Testiaineistosta kielenoppijakorpukseksi

Abstract

This article concerns language learner corpus based on data from the National Certifi cate of Language Profi ciency (NC) that provides tests in nine different languages on three different examination levels for adult language learners. The quantitative data of the corpus consist of background information of test takers, and test scores of reading and listening comprehension, speaking, writing and structure and vocabulary tests in numeric form, and the qualitative data contain speaking and writing performances of test takers in nine languages. The data are multilingual, and parallel across languages in terms of language profi ciency levels and functions of test tasks. The corpus design is looked at from the viewpoint purposes of the corpora, sampling, representativeness and usability of interface. Further, usability of the corpus is discussed in the light of practicality and future possibilities.