The goal of the FLUENT system is to provide authentic language practice for language learners. In learning systems, misunderstandings occur -- they are the basis for teaching and tutoring. There is the need, then, for error repair. This paper asks the question: if there is a learner error, what is the appropriate range of responses to resolve the misunderstanding? In starting to build such a module, a number of disciplines come into play: language teaching, intelligent tutoring systems, natural language processing and computational linguistics. We will touch on the difficulty in synthesizing each of these domains as we begin to analyze the handling of errors in a language learning environment.