L2 PROCESSING OF RELATIVE CLAUSES IN MANDARIN

Abstract

Studies investigating native processing of relative clauses in Mandarin Chinese have shown a processing asymmetry in terms of subject-gapped (SRCs) or object-gapped (ORCs) relative clauses. Unlike research that has been done in English, which revealed a clear SRC advantage in relative clause processing, results from the studies in Chinese did not reach a unanimous agreement. In this study, a self-paced reading task and an off-line questionnaire were conducted to examine the native and non-native processing of Chinese relative clauses (RCs) containing either a subject-gap or an object-gap, either at the matrix subject or object position of the sentences. Twenty-four native speakers of Mandarin (L1) and thirtytwo intermediate second language learners of Chinese (L2) participated in this study. The native languages of those L2 learners were either head-initial languages (e.g., English, Spanish, and French) or head-final languages (e.g., Korean and Japanese). Studies examining non-native processing of relative clauses in Mandarin Chinese have been limited, thus, research investigating how L2 learners of Chinese process relative clauses online can provide more information about whether L2 learners can construct abstract grammatical representations (e.g., filler-gap dependencies) as native speakers do in real-time processing. Results showed that, in answering the questionnaire, both L1 and L2 speakers found SRCs more difficult than ORCs. The reading time data showed that for native speakers, SRCs were read significantly more slowly than ORCs only when the RCs are modifying the matrix subject. Although the L2 data as a whole did not reveal a robust effect of gap site, dividing them into head-initial and head-final groups showed that learners from the head-initial group processed ORCs significantly more slowly than SRCs, irrespective of their positions in matrix sentences, which could be attributed to the result of L1 transfer. A main effect of the matrix position was also found for both the L1 and the L2 groups.