How People Talk to Computers, Robots, and Other Artificial Communication Partners

Abstract

End-to-end evaluations of more conversational dialogue systems with naive users have uncovered severe usability problems that, among other things, result in low task completion rates. First analyses suggest that these problems are related to the system’s dialogue management and turn-taking behavior. This paper starts with a presentation of experimental results, which shed some light on the effects of that behavior. Based on these findings, some criteria which lie orthogonal to dialogue quality are spelled out. As such, they nevertheless constitute an integral part of a more comprehensive view on dialogue felicity as a function of dialogue quality and efficiency. Since the work on spoken and multimodal dialogue systems presented and discussed herein is aimed at more conversational and adaptive systems, we also show that in certain dialogical situations it is important for such systems to align linguistically towards the users. After describing the corresponding empirical experiments and their results, pragmatic alignment will be introduced as more general framework for these types of adaptation to users which are, in the light of the aforementioned studies critical to building more conversational dialog systems.