| The social dimension of learning informs much of the current research and development of learning technologies. In some lines of research, the learning tool itself provides the social dimension, often in the form of an intelligent virtual agent. In other lines of research, a learning tool provides an environment in which multiple human learners can collaborate. Integrating these lines of research, this dissertation makes progress toward answering a question that demands increased attention: How can designers of learning technologies leverage the social capabilities of intelligent virtual agents to support collaborative learning? Specifically, I examine the effects of virtual learning companions on the discourse of pairs of middle school students collaborating in a game-based learning environment for computer science education, with a focus on equity.;To investigate this line of research, I crafted multiple individual episodes with virtual learning companions, designed for specific locations in the game where students could most benefit from their presence. Through their conversations with the virtual learning companions, I hypothesized that a wider range of students would benefit from productive learning experiences, and that students would collaborate more equitably. These hypotheses emerge out of prior research that has shown the benefits---particularly with respect to equity---that virtual learning companions can have for individual students. To investigate equitable collaboration among pairs of students, this work looks at the dialogue of student pairs as they collaboratively interacted in the learning environment. Following established methods for analyzing discourse, I have annotated transcripts of the student pairs by manually tagging each utterance.;These coded transcripts allow for the investigation of research questions concerning the impact of the virtual learning companions, including how the impact differs based on student characteristics such as gender and prior computing experience. They also lead to insights on the general nature of equitable collaboration within game-based learning experiences for middle school students. The quantitative and qualitative analyses included in this dissertation provide the groundwork for much future research in the fields of virtual agents, game-based learning, computer-supported collaborative learning, learning sciences and computer science education. |
