Design of Collaborative Learning Environment

A variety of peer learning environment designs to support effective collaborative learning has been attempted or proposed. The backbone of collaborative learning is the willingness of the peers to participate in collaboration in a constructive sense. This has been studied by a number of educational psychology researchers [Madden & Slavin1983,Slavin1978] who confirm that the peers in collaborating classes felt that their peers wanted them to learn. Slavin [Slavin1990] reports studies that confirm the willingness of peers to make the collaborative learning efforts succeed and the improvement in social status of the peers who achieved better than other peers.

Effective collaboration requires appropriate pairing of collaborating peers. There have been studies to identify the factors involved in combining peers. Slavin [Slavin1990] reports a study by Kuhn [Kuhn1972] who found that a small difference in cognitive level between collaborating peers was more conductive to cognitive growth than a larger difference. This supports the view that the collaborative peers should have almost equal knowledge levels to make the collaboration constructive. However, the study by Azmitia [Azmitia1988] found that when novices were paired with experts on a model building task they improved significantly while equal ability pairs did not. The contradiction in pairing of knowledge levels could be because of the difference in domains or characteristics of peers or the experimental setup. Azmitia's view is further supported by Rogoff [Rogoff1990,Rogoff1991] who found better results with adult-child than with child-child pairing. She also identified the intermediate variable that explains these variations. She found that effective adults involved the child in explicit decision making while skilled peers tended to dominate the decision making. Thus, a collaborative learning environment should also possess mechanisms to identify the appropriateness of peers to collaborate with each other.

There have also been studies to support collaborative learning conducted from the computational viewpoint. For instance, Blaye et al. [Blaye et al. 1990] use a (peer1-peer2-system) environment to solve information gathering and planning problems. While one peer attempted to solve the problems the other watched the problem solving process and provided critiques and suggestions. Chi et al. [Chi et al. 1989] present a restricted environment where the learner is forced to explain an example to himself or herself, thus leading to reflection.

The combination of tasks and the number of peers involved in the learning is determined by the subject domain being taught, the learning theory adopted, and the capability of the system. The following list presents a variety of collaborative environment designs:

1

Two or more peers collaborating with each other using the computer as a mediating tool. The system does nothing but provide the communication channels for collaboration without playing an active role.

2

Two or more peers collaborating with each other using an active tutor that controls and directs collaborative interactions.

3

Two or more peers working together on a problem at the same workstation, using the tutor in much the same way as when a peer works alone. The peers could be intelligently grouped based on the background knowledge about the peers.

4

Two or more peers working together on a problem from networked machines where peers take turns carrying out the next action. Communication windows are available so that students can send advice, suggest alternative actions, comment on their partner's actions, etc. Peers could work in reciprocal teaching mode or competitive fashion. On the other hand, one peer could act as a coach while the other would attempt to solve the problem.

5

Two or more peers working together with at least one of the peers being simulated by the system. Peers can take turns carrying out actions with a simulated peer. The simulated peer can carry out actions automatically or on demand. Simulated peers can be selected from a library of student models. VanLehn et al. [VanLehn & Ohlsson1994] and Kumar [Kumar1992] discuss the generation, utility and control of simulated peers in collaborative learning.

Collaborative learning research has not identified formalisms prescribed for an ideal collaborating environment design for peers. It is left as a subjective assessment and prescription depending on the application, peers, and collaborative environment.