Engagement in discussion forums is associated with better success rates and persistence in MOOCs, which have dropout rates of 90-95%. However, learners engage very little in MOOC discussion forums, which they perceive as unhelpful and not supervised by teachers. Since teachers cannot provide individual support to all learners in these courses, effective ways of supporting engagement must be provided for both large and small forums. This starts with a better understanding of engagement in MOOC forums as research has mainly focused on studying this phenomenon in a holistic way in distance learning and MOOCs. Furthermore, the concept of engagement in MOOC forums remains very confused as to what it means despite its widespread use. Research studies particular forms of engagement in these forums, sometimes with indicators that are not unanimously agreed upon.

The present thesis therefore aims to better understand and support student engagement in MOOC discussion forums through four main contributions. The first contribution is a conceptualization of engagement in MOOC discussion forums as perceived by students. The analysis of eleven semi-structured interviews with successful MOOC students revealed four dimensions (behavioural, affective, cognitive and social) of engagement and related indicators. The second contribution is an exploration of the potential factors of engagement in MOOC discussion forums as perceived by students through the prism of several theories and models: the self-determination theory and reasons for engagement, models of motivational dynamics, acceptability and acceptance of technologies. In addition, we constructed a model that shows how the 34 observed potential factors interact to influence learner engagement in MOOC forums. On the basis of these potential factors, suggestions for supporting engagement adapted to small forums were formulated. The third contribution concerns the development of a short questionnaire for assessing engagement in MOOC forums. It assesses the four dimensions of engagement and consists of the indicators from the first contribution. This grid is useful for teachers to review the pedagogical scenario of the MOOC forums according to the level of engagement of the learners, but also for the learners to objectively evaluate themselves. Finally, the fourth contribution explores a way of automatically managing MOOC discussion forums, as it is impossible for teachers to answer questions from thousands of learners. In particular, it highlights among the potential factors of Study 1, those that most prevent and limit engagement in MOOC forums. Based on this diagnosis, she argues from the literature that chatbots can help to better support engagement in MOOC discussion forums.