Poster #37 - Dialogue to Text for Assessment of Collaborative Learning

• In Event: 18.060 - MTCC Poster Session Two
  In Poster Session: 18.060-5 - Advanced Technologies for Learning Poster Session

Abstract

This study aims to leverage education research in a ‘post-truth’ era in two senses: 1) having students engage in dialogue to explore the truth and 2) digitalizing the dialogue into text to make it evident for us all. We used a Knowledge Constructive Jigsaw method and voice recognition system to transcribe all students’ conversations. Comparing dialogues in the jigsaw-activity with presentations in the crosstalk-activity, the recognition accuracy was lower in the dialogues than in the presentations. Another study in a think-aloud situation demonstrates that participant engaged more in exploratory talk. The results indicate that those speeches, even hard for AI to transcribe, are precious devices for students to learn deeper and for us to deepen understanding of how they learn.

Authors

• Hajime Shirouzu, the University of Tokyo

• Takahiro Nakayama

• Moegi Saito, University of Tokyo

• Shinya Iikubo

• Tohru Nagano