The Swiss learner corpus SWIKO

Abstract This article introduces the Swiss learner corpus SWIKO, a multilingual and multi-modal corpus developed at the Research Centre on Multilingualism in Fribourg (Switzerland). SWIKO comprises over 2,600 annotated oral and written productions by 340 lower-secondary school students. Between 2017–2022, data was elicited through eight communicative tasks systematically varied by text type, topic, and task constraint and produced under different conditions with respect to mode, modality, and language (German, French, and English; as both the L1 and L2/L3). All productions are part-of-speech tagged. In addition, 1,549 written L2/L3 productions were rated along the CEFR scales, and 890 written German L1 and L2 texts were semi-automatically error annotated. Due to its controlled task design, rich annotation, and high comparability of learner groups, SWIKO supports a wide range of empirical and pedagogical applications, including contrastive analyses and investigations of task effects. Open access to the corpus is provided via https://ifm-swiko.unifr.ch/ .