Stephen C. Levinson -
(a) In the Categories project we look for patterns ('semantic typology') underlying semantic diversity, and explore the correlates in non-linguistic cognition ('linguistic relativity'), for example in the perceptual and spatial domain. more >
(b) In the Interactional Foundations of Language project we explore the foundations for language use, on the hypothesis that a strong universal pragmatic base forms the infrastructure for language in all its diversity, allowing the child to bootstrap into the local language. This explains the species-unique capacity for language. more >
- My current research perspective
- Six strands of my work, which all exploit cross-linguistic variation as a source of insight into the relation of language to cognition:
- Language under research: Yélî Dnye, language of Rossel Island, Papua New Guinea (click here for more information about the Yélî Dnye fieldsite)
I believe the language sciences urgently need a change of paradigm. As more data comes in, language diversity is looking more and more unconstrained by strong language universals, and more and more a matter of cultural evolution (see Evans & Levinson BBS, 2009). Application of bioinformatic techniques reinforces this view, showing that languages retain phylogenetic signals over deep time, better than any other signal including even human genes. Languages vary not only in almost every aspect of form, but also in the concepts encoded in the lexicon (see Categories project at the institute). Our work suggests that these differences can make a real difference at other cognitive levels, for example how visual arrays are memorized.
If language diversity is largely driven by principles of cultural evolution, how are we to explain that humans alone master language, infants seem specially equipped to learn it, there are biological foundations for language, and so on? At least a part of the explanation is that there is a strong universal base not in language proper, but in the pragmatic foundations that support it (what I have called the 'interaction engine' in Levinson 2006). These include ethological features like multimodal signal exchange (gesture, gaze and vocalization), turn-taking, and universal functions (denying, requesting, greeting, etc.) and universal inferential systems for attributing communicative intents. Infants use this universal system to bootstrap into the local language. These issues are studied in the Interactional Foundations of Language project at the institute.
These lines of work converge with new results from cognitive neuroscience and human genetics to suggest a new Darwinian paradigm in the language sciences, focussing on human biological and cultural variation, and how this variation is on the one hand amplified to yield linguistic diversity, and on the other hand dampened to produce relative linguistic uniformity within a speech community.
I co-edited (with Gerald Gazdar) the first journal and wrote the first comprehensive book in this field (Pragmatics, CUP 1983, transl. Japanese 1990, German 1990, Spanish 1991, Chinese 2001, Portugese 2007). Numerous research papers followed, suggesting that many patterns in both syntax and meaning may have pragmatic origins. The monograph Presumptive Meanings (MIT 2000, transl. Spanish 2004, Japanese 2007) points to very detailed, stable and universal principles of preferred interpretation or language construal. Current work is focussed on verbal interaction.
With Penelope Brown I published a foundational book in this field, Politeness: Universals in language usage, and have published 22 papers on verbal interaction in the last decade. I have done extensive field work (with P. Brown) on Tzeltal Mayan Indians and Rossel Islanders (Papua New Guinea), comparing and contrasting interactional patterns (including between infants/care-givers). I have published on e.g. gaze, repair, person-reference, and kinship contexts in interaction across these societies. I have also tried to foster interdisciplinary collaboration in this field, between anthropology, linguistics and cognitive science (see Enfield & Levinson, 2006, Roots of Human Sociality: Culture, Cognition & Interaction, Berg). Current work in the Interactional Foundations of Language project pursues these themes.
With colleagues, I have developed the methods and authored some of the first papers in this field (e.g. Levinson, S.C. & Meira, S. 2003, Language, 79(3): 485-516) and coordinated the first large-scale systematic study, Grammars of Space (CUP, 2006). In the Categories Across Language and Cognition project at the MPI for Psycholinguistics, we investigate the role of language categories in cognition from a developmental and cross-cultural perspective (q.v. Bowerman & Levinson 2001, Language acquisition and conceptual development, CUP). We produce regular Field Manuals for stimulus-based comparative semantics in different field sites around the world which have provided this field with its first firm empirical foundations. Current systematic work in the Categories project is illuminating the language of perception (colour, taste, smell, sound, touch), and has shown striking mismatches in e.g. the ontology underlying toponyms and landscape terms (ideas picked up in Cognitive Geography).
I introduced some of the first systematic cross-cultural data into this field (Levinson 1996, 'Frames of Reference', in P. Bloom et al, Language and Space, MIT), summarizing a large cross-cultural study in the 2003 monograph Space in Language & Cognition (Chinese trans. 2008). Papers have appeared in Cognition, PNAS, TICS, Current Biology, Language (*3). This research has found substantial linguistic and cognitive diversity in this domain, unimagined by most cognitive scientists, and has therefore reopened discussion of 'linguistic relativity' (see e.g. Levinson et al. vs. Li et al. in Cognition 2002). With Gabriele Janzen, we have new intriguing fMRI results showing clearly differentiated neural circuitry for distinct frames of reference.
I have contributed to this rapidly changing field by organizing two major international conferences, by co-editing two significant volumes (Evolution & Culture, MIT 2006; Roots of Human Sociality, Berg 2006), by pioneering with colleagues the application of bioinformatics to language typology (papers in Science 2006, Language 2008), and by exploring the relation between our phylogenetically-inherited spatial cognition (via the experimental study of all the great apes) and human cultural variants (papers in PNAS, Current Biology). I believe that a co-evolutionary (twin-track, culture-biology) perspective should replace the simple nativist paradigm dominant in the cognitive sciences (see Evans & Levinson in BBS, 2009). I have a collaborative project with Mike Tomasello (funded by MPG, headed by Daniel Haun) putting together data from developmental, cross-linguistic, and cross-primate research in core cognitive (space, time) and social domains (emotion, imitation).
With the technical group at the MPI for Psycholinguistics (headed by Peter Wittenburg) we have organized the world’s largest digital archives (c. 45 TB) of endangered languages, with state of the art annotation and retrieval software, funded by MPG, EC and VW. MPI for Psycholinguistics has also hosted 5 externally-funded documentation projects, with 3 current PhD students. I have conducted anthropological and linguistic fieldwork in India, Mexico, Australia and Papua New Guinea. In the last 10 years I have made 8 field expeditions (2-3 months each) to New Guinea, and organized Eurocores and NWO-funded major expeditions to Melanesia. I have published some dozen papers on Yélî Dnye, Papua New Guinea, with grammar and dictionary in draft (see Rossel Island fieldsite). More generally I have contributed to eHumanities, fostering cooperative, cumulative science in the humanities (e.g. under the ECHO project I headed an initiative for a Dutch Sign Language (NGT) archive, now a major resource).
Within the Categories project, we are now focussing (a) on the language of perception, looking at the relative codability of different perceptual domains (color, taste, smell, touch) across diverse languages, (b) the linguistic coding of emotion across cultures. We ask: why are some domains more codable than others?; To what extent does this reveal architectural constraints on the access of language to perception?; To what extent does this depend on elaborated cultural traditions?
Within the Interactional Foundations of Language project we investigate the pragmatic infrastructure for language use. Current research is focussed on 'action ascription', the realtime assignment of 'speech acts' to utterances. Research on question-response sequences has revealed a remarkably similar pattern of timing across many languages (Stivers et al, PNAS 2009), where modulations in the timing signal modulations in the action.