LeKoDe-CH: a Swiss learner corpus of L2 German for language acquisition research, teaching, and learner development

Description

As extensive collections of learner-language data, learner corpora constitute an important source of evidence for research on foreign- and second-language (L2) acquisition. Depending on the corpus's composition, learner varieties from different first languages (L1s) and proficiency levels can be empirically examined from multiple perspectives (e.g. lexical, morphosyntactic and pragmatic) (Granger, 2008; Schmidt, 2011). Notably, insights gained from learner corpus analyses are relevant not only for L2 acquisition research, but also for language pedagogy. In this context, they provide a basis for developing learner-friendly teaching materials and offer scientifically grounded support for the language acquisition process (Granger et al., 2015).

Despite their potential for L2 acquisition research and pedagogy, only a relatively small number of learner corpora are currently accessible to researchers, teachers and learners. This is true of the German corpus landscape in general, and Switzerland in particular, where no accessible, continuously expanding DaF/DaZ learner corpus exists. Against this backdrop, the project is compiling version 1.0 of the LeKoDe-CH corpus, the first accessible, extensively annotated Swiss DaF/DaZ learner corpus.

The corpus will be regularly expanded and improved, and will serve as an important resource for research into language acquisition, pedagogy and learner development. Creating version 1.0 involves considering different approaches to the following challenges:

Data protection challenges involved in collecting learner productions.
Automatic processing of learner texts using existing computational linguistic tools.
Development of learner-appropriate, differentiated linguistic annotation.