Variation in interlanguage morphology: (s) Plural-marking in the speech of Chinese learners of English

Much work by second language acquisition researchers has focused on how interlanguage varies across different learners or within the speech of one learner across different tasks. Previous studies of variation in English as a second language, however, have led to conflicting and contradictory claims concerning the causes of variation. The present study suggests that such contradictions are due to the descriptively inadequate model of variation used in earlier studies. A multivariate model is proposed and applied to the analysis of variation in English (s) noun plurals by NSs of Chinese. Four major factors are hypothesized to influence variation: convergence toward the speech patterns of an interlocutor predicted by speech accommodation theory; level of proficiency in English; the phonological environment of the -$s$ inflection together with the syntactic function and semantic features of the noun; and a tendency to omit marking of plural when it is redundant and to retain it when it transmits necessary information (the functional hypothesis). Six low proficiency and six high proficiency NSs of Chinese were interviewed in English by a fellow Chinese interviewer and a NS interviewer. A multivariate analysis of the data confirmed that ESL proficiency had a strong influence on variation, but that the pattern of variables affecting variation differed for low and high proficiency learners. For low proficiency learners (s) pluralization was promoted most strongly by phonotactic factors. For high proficiency learners, however, phonotactic factors did not have a significant effect on variation. Contrary to the prediction of speech accommodation theory, the frequency of (s) plural marking in the speech of Chinese and NS interviewers had no observable effect on variation in (s) plural marking in the speech of the informants. And contrary to the prediction of the functional hypothesis, informants either marked plural redundantly or did not mark it at all. It appears that informants acquired a set of structural number contrasts and concord rules rather than a salient semantic notion of number. In addition, informants applied the pluralization rule almost categorically in a limited set of prototypical expressions consisting of numerals or quantifiers modifying plural nouns, with the NP functioning syntactically as an adverbial