Academic journal article
By Lonchamp, Jacques
Educational Technology & Society , Vol. 13, No. 2
Computer-supported collaborative learning (CSCL) emphasizes the importance of social processes as an essential ingredient of learning. CSCL has been recognized as a possible way to prepare people for the knowledge society, to achieve deeper learning than traditional methods provide, and to better meet the expectations of the Net generation (Resta & Lafferiere, 2007). There exist two main approaches for supporting collaborative learning (Jermann, Soller, & Muehlenbrock, 2001). The first one structures the situation in which the collaboration takes place: the task (e.g., with a learning script), the group of learners (e.g., with specific roles), the interaction (e.g., with sentence openers), and the artifacts. The second approach involves structuring the collaboration itself through coaching and selfregulation: as the collaboration progresses, the state of interaction is automatically evaluated by the system with respect to a desired state, and remedial actions may be proposed. In this article, we focus on the coaching and selfregulation approach as a complement to the structuring approach in the context of synchronous CSCL environments, that is, same time/different places or same time/same place systems. The coaching and self-regulation approach requires storing the stream of all relevant interaction events, computing on-demand interaction analysis (IA) indicators that support tutoring activities, and computing periodically synthetic visual representations of IA indicators that support learners' self-regulation. The tutor and the system together guide the learners toward effective collaboration and learning.
The literature describes many computer-based IA indicators. They can be related to the cognitive, social, and affective dimensions of interactions. However, most of them are dependent of a specific learning activity or a specific learning environment (Dimitracopoulou et al., 2004). Building customizable IA tools is recognized as a promising research direction "that could help to face the problems of restricted validity field of IA tools, the one of low powerfulness of IA output, as well as this of not fulfilment of various users' profiles" (Dimitracopoulou et al., 2004). The idea of building more generic and customizable mechanisms is not restricted to the IA domain, but can impact all aspects of CSCL systems. The CSCL community in its first decade has produced a large number of ad hoc systems focusing on particular situations and contexts, and aiming at triggering specific learning processes. That was the case of all early specialized synchronous tools for structured discussion (e.g., Baker, Quignard, Lund, & Sejourne, 2003), collaborative design (e.g., Soller, Linton, Goodman, & Lesgold, 1999), collaborative knowledge construction (e.g., Suthers & Jones, 1997), collaborative modeling (e.g., Baker & Lund, 1996) and collaborative writing (e.g., Jaspers, Erkens, & Kanselaar, 2001). Many researchers claim that this first generation of ad hoc, specialized, and closed tools should be replaced by systems "richer and appropriate for various collaborative settings, conditions and contexts" (Dimitracopoulou, 2005); "reconfigurable, adaptive, offering collections of affordances and flexible forms of guidance" (Suthers, 2005); and "very flexible and tailorable" (Lipponen, 2002). Two research streams, following either the component-based approach (De Chiara, Di Matteo, Manno, & Scarano, 2007; Asensio,
Dimitriadis, Heredia, Martinez, Alvarez, Blasco, & Osuna, 2004) or the model-based approach (LAMS, 2009; Ronen, Kohen-Vacs, & Raz-Fogel, 2006; Bote-Lorenzo, Hernandez-Leo, Dimitriadis, Asensio-Perez, GomezSanchez, Vega-Gorgojo, & Vaquero-Gonzalez, 2004), aim at meeting these expectations. The system described in this article, Omega+, follows the model-based approach: an explicit model parameterizes a generic kernel for flexibly supporting different kinds of collaborative applications (Lonchamp, 2006). …