| This thesis examined the impact of two different tutorial strategies (Feedback and No-Feedback) deployed by a pedagogical agent (PA) in a multi-agent hypermedia-learning environment, MetaTutor, on college students' embodiment of discrete emotions. Participants were randomly assigned to one of the two PA tutorial conditions, which were designed to assist students with setting three appropriate sub goals for their learning session. The two conditions differed in that the Feedback tutorial strategy condition provided participants with more information on how they could improve their suggested sub goals, were given more opportunities to try again (when unsuccessful) and the PA's counterproposals integrated relevant components of participants' proposed sub goals, whereas the No-Feedback PA tutorial condition did not and simply proposed the most related sub goal. FaceReader, an automatic facial expression recognition program, was used to analyze videos of learners' facial expressions in order to determine their basic emotional states. Tens of thousands of emotional data points (frequency data) were provided by FaceReader, which were transformed into proportions for analysis. The results of this thesis came from two analyses (Analysis 1, University of Memphis and McGill University, N = 18; Analysis 2, McGill University, N = 24), which examined undergraduate students' emotional responses to the PA during the initial sub goal setting phase, while they set three sub goals for their two-hour learning session with MetaTutor. Results demonstrated that learners' experienced more negative than positive or neutral emotions during their interaction with the PA and less anger and happiness, but more surprise in the Feedback tutorial condition. This thesis provided evidence for recommending tutorial strategies similar to the Feedback tutorial strategy examined and the use of the control-value theory of achievement emotions to assess the emotional impact of tutorial strategies. |
PDF Full Text Request