Modeling Life-Like Behaviors of An Animated Virtual Tutor

The aim of this study is to understand and model the coordinated verbal and nonverbal behaviors that occur during one-on-one tutoring of students by expert reading tutors, and to develop learning tools in which a virtual tutor-a lifelike 3-D computer character-models these coordinated behaviors, and thus behaves more like an effective and sensitive human tutor. We hypothesize, and test the hypothesis, that virtual tutors that model the contextually appropriate exchange of signals and cues between an expert teacher and a student will produce better user experiences and more effective learning outcomes.