Regular seminar for MA students, but open to anyone interested.

This hands-on course takes you through the processes involved in working with multimodal corpora: the manual annotation of relevant phenomena and transcription of speech; techniques for speeding up manual work and making it more reliable; the construction of validation criteria and the creation of tools to perform automatic validation; the creation and execution of tools performing automatic annotation; and the analysis of annotated data with the aim of answering research questions and creating computational models of conversational phenomena. The course also includes a discussion of the practical purposes of annotation, what annotations to aim for first, and the pros and cons of generality vs. purpose-build annotation schemes.