Greg Kochanski

My main line of research involves modeling the brain's implementation of prosody, the non-lexical part of human speech. Understanding prosody and how it is encoded can help to understand the control strategies for all motions. It will also allow us to better understand human-human communication and it may have medical diagnostic applications.

This is a promising new area, somewhere near linguistics, neurobiology and biomechanics, with important technological and social implications. The research that I and my collaborators have completed is one of the very few clean and quantitative experiments linking language to the measurable dynamics of the organism. We succeeded in building a model of speech production that links objective acoustic measurements of speech to some linguistic concepts like position in a sentence, or the part-of-speech of a word. The model works because it builds on hard-won knowledge of the physics of speech.

Older papers can be found here. Most recent papers: Talk on acoustic analysis of the question/statement distinction. Preprint on modeling of Mandarin intonation. Talk on modeling of Mandarin intonation.