Picture of John Pate I am an assistant professor in the Department of Linguistics at the University at Buffalo. I use machine learning methods to understand how people learn and use language.

Language acquisition

How children learn to group words into meaningful phrases in the face of uncertainty is a fundamental question for linguistics and cognitive science. In this thread of work, I build probabilistic models for syntax acquisition to understand how different learning biases can turn a child's observations into evidence for syntactic structure.

Language use

Natural languages exhibit several statistical properties that suggest they may be adapted in some ways to be efficient in the sense of the noisy channel theorem: natural language utterances are short and error-resistant. In this line of work, I develop the information-theoretic framing of natural language communication with insights from coding theory, and perform corpus studies to check for evidence of adaptation.



I am currently looking for students who have a strong background in programming and machine learning, and an interest in computational approaches to linguistic questions.



johnpate [at] buffalo.edu



Last updated: Sep 30, 2016