Issues for Tutoring Knowledge in Foreign Language Intelligent Tutoring Systems

The tutoring approach and the representation of the expert knowledge in a foreign language ITS depend on the particular skill being taught (grammar, dialogue, vocabulary). Rule-based knowledge representations are appropriate for grammar tutoring. Scripts or semantic network representations may be more useful when learning language in context. Pedagogical contexts for the tutor and several knowledge representation schemes useful for foreign language learning are discussed in this paper. Research on developing a discovery learning environment for vocabulary learning as an alternative to planned-based rules in the tutor model is presented. The environment used both associative and propositional network models to represent lexical knowledge in different contexts. Preliminary data from a study comparing the effect of each model on word recall suggest that an associative network representation for the lexicon is effective for vocabulary tutoring with more advanced learners. The word usage results suggest that either representation influences communication skills when the vocabulary is already known by the learner. Contrary to what was predicted, the use of a more complex representation (propositional network) did not influence word usage positively.