A study of the constraints affecting resumption in Turkish and Mandarin Chinese relative clauses, and the transfer of these constraints to English as a second language.

摘要

Odlin (2003) observes that there is no consensus among researchers regarding the importance of L1 transfer in second-language acquisition (henceforth SLA). To test whether L1 transfer is a significant factor in SLA of English relative clauses (RCs), an English-language acceptability judgment task (AJT) with a four-point rating scale was administered to 34 native English speakers, 33 native Turkish speakers, and 35 native Mandarin Chinese speakers.;The linguistic phenomenon under investigation was the acceptability of resumptive pronouns (RPs) in relative clauses, as in the following examples in bold:;a. Direct Object RC: *The relativesi [Swho we [ VPvisited themi last night]] enjoyed the evening.;b. Indirect Object RC: * The relativesi [swho we [VPpaid a visit [PPto themi]]] last night enjoyed the evening.;While such sentences are not grammatical in English, RPs are used in many languages, and also occur in the English of young children (McKee and McDaniel, 2001). For the current study, it was assumed that RPs were acceptable in Mandarin Chinese in non-subject RCs comparable to examples (1) and (2) above (Hawkins and Chan, 1997), but unacceptable in comparable Turkish RCs (Tezel, 1999). It was therefore hypothesized that L1 influence would cause L1 Mandarin Chinese speakers to accept RPs in non-subject positions on an English-language acceptability judgment task at a higher rate than L1 Turkish speakers.;The results of the English-language AJT showed that both the L1 Chinese and L1 Turkish groups accepted RPs at a slightly higher (but non-significant) rate than the native speaker group, but there were no differences between the two groups of non-native speakers. A likely reason for this result is that the originally hypothesized distributions of RPs in Mandarin Chinese and Turkish RCs were incorrectly estimated, as indicated by the results of a follow-up study in which Mandarin Chinese and Turkish-language AJTs were administered to 32 L1 Mandarin Chinese speakers and 16 L1 Turkish speakers, respectively. Results of these two AJTs demonstrate that in Mandarin Chinese, RPs were less acceptable in non-subject positions than originally hypothesized, while in Turkish, RPs were more acceptable in non-subject positions than expected. Thus, the empirical basis for the original transfer hypothesis is seriously challenged by the results of these two follow-up experiments.;The current practice in SLA studies of language transfer is to consult previous studies, which typically rely on individual linguists' intuitions about the acceptability of L1 syntactic structures. The current study shows that such studies may be fallible, and that future researchers interested in language transfer should take pains to design sentence judgment tasks in both the L1 as well as the L2.;Despite the lack of a transfer effect, these data are of interest because they indicate the possible influence of language processing constraints on grammatical constraints. In the English-, Turkish-, and Chinese language AJTs, RPs were rated as more acceptable in syntactically more complex non-subject RCs as compared with the syntactically simpler subject RCs, results that are fully consistent with John Hawkins's (2004) Performance-Grammar Correspondence Hypothesis.