We examine the implications of negativity in free-form dialogue between student tutors and a synthetic agent in APLUS, a learning-by-teaching online learning environment for Algebra. We attempt to determine whether the negativity of a student tutor's discourse with the agent indicates that the student is learning more or less of the material and whether the feedback they give the synthetic agent is more or less accurate. We found a weak negative correlation between tutor negativity and learning gains and a strong negative correlation between tutor negativity and accuracy of feedback. Negativity might indeed indicate that student tutors lack mastery of the subject matter and need assistance themselves and detecting negativity during tutoring and providing appropriate assistance might enhance the effectiveness of APLUS and other intelligent tutoring systems.
Bringula, Rex. (2013). Exploring the Implications of Tutor Negativity Towards a Synthetic Agent in a Learning-by-Teaching Environment. Philippine Computing Journal.