The validity of an automated experimental system incorporating a computer mediator for resolving conflicting requests to speak among four-member problem-solving teams was assessed. Learning how to use problem-solving strategies was affected by this new method for facilitating participation in dialogue. Strategy learning among shy and assertive groups was mediated by a computer-executed rule resolving conflicting requests to speak in favor of those having spoken least. Total time waiting for one's turn to speak and the rate of rejection were shown to be related to initial shyness level and also to vary over the course of three problem-solving games. A discriminant function analysis identified the set of variables which accounted for most of the variance in assignment of subjects to the four treatment conditions which were based upon scores on the Personal Report of Communication Apprehension. Analysis of variance was performed to detect shyness level by game interactions. (Author)