Using Multimodal Learning Analytics to Explore Collaboration in a Sustainability Co-located Tabletop Game

Serious Games (SGs) are particularly suitable to foster collaboration in complex domains that challenge formal education approaches. However, their effectiveness depends on their features as much as on the ability to assess their impacts on players, and analysing collaboration in games remains by and large an open problem. Research has traditionally used rich unimodal data to examine collaboration processes in games (e.g. video content analysis of verbal exchanges). Despite providing relevant semantic information, this can make data coding and analysis difficult and time-consuming. Furthermore, unimodal approaches can only partially capture complex processes defined by multiple interacting variables, such as collaboration. Recent research highlighted the potentialities offered by multimodal learning analytics (MMLA) to address these issues. MMLA integrates multiple types of data captured both in and out of the game system through different modalities to analyse complex processes. Although it has been highlighted as particularly suitable to investigate collaboration, research on MMLA in SGs is still scarce. This work contributes to the state-of-the-art by leveraging MMLA to explore collaboration indicators in a multiplayer, co-located SG for education in sustainable development. Our results corroborate the MMLA effectiveness in analysing complex collaborative dynamics, and identify key multimodal analytics useful to investigate collaboration in SGs.