Celebrating 25 years of research into language and cognition

On December 1, the Max Planck Institute for Psycholinguistics celebrated the work of the Language & Cognition Department and said farewell to Levinson, who retired at the end of 2017. Levinson joined the Max Planck Institute for Psycholinguistics in 1991 as director, where he spent 25 years of a much longer career probing essential questions about language diversity. In 2018, he will move from Nijmegen to Cambridge, England, where he has an adjunct position at Cambridge University. “Leaving here will be full of regrets,” he said, “because I just love working with young scientists. They're so enthusiastic, they have such good ideas, they're so much better trained than I was and their skills are amazing. That's what I'm going to really miss.” 

The following is a lightly edited excerpt of the interview.

Michael Erard (ME): In your obituary of sociolinguist John Gumperz, one of your academic supervisors at UC Berkeley, you wrote how he “encouraged students to find their own way, giving them a tape recorder, encouraging exploratory fieldwork, showing them how to transcribe and analyze, and receiving half-baked ideas with an enthusiasm that transformed them.” Were you describing your experience, and do you remember any of those half-baked ideas?

Steve Levinson (SL): Of mine?

ME: Yeah.

SL: Well, one of them is one of our most cited things. It's almost embarrassing because it was written when we were graduate students, a paper that my wife Penny [Brown] and I wrote on politeness. When we first thought of it, the idea was this: if you just transcribe conversations and things, you realize, wait a minute, actually what's going on isn't what's on the page. It's much more oblique. You come to realize that human affairs are dealt with largely in an oblique manner. Of course, that runs against Grice's cooperative principle [“Make your contribution such as it is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged” (Grice, 1989)—ed.], the idea that for successful communication, people need to be as direct and clear as possible. We said, "Wait a minute, so why aren't people being efficient?"

We had the idea that people aren't really efficient because they treat each other with kid gloves. They're very worried about offense and they go out of their way to not cause offense, and that is one of the main reasons for the obliqueness in the conversation. We identified two kinds of politeness, one where you magnify the social distance between each other, a sort of formal politeness. Then the other kind where you claim very close solidarity, and that’s what allows me to do these things with you that I wouldn't otherwise.

Then we went and applied this cross culturally and showed that actually it works. I was working in Tamil Nadu, doing social dialectology [the study of variation in language as a product of social grouping or affiliation—ed.] in a Tamil village, and Penny was working in Mexico on a Mayan language. We compared our notes on politeness and said, “Look, it works in these two unrelated cultures.”

It's one of those funny things, because often the reason that one writes things is because you try out an idea and someone hugely disapproves of it and thinks it's awful. Then you think, damn it, it's right! Then you're really motivated to work it out and present it. Our first presentation of those ideas got totally panned.

ME: What other examples of that do you have?

SL: Another example is our work on spatial cognition. I was working on an Australian Aboriginal language spoken in Cape York, and Penny was still working in Mexico. Again, we realized: wait a minute, these guys don't know their lefts and their rights, they don't talk in terms of left and right, they don't have any rituals where left and right matters. They are completely oriented by the compass, as if they had a mental compass. Then we thought, what are the implications of that?

Language and Cognition fieldwork and language documentation across the world

When you start to work it out, the implications are really huge. For example, actually we discovered that people's gestures, if you look at their gestures, they're not a veridical for left and right, they're veridical for north, south, east and west. It's a completely different system. If you look at all the literature on spatial cognition, there's this very strong assumption in the Western literature that all basic cognition is egocentric. So it couldn’t be the case that primary spatial cognition would be based on north, south, east and west.

We did a whole bunch of experiments that seemed to support this. We did memory experiments, we did inference experiments. We were able to show that in these societies, people really seemed to do all their memory and spatial calculations based on a non-egocentric system.

ME: Was there push back on that idea?

SL: There was horror. That was the early 1990s, when nativism [the idea that universal language abilities or structures are hard-wired in a uniquely human language faculty—ed.] ruled the roost in the cognitive sciences. It's hard now actually to imagine what it was like, but as far as languages were concerned, people thought they were all identical except they just happened to have different sound systems, essentially. Quite different order in the surface elements, but essentially they were all the same. And every basic language was a mere expression of a fixed cognition. The idea that the underlying cognition may vary, possibly as a function of the language, was absolutely unorthodox. It was a real uphill battle -- people were appalled by it.

As I was coming from anthropology, the resistance seemed bizarre. We thought the whole point of culture was as an adaptive mechanism. That’s how come humans are so successful because we can adapt to any environment. So of course we'll adapt our thoughts and our language, too. I like to think about language as a bio-cultural hybrid, not reducible in either direction.

I like to think that over 25 years, we did nudge the cognitive sciences ever so slightly more into an open-minded state of mind on the importance of human variation. We were helped by the rise of parallel computing and connectionism, where a lot of the stuff that people thought could only be done by very specialized mechanisms built into our psyche could actually be done by a very general purpose machine.

In a way that's been my mission, trying to get cultural and linguistic diversity taken seriously by the cognitive sciences.

ME: How does the genetics of language fit into this?

SL: The interesting thing about that is you need variation in order to say anything. The variation is the lever that gives you insight into what the genes are actually doing. It's only by finding someone who's got a slightly different version of the gene and then looking at what the effect might be, that you have any insight at all. Variation is the very stuff of it, the heart of it.

ME: You were always interested in cultural variation or interactional variation, but the genetics is a different kind of variation. Do you see anything changing?

SL: I think it's an interesting part of the same picture, actually. What culture does is turn a bunch of individuals with all their individual variation into pseudo clones. Its function is to make us somehow more uniform that we actually are, so we speak the same language, we know the rules so that we're polite to each other, we drive on the right side of the road. We're turned into little behavioural clones by culture, but underneath of course we're all slightly different. I think the internal variation of a population is actually much greater than we realize because of this cultural uniformity. Even the algorithms that we each use to produce the similar result may be realtively idiosyncratic.

Again, that would still be thought to be radically heretical in cognitive science: To think that the uniformity you do see is a cultural artefact.

ME: What do you think is possible now in linguistics that wasn’t possible in the 1970s?

SL: In the 1970s, essentially the idea was that the core of language was a uniform mechanism. Languages were free to vary in their use more than in their structure, so that made social linguistics and linguistic anthropologists, okay, we can talk about cultural differences. Curiously, now the picture I discern is inverted, such that language structure is hugely diverse across languages but the usage characteristics are much more uniform than the linguistic anthropologists had thought. We've done a lot of work on this pragmatic infrastructure for language because we think that's what makes the diversity possible.

Levinson does a sound check in preparation for recording conversation in Tzeltal. © Elisabeth Norcliffe, Max Planck Institute for Psycholinguistics

For instance, if you go on holiday to Morocco or wherever, you're stumped for words but you're not stumped for the idea of a question, and nor is the other guy surprised that you would be asking a question about the price. He knows what a question is, you just don't know the same words for it. The idea of questions and answers is part of a universal substrate. It's obvious, really, but the richness of that substrate is much underestimated. That's what makes it possible for infants to bootstrap up into the local language, whatever it is and however complicated or diverse it may be. That's the inversion of the 1970s picture.

I think even things like our politeness phenomena, that's part of a universal system. When you go on holiday to Morocco and you think, oh I'm not going to offend this person, be careful when I question his price for his carpet. There’s also repair, turn taking, and the sequential organization of speech, things like question, answers, requests. It's actually pretty interesting that you can have a perfectly good dialogue that goes question 1, question 2, answer 2, answer 1. I say, "Are you coming to the party?"

You say, "Which party?"

Then I say, "My party."

"Oh yeah."

You can actually get that embedding deeper than you ever find it in linguistic structure, which is a little bit telling, because people say, "Oh, we can't do these embeddings beyond two deep in linguistic structure because of memory limitations.” But lo and behold you get these embeddings in discourse, six, seven, or eight deep. Again, it suggests that there is something special about this interactional system and its organization.

We looked at a dozen unrelated languages and looked at their system of question, answer and how you could build questions within questions and so on, and we found that all of the languages we looked at had the same properties.

ME: At some point in the last few years, the Directors have gone on research retreats, and people were asked to come up with a dream research project. What was yours?

SL: My idea was that I would have a shipping container, inside of which there would be all sorts of wonderful machines like imaging machines and eye trackers and I guess we'd need a little gene processing machine or two. We would have already done a survey of all the world’s languages, and we would know what all the outlier languages were. One would then say, let's have a look at all of these outliers and actually how are they processed. How do kids learn them? For example, there's not a single processing study of a free word order language. We don't know how they work -- we haven't a clue.

Like some of these Australian languages -- you could have some sort of sentence, it's totally jumbled up. You have whatever it would be, “the yellow butterfly is being eaten by the crocodile.” “Yellow” would be at one end of the sentence, “butterfly” at the other. “Crocodile” in the middle, between two bits of the verb, which are exploded completely everywhere. How do you process this? You can sit with a pen and paper and you can work it out, because these are case-marking languages, so this bit goes with this bit. But how do you do it as it's coming in, in an acoustic stream? We have no idea how it works.

So my dream study would be an attempt to really understand the variance in human languages. We’d ship this equipment around as a kind of lab that would drop in on all of these outlier languages, and we would try and figure out, are their genes different? Oh, no, their genes are the same, that's out. Now, do their brains seem to be using the same areas as our brain? Ah okay maybe, maybe not. Maybe it's more right hemisphere -- and so on.

After all, this is our central mission at the institute: what is it that makes language work, and how is it that minds process languages? So you've also got this huge range of different languages. How do similar but different minds process all this diversity?

Further reading

• Brown, P., & Levinson, S. C. (1987). Politeness: Some universals in language usage. Cambridge University Press. more >
• Levinson, S. C., & Brown, P. (1994). Immanuel Kant among the Tenejapans: Anthropology as empirical philosophy. Ethos, 22(1), 3-41. Retrieved from http://www.jstor.org/stable/640467. more >
• Majid, A., Bowerman, M., Kita, S., Haun, D. B. M., & Levinson, S. C. (2004). Can language restructure cognition? The case for space. Trends in Cognitive Sciences, 8(3), 108-114. doi:10.1016/j.tics.2004.01.003. more >
• Evans, N., & Levinson, S. C. (2009). The myth of language universals: Language diversity and its importance for cognitive science. Behavioral and Brain Sciences, 32(5), 429-492. doi:10.1017/S0140525X0999094X. more >
• Levinson, S. C. (2012). The original sin of cognitive science. Topics in Cognitive Science, 4, 396-403. doi:10.1111/j.1756-8765.2012.01195.x. more >
• Levinson, S. C. (2006). On the human "interaction engine". In N. J. Enfield, & S. C. Levinson (Eds.), Roots of human sociality: Culture, cognition and interaction (pp. 39-69). Oxford: Berg. more >
• Stivers, T., Enfield, N. J., Brown, P., Englert, C., Hayashi, M., Heinemann, T., Hoymann, G., Rossano, F., De Ruiter, J. P., Yoon, K.-E., & Levinson, S. C. (2009). Universals and cultural variation in turn-taking in conversation. Proceedings of the National Academy of Sciences of the United States of America, 106 (26), 10587-10592. doi:10.1073/pnas.0903616106. more >