A cognitive framework for second language error diagnosis

Abstract

The problem of identifying learners’ misconceptions hinges on making a correct analysis and diagnosis of the learners’ input. In computer systems for Second Language teaching, past approaches have used specialised knowledge to infer the learner’s misconceptions. This knowledge was specific to the language being taught and sometimes to the learner’s native language. An alternative approach is to use general knowledge which can be applicable to a variety of languages. Such an approach is presented here. This approach uses a unifying linguistic theory which encompasses all human languages in a universal framework. This theory describes languages as a set of interlocking principles and introduces parameters to account for the variations amongst languages. In this approach, the diagnoses are made following some patterns of acquisition that this theory defines.