Title: Question Personalization in an Intelligent Tutoring System

Abstract: This paper investigates personalization in the field of intelligent tutoring systems (ITS). We hypothesize that personalization in the way questions are asked improves student learning outcomes. Previous work on dialogue-based ITS personalization has yet to address question phrasing. We show that generating versions of the questions suitable for students at different levels of subject proficiency improves student learning gains, using variants written by a domain expert and an experimental A/B test. This insight demonstrates that the linguistic realization of questions in an ITS affects the learning outcomes for students.