A Study On The Universality Of Subject Relatives P Reference: ERP Evidence From Chinese

The structural complexity and the typological universals associated with relative clauses (RC) have made this structure particularly interesting to linguists. Concerning RC processing, theoretically, Noun Phrase Accessibility Hierarchy (NPAH) declares a universal subject relatives preference because the subject is on the top of the accessibility hierarchy which reflects“the psychological ease of comprehension”(Keenan & Comri 1997:68). Similar to NPAH, Experience / Frequency-based accounts (Mitchell, Cuetos, Corley & Brysbaert 1995; Hale 2001) support this universality too due to the cross-linguistic fact that subject-extracted relative clauses (SRCs) tend to occur more frequently than object-extracted relative clauses (ORCs).Empirically, a wide variety of work explores a robust processing asymmetry such that SRCs are easier to process than ORCs. For example, English shows a subject advantage, as demonstrated by a number of studies involving different measures including: self-paced reading (King & Just 1991), eye-tracking (Traxler, Morris & Seely 2002), ERP (King & Kutas 1995), fMRI (Just, Carpenter, Keller, Eddy & Thulborn 1996); and PET (Stromswold, Caplan, Alpert & Rauch 1996).Many findings in the literature suggest that subject preference seems to be a universal processing phenomenon in RCs. Some pieces of evidence come from Dutch (Frazier 1987), French (Frauenfelder, Segui & Mehler 1980), German (Schriefers, Friederici & Kuhn1995), Japanese (Miyamoto & Naka-mura 2003) and Korean (Kwon, Polinsky & Kluender 2006).In order to account for RCs processing asymmetry, several models are proposed, such as working memory accounts of Storage Resources and Integration Cost (Gibson 1998), structure-based theories of Filler-Gap Domain (Hawkins 2004) and Phrase-Structural Distance Hypothesis (O’Grady 1997), pragmatic functional account of Perspective Shift (MacWhinney 1977, 1982; MacWhinney & Pleh 1988) and Word Order (Bever 1970; MacDonald & Christiansen 2002).Chinese is the only language with a combination of SVO word order and head-final property (Dryer 1992). Its unique typological feature allows testing this universal processing preference. Current models of sentence processing make contrasting predications, and existing research has found mixed results suggesting either a subject preference or an object preference.In the face of such controversial predictions from theories and contrasting results from previous experiments, this study aims to 1) investigate whether processing preference is universal for typologically distinct languages? 2) Whether subject preference is applicable to Chinese? 3) Concerning structures, Chinese has subject-modifying, object-modifying and topicalized RCs. How do the extraction site, syntactic role of head noun, and topicalized position guide real-time processing of Chinese RCs? 4) What sentence processing models are more universal for accounting for RC processing with different typological characteristics?The theoretical part of the dissertation presents typology of RCs, explaining why Chinese provides a good case for cross-linguistic comparison. Sentence processing models and previous general findings in RC processing are discussed, and the comments of each experiment are given in particular.The experimental part presents ERP evidence from Chinese RC processing with subject-modifying SRCs / ORCs, object-modifying SRCs / ORCs, and topicalized SRCs / ORCs as experimental stimuli. Statistics are done with three levels including testing previous results, involvement of topicalization effect and expanding above with minimal pair analysis. The results show a significant main effect of RC-type, and SRCs consistently elicit larger N400, P200 as well as P600, indicating that SRCs are less expected and more difficult to process. That is, in Chinese it is object relatives that are easier to process. Universal processing preference is not applicable to Chinese. The results also clarify the processing asymmetry of each RC type in Chinese. All of them show object preference, suggesting object preference is general for Chinese.The findings of this study claim that working memory accounts are more prevalent to explain processing, but they also cast doubts about it. In addition, this study proposes the function of canonical word order to alleviate difficulty caused by syntactic and semantic violation. Moreover, it hypothesizes an unmarked RC form both in English and in Chinese.All in all, this study is the one until now exploring universality of subject relatives preference with comparatively complete types of Chinese RCs. Empirically, it adds the evidence that processing preference is not universal. Theoretically, it contributes to a comprehensive model of how complex structures are processed.