Computational Extraction of Social and Interactional
Meaning from Speech

Instructor(s): Dan Jurafsky and Mari Ostendorf

Computational methods for extracting social and conversational meaning, mainly from conversational speech but also from text. Topics will include automatic determination of emotions (angry, annoyed, happy), personality (extrovert, introvert), medical states (depression, intoxication), style (flirtatious, polite, funny), register (formality, audience), intention (deception), positioning, as well as the role of accommodation. Focus on extraction of cues from speech and text, with special attention to prosodic, acoustic, and repair/disfluency structure.

Familiarity with Praat, basic ability in programming (at least one or two college-level courses) using Perl, Python, Java, or Matlab.