Self-Regulated Learning in Learning Environments With Pedagogical Agents That Interact in Natural Language

Abstract

This article discusses the occurrence and measurement of self-regulated learning (SRL) both in human tutoring and in computer tutors with agents that hold conversations with students in natural language and help them learn at deeper levels. One challenge in building these computer tutors is to accommodate, encourage, and scaffold SRL because these skills are not adequately developed for most students. Automated measures of SRL are needed to track progress in meeting this challenge. A direct approach is to train students on fundamentals of metacognition and SRL, which is the approach taken by iSTART, MetaTutor, and other agent environments. An indirect approach to promoting SRL, taken by AutoTutor, is to track the student's knowledge and SRL based on the student's language and to respond intelligently with discourse moves to promote SRL. This fine-grained adaptivity considers the student's cognitive states, the discourse interaction, and the student's emotional states in a recent AutoTutor version.