Supporting Productive Dialogue with Collaborative Conversational Agents: Frequency and Sequence of Learner Behavior

This study presents the impact of a collaborative conversational agent on the chat dialogues of 17 dyads. Specifically, participants were asked to study two cases of climate change using the chat tool of a learning platform to communicate. The agent was available only in the second case and was able to monitor the participants’ discussion and intervene using the talk moves of the academically productive talk framework (e.g., prompting for elaboration, examples, etc.). Analysis of the chat logs showed higher levels of elaboration for the second case, while the agent acted as a catalyst increasing the chances for elaboration and transactivity.