Stephan Meylan


I am a postdoctoral associate in the Bergelson Lab at Duke University, in the department of Psychology and Neuroscience. I recently completed my Ph.D. in the Computational Cognitive Science Lab at UC Berkeley (Advisor: Dr. Tom Griffiths), focusing on probabilistic models of language learning and language processing. Now I am working with Dr. Elika Bergelson and Dr. Roger Levy (MIT Computational Psycholinguistics Lab) developing models that make use of the newly collected SEEDLingS corpus.

In my research, I use a combination of computational models, corpus studies, web-based experiments, and in-lab experiments. I am currently learning how to run experiments with infants and toddlers in the Bergelson lab.

My primary research goals are:

  1. to characterize how natural languages are shaped by the need for efficient online processing
  2. to characterize the origins and uses of abstract linguistic knowledge in early development
  3. to improve machine representations of word and utterance meaning

I am particularly interested in how people learn and use so-called “function” or “closed-class” words (articles, pronouns, prepositions, etc.) and inflectional morphology (systematic markings for number, gender, etc.). See Publications and Projects for more details.