An Experimental Investigation on Using Pedagogical Conversational Agents: Effects of Posing Facilitation Prompts in Oral-Based Peer Learning

Abstract

This study investigates the use of pedagogical conversational agents (PCAs) that intervene in learner-learner collaborative learning activities. In addition, this study investigates how the quality of learning performance in a simple concept explanation task may change due to the use of multiple PCAs that pose different types of facilitation prompts to the learners. A controlled experiment was performed by comparing a condition using multiple PCAs wherein each PCA provided different types of facilitations with a condition using multiple PCAs wherein each PCA provided a mixture of the two types of facilitations. Using the WOZ method, this study reports the preliminary results of an analysis of oral-based peer learning. Lexical network analysis was used to understand the complexity of learner’s semantic knowledge over two-time series. The results of the analysis show that when multiple PCAs were used with different facilitation prompts, the lexical network became more complex, showing that the learners developed a more sophisticated knowledge about the concept throughout their explanations. The significance of using multiple PCA is that it allows different types of contents to be considered during interaction.