Intelligent Tutoring Systems (ITSs) are situated in a potential struggle between effective pedagogy and system enjoyment and engagement. iSTART (Interactive Strategy Training for Active Reading and Thinking), a reading strategy tutoring system in which students practice generating self-explanations and using reading strategies, employs two devices to engage the user. The first is natural language processing (NLP). Incorporating NLP within iSTART allows students to use their own thoughts and ideas to communicate with the system, and serves as the core intelligence of the system that is used to drive the feedback and the adaptive interactions during practice. Studies have shown that the NLP algorithms within iSTART perform comparably to human raters and provide a good measure for the sophistication of student self-explanations. The second device is the use of game-based practice. Skill mastery requires a significant commitment to practice over extended periods of time. Unfortunately, this persistent and repetitive practice is also associated with disengagement from the target educational task. Therefore, a gaming environment was developed that integrates multiple combinations of enjoyable, engaging game elements with the target practice tasks. This paper describes these two principle aspects of iSTART and research on their effectiveness.