Publications

We present a methodology for building a system that corrects ill-formed English text produced by writers learning English as a second language. This methodology is described within the framework of a system to correct English text produced by deaf writers whose native language is American Sign Language (ASL). ASL is a visual-gestural language which significantly differs from English in both its sentence level grammar and its discourse strategies. Our model is based on the hypothesis that there is language transfer (LT) from ASL to English for these writers, and that transfer should be exploited in the identification and correction processes. We provide examples of sentence-level errors that are attributable to LT, thus supporting our belief that LT occurs between ASL and English, While researchers in second language acquisition have hypothesized and demonstrated language transfer at the sentence level, we provide examples of discourse errors which are attributable to LT. These errors could be corrected with the aid of a mechanism that tracks the discourse focus of written English and knowledge about ASL discourse strategies/rul