Corpus based language technology for computer-assisted learning of Nordic languages: Squirrel Progress Report September 2001

1 Background and objectives of the project 1.1 Background The most prevalent forms of information technology in computer-assisted language learning (CALL) applications are email, multimedia and computerised multiple-choice test forms. Until recently, the contribution of language technology (LT) to CALL has been next to nonexistent, although there is now a growing amount of work in the LT community on using natural language processing (NLP) technologies and speech technology for creating 'intelligent' CALL (ICALL) applications , where language analysis and learner modelling are used in order to analyse and respond properly to free-form linguistic input from the learners (e.g. answers to exercise questions) or even to score essays. Text corpora and corpus tools, another kind of LT, are also used in CALL, indirectly in the form of so-called learner corpora, i.e. collections of second or foreign language texts produced by language learners. These corpora can be examined for interlanguage traits which in turn can inform language teaching. There is also a more direct use of corpora in so-called " data-driven learning " , where students use concordancers and search tools for empirical investigations of the use of words and constructions in large text materials. These tools are charac-terisable as language technology only in a very wide sense, since their linguistic sophistication is of a very low order (usually not even POS tagging is involved, let alone more sophisticated linguistic analysis).