A Learner Corpus for Listening and Spelling Errors

A learner corpus is a valuable resource for developing automatic assessment techniques for implementation in a computer-assisted language learning system. Currently, there are numerous spelling error corpora for English. Most of them are focused on learner-written texts. However, another important aspect of spelling is phonological awareness and auditory discrimination of speech sounds. In this paper, we propose the scheme of a spelling error corpus which is intended to be used for analysis of listening and spelling errors of English language learners. In our data collection method, words are dictated to the user by a computer program and their inputs and interactions are recorded to provide the corpus data. The resulting data can be analyzed either by human or automatically by machine learning methods to assess listening and spelling performance of an individual or a group of learners

Keywords:

Learner corpora , error-analysis , computer-assisted language learning