Published online by Cambridge University Press: 01 September 2004
Pinker and Prince (1988) argued that two dissociable systems underlie the development of linguistic representations: one rule governed and the other associative. These two dissociable systems of representation and processing are claimed to be a linguistic universal (Pinker, 1999). Therefore, one should expect that nonnative speakers of a language also manifest the same kinds of dissociations in performance between rule-based and associative features of language as native speakers. The study reported here extends the work of Prasada and Pinker (1993) into the second language (L2) domain to test whether nonnative speakers (a) perform similarly to native speakers and (b) dissociate rule-based from associative features of language. In Prasada and Pinker, the degree of similarity between a nonce verb and a real English verb influenced past-tense generalizations on nonce irregular verbs but did not influence generalizations on nonce regular verbs. In the experiment reported here, first language (L1) and L2 participants of different ages and language groups were compared on the same task as that used by Prasada and Pinker. Participants overall produced more verbs with a past-tense suffix for nonce regular items than for the nonce irregular items. Significant group effects indicate that participants' varying levels of experience with English play an important role in their past-tense productions of nonce verbs. Furthermore, similarity influenced both the regular and irregular verbs on a production task. These results are discussed in terms of whether there are two dissociable systems underlying L2 linguistic knowledge and whether an alternative single associative learning mechanism could be responsible for the development of the system of regular and irregular inflectional morphology in both L1 and L2 learning.This research was partly funded by a Social Sciences and Humanities Research Council (SSHRC) grant (no. 752-94-1808). I appreciate the help of Jennifer Graham in collecting the data and am grateful to Nick Ellis, Nina Spada, Lydia White, and five anonymous SSLA reviewers for helpful comments on earlier drafts of this manuscript.