2. THE MECHANISMS OF COLLABORATIVE LEARNING

2.8 Social grounding

Social grounding is the mechanism by which an individual attempts to maintain the belief that his partner has understood what he meant, at least to an extent which is sufficient to carry out the task at hand [19]. This mechanism involves that the speaker monitors the listener's understanding and, in case of misunderstanding, attempts to repair communication. Verbal and non-verbal cues are important to detect misunderstanding. Repairs involve disambiguating dialogues, pointing to shared souvenirs, to visible objects, drawing a schema, and so forth. Through this mechanism, both partners progressively build a shared understanding of the problem.

For Clark and Brennan [19] the cost of social grounding changes with the medium. They decompose this cost into eleven factors, among which two factors appear to us as especially relevant here:

- The speaker change costs.
In face-to-face conversation, participants manage turn taking quite easily. There is an implicit rule that only one agent speaks at a time. When this rule is perturbed, for instance, when satellite-based communication introduces a one-second delay between the emission and the reception of a message, turn taking mechanisms are deteriorated. In Internet-based tools, one has to address those costs. Some tools for instance, require the partners to use some commands to inform that they take turn or that they end turn.
- Display costs
Social grounding is easier when participants can monitor the facial expression of their partners, to find hints of misunderstanding or to know what they are looking at. When those facilities are not provided, it takes more efforts to detect and to repair misunderstanding. Note, for instance, that in text-based MOOs textually displayed "emotions" or "thinking aloud" fulfill this role.

These three last mechanisms illustrate a new theoretical perspective inherited from the situated cognition approach [19], and refered to as 'socially shared cognition'. This theory views a group as a single cognitive system distributed over individuals. It does not focus on individual contributions, but on the shared representation built by the group. Within this perspective, the main reason why collaborative learning is efficient is that members learn to think interactively: thinking is not only manipulating mental objects, but also interactions with others and with the environment.

ICCAI 95 article - 08 FEB 95

[Next] [Previous] [Up] [Top]

Generated with CERN WebMaker