| In this dissertation I examine the design of an anchored discussion system to facilitate online learning conversations. I conducted a sequence of three quantitative studies. The first study compared two forms of an open source anchored discussion system with each other and with a conventional discourse system. Thus, the purpose of the first study was to investigate which functional characteristics might help in the promotion of rich and constructive interactions. Results showed that the online presence of the learning material supported sustained discussions centered on understanding the meaning of a text. Furthermore, annotating messages to parts of a text promoted more complex forms of interactions with the learning material and re-focused the discussion when it digressed. Based on these results, I conducted the second study to determine how collaborative knowledge construction activities facilitated by two forms of an open source anchored discussion system affected individual learning outcomes. I presented a theoretical research model combining social constructivism, grounding theory, and elements from cognitive load theory. I found that the annotating functionality indeed decreased the coordination of online social interaction. Additionally, less need for coordination left more room for knowledge generating activities. Finally, more knowledge generating activities led to a deeper understanding of the learning material. In the light of the second study, I re-designed and evaluated the open source anchored discussion system. The goal of the re-design was to focus social construction of knowledge on relevant information from online academic texts. I developed three prototype software environments: teacher-based attention guidance, peer-oriented attention guidance, and control condition. I found that teacher-based attention guidance helped treatment group students to select relevant information and discuss it with higher quality interaction patterns. Moreover, when treatment group students switched to peer-oriented attention guidance, they maintained focus on central domain principles and their interrelations. However, the socio-cognitive process for discussing the importance of candidate areas did not differ between peer-oriented attention guidance and control conditions. |
