Learning companion robots that support spoken natural language dialogue present exciting opportunities for adaptive and personalized interaction. We hypothesize that adaptive robot behaviors can have a positive effect on student motivation. Supporting these adaptive behaviors in real-time requires that learning companion robots construct and dynamically update models of student motivational state. Our project examines the utility of speech prosody in contributing to a dynamic model of student motivational state in human-robot peer tutoring interaction. The project involves collecting spoken dialogue data with a learning companion robot, measuring the motivational state of students, and modeling the relationship between speech prosody and student motivation.