Collaborative tasks do not always promote equal learning. Varying levels of social interactions and regulation at the individual and group levels can influence knowledge construction efforts and learning success. To understand which collaboration patterns may be more conducive to learning, this study examined the relation between social exchange, regulation, and learning outcomes. Four project-based engineering undergraduate teams were audiotaped in collaborative tasks (7514 talk turns). Discourse was coded for regulation processes and types (self and socially shared regulation), and analyzed with Epistemic Network Analysis and Process Mining. We find that teams who reported more frequent social exchange engaged in shared regulation together with planning and monitoring more frequently, while teams with less exchange engaged in long durations of collaboration. Furthermore, students in teams with more engaged regulation reported enhanced beliefs in group efficacy to solve collaborative tasks. The study illustrates the potential of applying quantitative approaches to analyzing rich discourse.

协作任务并不总是能促进平等学习。个人和团体层面上不同程度的社会互动和调节可以影响知识建设的努力和学习的成功。为了了解哪种协作模式可能更有利于学习，本研究考察了社会交流，监管和学习成果之间的关系。在协作任务中录制了四个基于项目的工程学本科团队的录音（7514个谈话回合）。话语针对监管流程和类型（自我和社会共享监管）进行了编码，并通过认知网络分析和流程挖掘进行了分析。我们发现，报告了更频繁的社会交流的团队参与了共同监管，同时也更加频繁地进行了计划和监控，交流较少的团队则需要长期合作。此外，团队中有更多参与法规的学生报告说他们对解决集体任务的团队效能的信念增强。该研究表明了应用定量方法分析丰富话语的潜力。