ABSTRACT
Curiously, the development of a true intelligent computer assisted language learning (ICALL) system requires that a number of fundamental issues about second language acquisition and instruction be addressed, yet relatively little research is available that links basic questions about acquisition to applied questions about engineering a useful tutoring system. This chapter presents a theoretical model of input and interaction in second language acquisition and a research protocol-the FlatLand protocol-employed to provide some empirical grounding to this model. The model represents an overall picture of how language learning activity involving a single human tutor and a single human learner transpires on an utterance-by-utterance basis, thus providing insight into how one might conceptualize the control structures needed for an ICALL system.
