engleski

Preparing the NooJ German module for the analysis of a learner spoken corpus

This project merges the knowledge of a foreign language teaching specialist and a language processing specialist to detect and classify different types of errors found in the spoken classroom discourse of Croatian learners of German as a foreign language. The role of NooJ in this project is to detect and annotate different types of errors in order to facilitate the analysis of such data. The preliminary corpus consists of 5 different classroom interactions of Croatian master's level students of teaching German as a foreign language that were recorded during their regular classes. The recordings have been faithfully transcribed, keeping the mistakes originally introduced by the student speakers. In order to evaluate the student-speakers' linguistic, lexicosemantic, and pragmatic competences, the rated criteria fall into the areas of general linguistics, vocabulary, vocabulary control, grammatical accuracy, coherence, and pragmatic appropriateness. This classification conforms to the recommendations by the Common European Framework of Reference for Languages. The project uses the existing NooJ resources for German that had been enhanced with new dictionary entries, an additional morphological grammar constructed to recognize misspellings, and syntactic grammars introduced to find and annotate errors at the higher levels of syntactic analysis.

student learner corpus ; detecting errors ; German language ; dictionary ; morphology ; syntax ; NooJ

nije evidentirano