We are interested in human--human dialogue, and the particular aspects we are looking at happen to, for reasons not just of lexical coincidence, all start with the syllable "co": context, coherence, coordination, and collaboration. (It's just that cocococo-lab would have been a bit excessive.) More precisely, we are interested in how participants in a dialogue coordinate what they do, that is, collaborate, in order to create content that is, within its context, coherent. There are two main methods we use: we observe people doing this in real, spontaneous dialogues (perhaps under specially created conditions), and then derive models from the observation, which we implement and evaluate.

das, 8.5.07