denoet the speed and force of an impact. Gesturing played a big part in the decisions that wre made about what sorts of experiments to perform with the simulation, i.e. how and when to change the salient variables for mass and velocity. Quantifications were therefore indicated through speech and gesture.
With gazing and domain interactions there was not a significant association with Speaker discussing the motion of the ice pucks. It was only when the Hearer looked at the listener who was talking about the motion of the pucks that a significant relationship occurred. The problem space was mainly controlled by the Speaker and it was the Speaker's actions rather than gaze which were more important here than when the subject domain per se was discussed. This finding could well be an atypical one since the topic of elastic collisions involves notions which are perhaps more easily described via gestures. Therefore this type of analysis needs to be applied to another subject discipline which does not focus around the subject of dynamics.
Another important category that prompted gaze and gesture was that of cognitive planning which in our framework is referred to as "Progressing the Investigation". How did gesture and gaze with respect to this event category influence the problem solving process? The data from this study suggested that a number of incidents of eye gazing during the planning phase correlated with problem solving success. This data supports Taylor et al's ( 1993) observation that eye contact was important in the planning stage of an empirical investigation. In addition Mutual gesturing and strategies employed to progress the investigation correlated significantly with problem solving achievement and hence the non-verbal interactions of gaze and gesture were acting in different ways with respect to problem solving and planning activities.
The other event categories concerned with strategies to "progress understanding" were associated with mutual gaze related incidents or those of the Hearer only. Mutual gesturing events were also associated with this category but the non-verbal interactions were not as frequent and significant with respect to joint problem success as the other event categories mentioned above. What is perhaps more surprising is the role of gesture when gestures were offering both negative feedback and positive support to one another. Gazing was associated with these types of interactions and there were no significant differences with respect to gender for these latter event categories.
There were, however, significant differences with respect to gender and gaze. Male Speakers gazed more when talking about the motion of the pucks. Female Speakers differed in that they looke at their partners more when trying to progress mutual understanding of the problem space. It was the mixed gender pairs who gazed less than the two other gender groups when providing emotional support or progressing the investigation.
Our categories of talk which precipitated gaze and gesture were in agreement with many accounts of discourse and language use (e.g. Grosz and Sidner 1986 and Litman and Allen 1987) which assume that was we talk we tend to look for evidence that we have been misunderstood and then attempt to repair the problem. This process was called "grounding". Other forms of grounding were seeking positive understanding such as acknowledgements which take the form of "yeah", "yes" "mmm" ( Schegloff 1981) and gestures such as head nodding ( Goodwin 1981). There was also the "next turn" phenomenon which indicates comprehension has taken place and the third and most basic form of positive evidence was continued attention. Mutual gazing was an indicator of this latter phenomenon and was prompted by discussions about how to progress understanding of the problem space and talk which offered positive emotional support.
To conclude, our main finding suggests that gesture and gaze play different roles in constructing the problem space and continuing the collaboration process between the dyads. Understanding how this interplay also effects problem solving when subjects are linked with a video conferencing system will be part of our future work. This will allow us to make recommendations about the quality of visual links that are required to sustain collaborative learning at a distance.
Azimita, M. and Montgomery, R. ( 1993) "Friendship, transactive dialogues and development of scientific reasoning". Social Development, 2, 202-221.
Barfurth, M. A. ( 1995) "Understanding the Collaborative Learning Process in a Technology Rich Environment: The Case of Children's Disagreements". In J. L. Schnase and E. L. Cunnius (Eds.) CSCL 95 Computer Support for Collaborative Learning pp.8-13.