Iterative Evaluation of a Large-Scale, Intelligent Game for

Abstract. Educational content developers, including AIED developers, traditionally make a distinction between formative evaluation and summative evaluation of learning materials. Although the distinction is valid, it is inadequate for many AIED systems because they require multiple types of evaluation and multiple stages of evaluation. Developers of interactive intelligent learning environments must evaluate the effectiveness of the component technologies, the quality of the user interaction, and the potential of the program to achieve learning outcomes, in order to uncover problems prior to summative evaluation. Often these intermediate evaluations go unreported, so other developers cannot benefit from the lessons learned. This paper documents the iterative evaluation of the Tactical Language Training System, an interactive game for learning foreign language and culture. This project employs a highly iterative development and evaluation cycle. The courseware and software have already undergone six discrete stages of formative evaluation, and further formative evaluations are planned. The paper documents the evaluations that were taken at each stage, as well as the information obtained, and draws lessons that may be applicable to other AIED systems