The paper describes a learner corpus of Czech, currently under development. The corpus captures Czech as used by non-native speakers.
We discuss its structure, the layered annotation of errors and the annotation process.