This dissertation, "Characterizing the Discourse Patterns of Collaborative Knowledge Building" by Lai-fan, Fu, 傅麗芬, was obtained from The University of Hong Kong (Pokfulam, Hong Kong) and is being sold pursuant to Creative Commons: Attribution 3.0 Hong Kong License. The content of this dissertation has not been altered in any way. We have altered the formatting in order to facilitate the ease of printing and reading of the dissertation. All rights not granted by the above license are retained by the author. Abstract: This study aimed to develop a holistic understanding of knowledge-building discourse supported by Knowledge Forum among primary-and secondary-school students in Hong Kong. It is argued that prior studies of knowledge building did not adequately address the important question of how ideas are progressively improved because these studies employed cognitively oriented approaches that discarded the sequential, structural, and situational information about the process of group interactions. To better understand this question, the author applied methods from qualitative traditions to the study of knowledge-building discourse.The study was part of a five-year professional development project, "Developing a teacher community for classroom innovation through knowledge building." The author and other project members collaboratively analyzed more than hundreds of Knowledge Forum views to gain an initial understanding of productive group interactions. The selection of data set for the study utilized purposive sampling. The author evaluated the online discourses of several dozens of classes, with the criteria of productive group interactions. Three classes from different schools were selected: Grade 5 Science, Grade 10 Liberal Studies, and Grade 10 Visual Art. These classes offered diverse examples to enhance the transferability of the findings. The data set comprised 764 Knowledge Forum messages, which were examined in great detail by a four-stage qualitative method. The first stage was a thematic analysis at the thread level to pre-process the online discourses for the subsequent analyses. The second stage was a qualitative coding at the action level to characterize the discourse components of the threads. The coding utilized 7 main codes that were adapted from van Aalst (2009): community, information, question, idea, linking, agency, and meta-discourse. This coding scheme formed a foundation of the data analysis, and this study extended the scheme in two ways. First, it gave the main codes a more theoretically solid foundation by conducting a literature review to further conceptualize or re-conceptualize the main codes. Second, it went beyond conducting the qualitative coding to seek for general patterns of interactions in the third-stage analysis. The third stage was a narrative analysis at the episode level to identify discourse patterns. Eleven patterns were identified to demonstrate productive and unproductive group interactions. The findings from the three stages of analysis were then interpreted to provide a comprehensive profile of the class discourses in the final-stage analysis. The relationship between the discourse profiles and idea improvement was explained. Finally, a validity check was conducted and the findings suggested that the discourse patterns could be used as a heuristic device to provide a basis for understanding other discourses.The implications of this study are threefold. Methodologically, the study has identified eleven discourse patterns that can be conceived as an extensive classification scheme allowing researchers to understand different types of group interaction in asynchronous online discussion forums. Theoretically, the discourse patterns contribute to the literature concerning the process of computer-mediated group interactions. Pedagogically, it is hoped that the discourse patterns can be used as conceptual tools for scaffolding students toward productive g