Using Automatic Roundtrip Translation to Repair General Errors in Second Language Writing

In this paper, we evaluate for the first time the use of Machine Translation technology to repair general errors in second language (L2) authoring. Contrary to previously evaluated approaches which rely exclusively on unilingual models of L2, this method takes into account both languages, and is thus able to model linguistic interference phenomena where the author produces an erroneous word for word translation of his L1 intent. We evaluate a simple roundtrip MT approach on a corpus of foreign-sounding errors produced in the context of French as a Second Language. We show that the roundtrip approach is better at repairing linguistic interference errors than non-interference ones, and that it is better at repairing errors which only involve function words. We also show that the first leg of the roundtrip (inferring the author's L1 intent) is more sensitive to error type and more error prone than the second leg (rendering a correct L1 intent back into L2).