We develop a grammatical error correction (GEC) system for German using a small gold GEC corpus augmented with edits extracted from Wikipedia revision history. We extend the automatic error annotation tool ERRANT (Bryant et al., 2017) for German and use it to analyze both gold GEC corrections and Wikipedia edits (Grundkiewicz and Junczys-Dowmunt, 2014) in order to select as additional training data Wikipedia edits containing grammatical corrections similar to those in the gold corpus. Using a multilayer convolutional encoder-decoder neural network GEC approach (Chollampatt and Ng, 2018), we evaluate the contribution of Wikipedia edits and find that carefully selected Wikipedia edits increase performance by over 5%.

The MERLIN corpus is a written learner corpus for Czech, German,and Italian that has been designed to illustrate the Common European Framework of Reference for Languages (CEFR) with authentic learner data. The corpus contains 2,290 learner texts produced in standardized language certifications covering CEFR levels A1-C1. The MERLIN annotation scheme includes a wide range of language characteristics that enable research into the empirical foundations of the CEFR scales and provide language teachers, test developers, and Second Language Acquisition researchers with concrete examples of learner performance and progress across multiple proficiency levels. For computational linguistics, it provide a range of authentic learner data for three target languages, supporting a broadening of the scope of research in areas such as automatic proficiency classification or native language identification. The annotated corpus and related information will be freely available as a corpus resource and through a freely accessible, didactically-oriented online platform.

This paper describes the Error-Annotated German Learner Corpus (EAGLE), a corpus of beginning learner German with grammatical error annotation. The corpus contains online workbook and and hand-written essay data from learners in introductory German courses at The Ohio State University. We introduce an error typology developed for beginning learners of German that focuses on linguistic properties of lexical items present in the learner data and present the detailed error typologies for selection, agreement, and word order errors. The corpus uses an error annotation format that extends the multi-layer standoff format proposed by Luedeling et al. (2005) to include incremental target hypotheses for each error. In this format, each annotated error includes information about the location of tokens affected by the error, the error type, and the proposed target correction. The multi-layer standoff format allows us to annotate ambiguous errors with more than one possible target correction and to annotate the multiple, overlapping errors common in beginning learner productions.