Second Language Acquisition of Reflexives Revisited
BP YUAN
DOI: https://doi.org/10.2307/416484
1994-01-01
Language
Abstract:DISCUSSION NOTE Second language acquisition of reflexives revisited BopiNG Yuan University of Cambridge* Quite a few researchers have used data on the acquisition of English reflexives by native speakers of Chinese, Japanese, and Korean as evidence for the claim that Universal Grammar (UG) is available to L2 learners and parameter resetting is possible in second language acquisition (SLA; cf. Finer & Broselow 1986, Finer 1990, 1991, Hirakawa 1990, Thomas 1989, 1991). In this discussion note, I will show that the acquisition ofEnglish reflexives by Chinese, Japanese, and Korean learners is not a valid test for parameter resetting in SLA, because English reflexives share the same setting as phrasal reflexives in Chinese, Japanese , and Korean, and it is therefore very likely that no parameter resetting is involved in the acquisition of English reflexives by the learners. According to Chomsky (1981 , 1986), anaphor, which includes reciprocals and reflexives, is subject to Principle A of the Binding Theory: (1)Principle A: An anaphor is bound in its governing category. While Principle A is a principle of UG, the notion of 'governing category' is a parameter in UG, which means that it varies from language to language. Wexler & Manzini propose that UG provides the settings in 2a-e for the governing category parameter (1987:53): (2)Governing category parameter (GCP): ? is a governing category for a iff ? is the minimal category which contains a and a.has a subject, or b.has an Infl, or c.has Tns, or d.has an indicative Tns, or e.has a root Tns. According to Wexler & Manzini's definition of governing category, languages differ with respect to how far away the antecedent can be from the reflexive. From the English sentence in 3 (= Thomas 1991, ex. 6), we can see that English reflexives are subject to setting 2a of the GCP, and that only Sue can bind the reflexive herself. (3)Alice, thinks that Sue, loves herself*,ij. (4)Alice, wa Suej ga zibun-,ij o aisite Alice top Sue nom self acc love iru to omotte iru. is comp think is 'Alice, thinks that Sue, loves self,/,.' * I would like to thank three anonymous Language reviewers for their valuable comments. The abbreviations used in examples in this paper are: acc = accusative; comp = complementizer; nom = nominative; sub = subject; and top = topic. 539 540LANGUAGE, VOLUME 70, NUMBER 3 (1994) (5)Alice¡ rènwéi Suej ai zijínj. Alice think Sue love self 'Alice¡ thinks Sue loves self¡/j.' From the Japanese sentence in 4 (= Thomas 1991, ex. 7) and the Chinese sentence in 5, we can see that the Japanese reflexive zibun and the Chinese reflexive ziji are both associated with setting 2e of the GCP, taking the root sentence as their governing category. The Japanese reflexive zibun in 4 and the Chinese reflexive ziji in 5 can be bound by Alice as well as by Sue. With regard to the possible antecedents of the reflexive, the five types of language in 2 form an inclusive hierarchy, with a language like English being least inclusive and languages like Japanese and Chinese being most inclusive. Wexler & Manzini (1987) propose that children always form conservative hypotheses about what can serve as the antecedent of a reflexive, and that they will adopt a more inclusive setting when they encounter reflexives bound within larger domains. In SLA, the question arises as to how reflexives are acquired if the governing category of reflexives in the Ll is more inclusive than that in the L2, as is the case with Chinese, Japanese, and Korean learners of English. In a small pilot study, Finer & Broselow (1986) investigated the interpretation of English pronouns and reflexives by six adult native speakers of Korean. Korean is like Japanese and Chinese in that the Korean reflexive casin can be bound in the root sentence as well as in the embedded sentence, so that it has the GCP setting 2e. In a picture identification task, Finer & Broselow found that their subjects bound 91.7% of reflexives to local antecedents in sentences like 6a, when the reflexive was inside a tensed embedded sentence. However, in infinitive sentences like 6b, 37.5% of the reflexives were bound long-distance, and only 58...