Spelling: computerised feedback for self-correction

Abstract

Research has shown that any assumption that L2 learners of English do well to rely on the feedback provided by generic spell checkers (for example, the MS Word spell checker) is misplaced. Efforts to develop spell checkers specifically for L2 learners have focused on training software to offer more appropriate suggestion lists for replacing misspellings, but research suggests that students often select the wrong word from such lists. Moreover, since their spelling errors are often competence errors arising from misconceptions about the target language, students might benefit from feedback that addresses the underlying problem rather than simply suggesting the correct word. This article describes the use of a corpus of learner English to identify common misspellings produced by Spanish learners of English. For each of these errors, hand-crafted feedback has been written explaining how the correct spelling is consistent with the orthography or morphology of English. This material has been incorporated into a prototype spell checker for learners of English. Meanwhile, the software simply flags other errors as “not amongst the 90,000 commonest words of English”. Because it also detects unusual bigrams (word pairs), the program finds more spelling errors than generic spell checkers and additionally detects some non-spelling errors. Tests indicate that such easy-to-build spell checkers might become useful tools for L2 learners, and, along with other recent technological developments, suggest that student self-correction of compositions (including grammatical and lexical errors as well as spelling errors) is becoming increasingly feasible.