The purpose of intelligent tutoring systems is to provide students with personalized instruction and feedback. The focus of these systems typically rests in the adaptability of the feedback provided to students, which relies on automated assessments of performance in the system. A large focus of my previous work has been to determine how natural language processing (NLP) techniques can be used to model individual differences based on students' natural language input. My proposed research will build on this work by using NLP techniques to develop stealth assessments of students' individual differences and to provide more fine-grained information about the cognitive processes in which these students are engaged throughout the learning task. Ultimately, my aim will be to combine this linguistic data with on-line system data in order to develop more robust student models within ITSs for ill-defined domains. [For complete proceedings, see ED560503.]