David Farwell, Louise Guthrie, and Yorick Wilks
Our position is not very contentious. Both monolingual and bilingual machine readable dictionaries (MRDs) have information related to words and their contexts. Lexicons for natural language processing (NLP) need information related to words and their contexts. Why not, then, use that information in an MRD which is relevant to the lexical specifications of an NLP system and which can be extracted automatically either to construct or partially construct the system’s lexical items?