Representing Regularity: The English Past Tense

For the last ten years, the English past tense has been an important test-case in the debate between rule-based and connectionist accounts of human language processing (Pinker, 1991; Rumelhart and McClelland, 1986). The work we report here focuses on a particular psychological property of regular and irregular past tense verbs; namely the demonstration that regularly inflected verbs prime their stems whereas irregular verbs do not (Marslen-Wilson, Hare and Older 1993; Stanners et al. 1979). This result has been interpreted as supporting a dual mechanism account (Pinker, 1991) in which the past tenses of regular verbs are generated by a rule-based mechanism, whereas irregular past tenses are stored in an associative memory system. Our purpose here is to investigate whether these representational differences between regular and irregular verbs can be accounted for by a single mechanism, connectionist model.