Many are the ways to learn identifying multi-modal behavioral profiles of collaborative learning in constructivist activities

• DOI: 10.1007/s11412-021-09358-2
• Publication Date: 2022-01-21T03:02:25Z
• Authors: Jauwairia Nasir; Aditi Kothiyal; Barbara Bruno; Pierre Dillenbourg
• Keywords: Learning Analytics; Affect

Abstract Understanding the way learners engage with learning technologies, and its relation their learning, is crucial for motivating design of effective interventions. Assessing learners’ state engagement, however, non-trivial. Research suggests that performance not always a good indicator especially open-ended constructivist activities. In this paper, we describe combined multi-modal analytics interaction analysis method uses video, audio log data to identify collaborative behavioral profiles 32 dyads as they work on an task around interactive tabletops robot mediator. These profiles, which name Expressive Explorers , Calm Tinkerers Silent Wanderers confirm previous findings. particular, amount speech overlap between pair are behavior patterns strongly distinguish non-learning pairs. Delving deeper, findings suggest overlapping can indicate engagement conducive learning. When more broadly consider learner affect actions during task, better able characterize range exhibited among those who learn. Specifically, discover two dimensions along learn vary, namely, problem solving strategy (actions) emotional expressivity (affect). This finding behavior; one leads frustration compared another. have implications real-time interventions support productive in tasks.