Key points are not available for this paper at this time.
This paper describes preliminary work in exploring the relative effectiveness of speech versus text based tutoring. Most current tutorial dialogue systems are text based (Evens et al., 2001; Rose and Aleven, 2002; Zinn et al., 2002; Aleven et al., 2001; VanLehn et al., 2002). However, prior studies have shown considerable benefits of tutoring through spoken interactions (Lemke, 1990; Chi et al., 1994; Hausmann and Chi, 2002). Thus, we are currently developing a speech based dialogue system that uses a text based system for tutoring conceptual physics (VanLehn et al., 2002) as its "back-end". In order to explore the relative effectiveness between these two input modalities in our task domain, we have started by collecting parallel human-human tutoring corpora both for text based and speech based tutoring. In both cases, students interact with the tutor through a web interface. We present here a comparison between the two on a number of features of dialogue that have been demonstrated to correlate reliably with learning gains with students interacting with the tutor using the text based interface (Rosé et al., submitted).
Rosé et al. (2003) studied this question.