SLaTE 2009 Workshop Technical Programme and Abstracts

In this work we introduce a set of response strategies that capture the effect of cultural norms on the behavior of conversational agents in language and culture training systems. Response strategies cover behavior such as deception, vagueness, and distraction. Starting from a vocabulary of strategies derived from the literature on compliance and cooperation, we compare this explicit list to evidence of implicit strategizing in hand-authored dialogs captured from a serious game system. As a result, we find that the strategy representation codifies a layer of communicative information that seems to be necessary for believable dialog in the context of teaching cultural communication skills. We also explore how dialog strategies can be explicitly authored and tested, presenting results from an implemented prototype. O1.2 Rainbow Rummy: A Web-based Game for Vocabulary Acquisition using Computer-directed Speech Brandon Yoshimoto, Ian McGraw, Stephanie Seneff MIT Computer Science and Artificial Intelligence Laboratory, 32 Vassar Street, Cambridge, MA 02139 [email protected], [email protected], [email protected]