E. André et al. (Eds.): ADS 2004, LNAI 3068, pp. 254–264, 2004. © Springer-Verlag Berlin Heidelberg 2004 Generating Socially Appropriate Tutorial Dialog W. Lewis Johnson 1 , Paola Rizzo 2 , Wauter Bosma 3 , Sander Kole 3 , Mattijs Ghijsen 3 , and Herwin van Welbergen 1 1 CARTE, USC / Information Sciences Institute, 4676 Admiralty Way, Marina del Rey CA, 90292 USA {Johnson, Bosma,Ghijsen, Herwin}@isi.edu 2 Dept. of Philosophy and Education Sciences University of Rome “La Sapienza” P.le Aldo Moro 5, 00189 Rome, Italy rizzo.paola@virgilio.it 3 University of Twente, Enschede, Netherlands; {bosmaw,kole, mattijs}@cs.utwente.nl Abstract. Analysis of student-tutor coaching dialogs suggest that good human tutors attend to and attempt to influence the motivational state of learners. Moreover, they are sensitive to the social face of the learner, and seek to miti- gate the potential face threat of their comments. This paper describes a dialog generator for pedagogical agents that takes motivation and face threat factors into account. This enables the agent to interact with learners in a socially ap- propriate fashion, and foster intrinsic motivation on the part of the learner, which in turn may lead to more positive learner affective states. 1 Introduction Animated pedagogical agents, or guidebots, exploit human-like communication mo- dalities, such as speech and nonverbal gesture, in order to promote more effective learning [10]. Learning materials incorporating animated agents can engender a higher degree of learner interest than similar learning materials that lack such agents [16], and reduce the perceived difficulty of the learning material [1]. They can also produce a positive affective response on the part of the learner, sometimes referred to as the persona effect [13]. This is attributed to the natural tendency for people to relate to computers as social actors [23], a tendency that animate agents exploit. Educational researchers increasingly recognize the importance of learner affective states in promoting effective learning. Of particular importance factors such as self- confidence and interest that contribute to learner intrinsic motivation [25]. Expert human tutors are also able to recognize when learners have negative affective states due to poor motivation or low sense of self-efficacy, and can try to influence learner motivation through encouragement and other motivational tactics [12]. Since ani- mated agents can display emotion via speech and gesture, it is reasonable to suppose