Spell checking in computer-assisted language learning: A study of misspellings by nonnative writers of German

This thesis investigates spell checking in Computer-Assisted Language Learning. It analyzes nonnative misspellings and evaluates the performance of a generic spell checker. A total of 1027 unique misspellings were collected from 32 beginners and 16 intermediate university learners of German who worked on two different task types. CLASSY, a classification system for nonnative misspellings that has been developed for this thesis, is introduced. CLASSY categorizes misspellings along four taxonomies: competence vs. performance, linguistic subsystem, language influence, and target modification. Results show that 72% of the nonnative misspellings are competence-related rather than accidental typographical mistakes. Furthermore, the generic spell checker tested on the misspellings of this study corrects only 62% of them. The study also indicates that both proficiency level and task type influence the learners' misspellings and affect the spell checker's correction success. Finally, the thesis makes computational and pedagogical suggestions to enhance spell checking for foreign language writers.