts of his thesis. He kept revising and revising it, and it became a very nice book, and he got a job at the University of Chicago. After that, we decided to do this other book on poetics, poetry. And that is where it came from.

In 1995 with «The neurocognitive self: Conceptual system research in the 21st century and the rethinking of what a person is», you seem to depart from the linguistic proper study of metaphors to the more «scientific» approach to neuroscience. 
Well, first of all, cognitive science is a scientific approach, and even the way we did generative semantics was a science. We did it as a scientific approach, with our data. Now people used to call us armchair linguists, but we were gathering vast amounts of data, huge numbers, it was not two or three sentences, huge numbers of things. We were thinking about these all the time, and noticing things, and every time we found there was a counter example, we had to figure it out. So we had very much a theoretical approach and an empirical approach, but it was not the kind of empirical approach that is done by machines or something. So it was always, I always saw myself as a scientist, but a scientist who was studying something that would apply, as Jakobson would say, to all humanistic and social issues. From the beginning, that was not new, but then what happened with neuroscience was interesting.

And what happened to Neuroscience at that time?
As I mentioned, Paul Kay published a paper around 1978 on language related to color. Taking the work of Russell De Valois, who was in the Psychology department here, which showed that color is not in the external world, color is a matter of wavelengths and reflectances. But wavelengths are not colors, they have to go through the retina and then connect with the brain. It is from the brain circuitry and the retina that you get color. Now that turns out to be different for men and women and it is a very complicated story, but it is not external, it is interactional, it is an interactional issue. I knew from 1975 on, when neuroscience began its story, that I would have to worry about it.

«I knew from 1975 on, when neuroscience began its story, that I would have to worry about it»

Then you started working on computational neural models. 
I had a friend at the University of California, San Diego, who I worked with on grammar. He was in the Psychology Department and then he figured out how to do neural work computation, and he started the whole theory of artificial neural networks, a parallel distributed processing. He was an old friend. My brother at the time was the Chair of the Political Science Department at UC San Diego, so I used to visit him. When I visited him I went over to see Dave, and Dave gave me a tutorial on how to do this, he was very kind. I had previously given him tutorials on the structure of narratives and when I was doing that I gave him my own papers, my thesis, my previous work on this and he did a lot of historic grammar work on this. Then he moved on to this, and then started teaching me.

And here comes the parallel distributed processing, which means…
Yes, I started learning parallel distributed processing and how that worked. I tried to figure it out. The assumption was that it was how the brain worked, and so I learned it, but it was highly technical. So I started studying mathematics, I went on learning the mathematics, and I even got to the point of teaching a course at the summer school about all this. But it became clear as I went along that this wasn not going to fit. Metaphor is not going to fit grammar and it was not going to make it, but along the way I ran into a work by Jerry Feldman, On structured connectionism (1982), and I decided his idea was probably more interesting than the other I had been working on up to that time. Then in 1988 he came to Berkeley and started the International Computer Science Institute. We started a project on the neural theory of language, and we took the structure approach rather than parallel distributed processing approach.

Photo: Esther Enjuto Castellanos.

How did you get to study metaphors from a more «empirical» approach, such as the neural models? 
In 1992 Srini Narayanan came to our project and changed everything. He was in India, he knew his mathematics stuff, and he figured out how you have a neural model of motor control, because there was another graduate student, named David Bailey, who was trying to model hand movements, looking at verbs of hand movements in various languages. And to do that, he needed a model of the body. There was an online computer project in the University of Pennsylvania studying car crashes. They had a model of every bone and muscle, but it did not move. They said: «Oh! You can have our model, but you have to get it to move», and David did not know how to get it to move. Srini had motor control models for all of these, and he returned to our group to give a report and he said: «There is something strange about my findings. There is a hierarchy of structures with the highest level, and then there are various bindings to the lower levels, fine details of how the body works». And now, given that, I said: «Which is interesting because it suggests that we understand the structure of actions and events in terms of what our bodies can do even if they are abstract actions and events». I knew at the time, having studied logic, that there was no logic of aspect. No logician had figured it out. So he said: «Ok, I will work on that in my thesis», and he added: «Abstract ideas are basically metaphorical ideas». What he did was construct a neural theory of metaphor.

You wrote along with Rafael Núñez Where mathematics comes from: How the embodied mind brings mathematics into being (2000), and also your subsequent article «The cognitive foundations of mathematics: The role of conceptual metaphor» (2005). Do you think that mathematics has something to do with metaphors and cognitive linguistics?
Oh, absolutely. I will tell you how this happened. In 1992 I had a student in my metaphor class, named Min Min Chung, who is now a professor at the University of Illinois, but he was studying mathematics education. Very, very bright, very sweet, nice man. He wanted to know how you could understand arithmetic, and how you could understand multiplication by negative numbers. What is it and how could you even understand what a negative number is? What he did was write a paper that began to do that and in his dissertation the following year he wrote a chapter on a lot of the details of the metaphors for this. For example, using a word like add, is like «putting something in a container», and then subtracting, is «taking away». So, for instance, you have four, you take away three, you get one, right? You have the notion of something being in the container or not, or taking steps in a certain direction and then turning around in the opposite direction. You start at zero, and then when you pass zero you are into the negative numbers.

I heard you then decided to give a seminar so you might discuss all these mathematical issues with your colleagues and the postgraduate students.
I realized that when I was at MIT – I was double majoring in mathematics and literature – many of the things I learned in mathematics were things I did not believe. They would prove them, for instance, the number 0.99999 equals to 1. I could prove it, they showed me the proof, but I just didn’t believe it. There are all kinds of proofs that I simply did not believe, like the proof in logic, there are many things that did not make any sense to me, but you could prove them. I simply could not believe them. The assumption was anything you can prove it must be true whether or not you can make any sense of it. And I said: «There is something wrong in here, something is not right».

So you really think there are some obscure concepts in mathematics which very few people if any can understand even if you see the mathematical demonstration?
Look, we [Rafael Núñez and I] started this seminar by asking the mathematicians: «Are there things you do not believe?» Every single person had a long list of things they did not understand or things they did not believe. The math professor had things he did not understand or believe. What does it mean to have something that is 2.3 dimensional, what does that mean? And if you have an infinity, you can have two infinities, infinities that go on and on and on, and there are the infinities that are things, like the set of all the real numbers that go on and on, for instance. Pi is a thing, but is an infinite decimal that goes on and on and on, but is a thing, how can that be? So we started trying to understand the things the people did not believe. And we could have written three more books on this, we had so much material… We started with six hundred pages but we could have written another thousand pages on this.

This interview was made possible thanks to the postdoctoral aid granted by the Office of the Vice-Principal for Research and Science Policy of the University of Valencia within the «Talent Attraction» subprogram, within the framework agreement signed by the University of California, Berkeley, Andalucía TECH and VLC/CAMPUS (Universitat Politècnica de València, University of Valencia and CSIC).