Universal Grammar and the interpretation of reflexives in a second language

UNIVERSAL GRAMMAR AND THE INTERPRETATION OF REFLEXIVES IN A SECOND LANGUAGE Margaret Thomas Harvard University This article addresses the debate about whether adult language learners have access to the principles and parameters of universal grammar in constructing the grammar of a foreign or second language (L2). I investigate the interpretation of English reflexive pronouns by native speakers of Japanese and of Spanish, and the interpretation of the Japanese reflexive zibun by native speakers of English and of Chinese. The data suggest that L2 learners observe constraints defined by universal grammar, constraints which they could not have derived solely from inspection of the input data, nor from the treatment of anaphors in their native language. Thus these results support the proposal that adult learners have access to universal grammar.* Introduction 1. The idea that there exists a 'logical problem of language acquisition' arose from the finding that child language learners seem to have knowledge about language which they cannot have gained by observation and imitation of adult speech (Baker & McCarthy 1981, Hornstein & Lightfoot 1981). In Chomsky's words, Our knowledge [oflanguage] is richly articulated and shared with others from the same speech community, whereas the data available are much too impoverished to determine it by any general procedure of induction, generalization , analogy, association, or whatever' (1986:55). An example of children's knowledge of language which is underdetermined by the input data is Crain & Thornton's 1988 report that 3- to 5-year olds seem to know the syntax of wa««a-contraction. In a production task, children spontaneously produced contractions in 59% of wH-questions in sentences like 1 , where the extraction site is in object position of the subordinate verb eat. But contractions appeared in only 4% of sentences like 2, where the extraction site is in subject position between want and to. (1)What do you want to eat __ ? (2)Who do you want __ to help you? Adult usage may inform children that wanna-contraction occurs in some contexts in English, but it does not necessarily indicate where contraction is impossible . Generative linguistics suggests that children make few mistakes in such cases because they have access to abstract and detailed linguistic knowledge which constrains their grammar. They do not contract in 2 because want and to are not strictly adjacent in this structure, and this fact is given by innate principles of language.' The challenge for this view of language acquisition is * I would like to thank three anonymous reviewers for their help, and to acknowledge the assistance of Susumu Kuno, Suzanne Flynn, and Catherine Snow in the larger project from which these results are drawn. The generous support of a Dokkyo University International Cooperation Research Fellowship made collection of the Japanese data possible. 1 In Pesetsky's account (1982:255-63), UG informs speakers ofEnglish that a wH-trace intervenes between want and to in 2 because WH-traces must receive Case and Case is assigned under adjacency . 211 212LANGUAGE. VOLUME 67, NUMBER 2 (1991) to specify the inventory, nature, and role of these linguistic principles and their associated parameters, which together define the limits of variability among human languages. That is, the task is to determine the contents and operation of universal grammar (UG). A separate but related issue is whether adults learning a foreign or second language (L2) exhibit this kind of subtle knowledge which cannot be accounted for by exposure to the surface facts of the language. Do adult learners have access to principles and parameters of UG in constructing the grammar of a second language? Lenneberg's 1967 Critical Period hypothesis argued that firstlanguage (Ll) learning after puberty cannot make use of an innate 'language acquisition device', an ancestor of the notion of UG. Dulay & Burt (1974), Bailey et al. (1974), and others challenged the relevance of the Critical Period hypothesis to L2 learning, chiefly by demonstrating that Ll and L2 learners follow similar sequences of acquisition of specific grammatical and morphological items. Although evidence based on order of acquisition eventually proved unconvincing (McLaughlin 1987:31-4, 66-9), increasingly sophisticated investigation of L2 data in the light of generative linguistic theory has evolved. At the center of this research is a...