Coding for mutuality strategies in dialogue
Brian Hansen · 2018
Mutuality strategies provide a way of predicting andproducing1600 41 that best convey the degree to which conversants have mutual understanding. We formulated the notion of a mutuality strategy model in an effort to account for the observed behaviors of human conversants engaged in task-orientedface-to-face38580-3740 While we have proposed to realize this model computationallyandtationa it both as a simulation of human behavior by computational agents and asa4399-3 application, it is first necessary to establish thatmutuality33200 0 can be identified with a reasonable level of reliability. In this paper we report the results of a study designed to determine the degree to which transcribedutterances41 be coded as conveying different levels of acceptance to prior utterances. Empirical analysis of dialogue typically has involved measuringeither 37687 task performance or within-utterance characteristics such as disfluencies.Forencies. 32698 systems, more useful characteristics for improvi...