Michael Gasser, Chan-Do Lee
Despite its successes, Rumelhart and McClelland's (1986) well-known ap(cid:173) proach to the learning of morphophonemic rules suffers from two deficien(cid:173) cies: (1) It performs the artificial task of associating forms with forms rather than perception or production. (2) It is not constrained in ways that humans learners are. This paper describes a model which addresses both objections. Using a simple recurrent architecture which takes both forms and "meanings" as inputs, the model learns to generate verbs in one or another "tense", given arbitrary meanings, and to recognize the tenses of verbs. Furthermore, it fails to learn reversal processes unknown in human language.