Acceso usuarios
Cecilie Carlsen
This article focuses on the proficiency level of texts in Computer Learner Corpora (CLCs). A claim is made that proficiency levels are often poorly defined in CLC design, and that the methods used for level assignment of corpus texts are not always adequate. Proficiency level can therefore, best be described as a fuzzy variable in CLCs, representing a potential source of error in CLC-based research. The article starts with a review of some of the most commonly used methods of proficiency-level assignment of texts in CLCs and a discussion of strengths and weaknesses of the different methods. A pioneer project to link a learner corpus of Norwegian (ASK) to the Common European Framework of Reference for Languages (CEFR) is presented to illustrate that a reliable assignment of corpus texts is viable by applying insights and practice from the professional field of language testing and assessment. Finally, some advantages of a learner corpus reliably linked to the CEFR are discussed in relation to SLA research, to language test development, and to a validation of the CEFR-level descriptions.
