The aim of this study is to examine means of fostering videoconference-based collaborative learning, by focussing on three issues: (1) to induce collaborative learners to write a co-construct, applying (in addition to their shared knowledge) their unshared knowledge, which tends to be neglected, according to the social-psychological research paradigm of information pooling; (2) to activate these learners in their dialogues to exchange unshared knowledge possessed by one learning partner, so that it becomes shared knowledge possessed by both partners (knowledge transfer); (3) to try out, as an instructional support measure, scripted, content-specific visualisation, combining a content scheme with an interaction script. An experiment was conducted with 30 learning dyads, divided into three conditions of videoconference-based learning with application sharing: without instructional support, with content-specific visualisation, and with scripted content-specific visualisation. As expected, the scripted content-specific visualisation led to a higher transfer of previously unshared knowledge to shared knowledge. But, contrary to expectation, the scripted content-specific visualisation did not induce the learning partners to apply more unshared knowledge in writing their co-construct. Instead, in all three experimental conditions, learners brought significantly more shared knowledge into the co-construct than would have been expected from the distribution of shared and unshared knowledge measured before collaboration.