Adaptive Support for Face-to-Face Collaborative Learning at Tabletop Computers

Abstract

Collaborative learning is a common practice in today's classrooms. Technology-supported collaborative learning environments are becoming increasingly sophisticated, enabling new ways for students to work together with technology. Research has shown that collaborative learning has many benefits, particularly for developing students' higher-order thinking and problemsolving skills. However, it has also been shown that students do not always know how to collaborate effectively, which can inhibit the success of collaborative learning. These findings suggest that collaboration itself is a skill that needs to be fostered and developed in the classroom. Tabletop computers have affordances for collaborative learning because of the large, shared interface that multiple people can see and interact with at once. Despite these affordances, small group work at a tabletop computer is just as susceptible to breakdowns in collaboration as group work using other kinds of tools. Through design-based research in classroom settings, I have investigated how tabletop computers can model social regulation—the processes that groups use to manage and monitor their collective work—in order to detect when a group of students is in need of support. While collaboration is driven by the verbal and gestural interactions between the learners, the tabletop is only able to capture direct interaction with the device. I have identified touch patterns that reflect the quality of social regulation and can be used to detect problems in the collaborative process. To enable the real-time use of these touch patterns, I developed a machine learning-based approach for distinguishing among simultaneous users at a tabletop computer. I also present software adaptations designed to encourage more effective collaboration that are triggered when breakdowns in collaboration are detected. A classroom evaluation of these adaptations showed that they deterred disruptive behavior and reduced the length of periods of sustained, low-quality collaboration. My dissertation demonstrates the following thesis: Interactive tabletop software that can automatically detect breakdowns in collaboration and adapt in real-time to scaffold effective social regulation can improve secondary school students' collaboration skills.