Traffic lights and smiley faces : do children learn mathematics better with affective open-learner modelling tutors?

摘要

This PhD thesis investigates how the use of open-learner modelling (OLM) techniques and the inclusion of affective components in the design of intelligent learning environments can facilitate learning and enhance software usability by increasing children’s motivation and engagement in the learning process. The research solely focuses on mathematical applications, given to English and French children aged seven to nine years. The main contribution of this PhD concerns the study of children’s willingness and ability to use affective OLM applications for better learning. The results show that the way children interact with an OLM application depends on its level of openness and student’s control over the learning process. Children seem to want to access their learner model components. Such access, as well as the understanding of the learner model content, is facilitated by affective embodied pedagogical agents. The children using an intelligent tutoring system with a negotiated learner model appeared to learn more than children who used an environment with an editable or inspectable learner model, as their learning gain from during each learning session on software revealed to be higher . The use of two different representations of the learner model content - one representing the children’s self-beliefs, and the other the system’s assessment of knowledge acquisitions – has proven to lead children to be more involved in the representation of what they know by visually comparing their views of how much a specific concept is grasped to the system’s assessment, and engaging in a negotiation process when a disagreement was found, which led them to learn more from the sessions on software. The results and contributions of this thesis should help give evidence of which theories of emotions better apply to children aged seven to eleven working on OLM applications, how children can, want, and effectively use learner model components according to its representation, content, and method of interaction, and therefore help in the design of future affective OLM educational applications for primary school children.