Paper: ArCADE: An Arabic Corpus of Auditory Dictation Errors

ACL ID W14-1813
Title ArCADE: An Arabic Corpus of Auditory Dictation Errors
Venue Innovative Use of NLP for Building Educational Applications
Session
Year 2014
Authors

We present a new corpus of word-level lis- tening errors collected from 62 native En- glish speakers learning Arabic designed to inform models of spell checking for this learner population. While we use the cor- pus to assist in automated detection and correction of auditory errors in electronic dictionary lookup, the corpus can also be used as a phonological error layer, to be combined with a composition error layer in a more complex spell-checking system for non-native speakers. The corpus may be useful to instructors of Arabic as a sec- ond language, and researchers who study second language phonology and listening perception.