Contents

Two competing forces are at work in developing Intelligent Tutoring Systems: the desire to provide comprehensive feedback, and the desire to ensure that the knowledge learned will transfer to real life situations. One factor that can prevent comprehensive feedback being given is the problem of ambiguity, Menzel [13]. Methods for dealing with ambiguity exist. We explored one such method in this study, by constructing two systems for teaching German adjective endings. Tutor 1 requires the student to specify their complete mental model of the problem, in addition to answering the question. The system can thus correct their specific misconceptions. Tutor 2 only requires the student to answer the question. The two systems were evaluated by first year German students at the University of Canterbury. Results of this evaluation are unclear, due in part to the small number of participating students. Tutor 1 gave greater learning overall, however the degree to which students learnt the constraints that map only to the adjectiv