漢語母語者與學習者對反身代詞「自己」的指涉研究──基於事件相關電位技術的證據

Abstract

當漢語反身代詞「自己」出現在結構如「張三認為李四喜歡自己」時，「自己」可以同時與「張三」建立指涉（長距離指涉，long-distance reference），也可以與「李四」同指（局部指涉，local reference），而前者的機制在許多語言中（如英語）是不存在的。為了揭示獲得上述兩種指涉時語言處理歷程的異同，以及學習者母語的機制對獲取長距離指涉造成的影響，本文使用事件相關電位技術（Event-related potential）收集漢語母語者及華語學習者閱讀帶長距離及局部指涉反身代詞「自己」的電生理反應，並進一步與語法錯誤致「自己」無法獲得任何指涉的句子相比（如「張三希望李四難過自己」）。結果發現，對於漢語母語者而言，閱讀無法獲得指涉的句子時，反映句法整合困難的P600效應的振幅大於解讀長距離及局部指涉句；解讀長距離指涉句時，P600振幅也較解讀局部指涉句大，這是由於長距離指涉的獲得必須跨越反身代詞的句法限制，並經過語用、篇章的訊息整合後才能獲得。此外，相較於局部指涉句，解讀長距離指涉句更產生了與頻率相關的P300效應，這反映了漢語反身代詞「自己」帶長距離指涉的頻率較帶局部指涉得低，因此在解讀時需要分配更多的注意力資源。而華語學習者的母語反身代詞沒有長距離指涉的機制，我們發現學習者處理長距離指涉時產生的P600效應振幅，與處理語法錯誤句的P600振幅皆與華語程度、指涉選擇的答對率及居住在官方語言為華語地區的時長這三個反映學習者語言能力的變項呈現正相關，而這三個變項與處理局部指涉時的振幅沒有相關性，說明學習者處理長距離指涉的機制較近似於處理語法錯誤句。雖然這三個變項的增加，皆與學習者長距離及語法錯誤句振幅成正相關，在處理語法錯誤句振幅的增幅卻多於處理長距離指涉的增幅，與母語者的腦電波資料結果一致，即，語法錯誤句的P600效應振幅最大，長距離指涉句次之，而局部指涉最末。這也說明隨著學習者語言能力的提升，學習者處理指涉的機制或許能越近似於母語者的處理機制。 本研究的結果除了支持漢語長距離及局部指涉的反身代詞「自己」為兩種具有不同功能的語言成分，對文獻中爭論不休的議題提供了實證性的證據外，同時，對於華語學習者而言，本研究的結果也說明現有的華語教材及課程對提升學習者長距離指涉的意識仍然十分有限。

This study aims to analyze and compare differences in the neural processing mechanisms for interpreting Chinese reflexive ziji. Chinese reflexive ziji is believed to have some properties different from the reflexives of other languages—while reflexives in most languages are only allowed to refer to the antecedents in their governing categories, Chinese reflexive ziji can be bound not only by the antecedent within its governing category (i.e., local reference) but also outside its governing category (i.e., long-distance reference). We assessed Event-Related Potentials (ERPs) from Chinese native speakers and L2 learners of Chinese while they read grammatical sentences containing ziji with long or local antecedents determined by a preceding verb as well as ungrammatical sentences containing ziji with no suitable antecedents at all. The results showed that for Chinese native speakers, ungrammatical sentences elicited larger P600 responses than long-distance or local reference conditions did. In addition, long-distance reference sentences also elicited larger P600s than did local reference sentences, suggesting syntactic integration difficulties from breaching the syntactic restriction of reflexive and integrating non-syntactic factors in the language, such as semantic, pragmatic, and contextual information. Furthermore, compared to processing the local reference, processing the long-distance reference of ziji elicits a P300 effect, reflecting that the frequency of Chinese reflexives with the long-distance reference is lower than that with the local reference. The P300 effect implies that Chinese native speakers allocate more attentional resources when interpreting long-distance references. Regrading Chinese L2 learners, their amplitudes of P600 responses in reading long-distance reference sentences and ziji in ungrammatical sentences were positively correlated with three variables that reflect the Chinese fluency of these learners: the number of verbs in the stimuli the participant understands, the accuracy rates of the antecedent selection questionnaire, and the time of living in areas where the official language is Chinese. These same variables, however, showed no correlations with the amplitude for local reference sentences. Additionally, although there are positive correlations between the three variables and the amplitudes of P600 effects for long-distance and ungrammatical sentences, the coefficient is much larger for the latter than for the former. Therefore, consistent with the data of native speakers, when these three variables are matched, the amplitude of the P600 effect is estimated to be the larger for processing ungrammatical sentences than interpreting long-distance reference sentences. Together, these results indicated that with improving learners’ language fluency, learners’ reference processing may become more similar to native speakers. The results of the study support the view that Chinese long-distance and local references involve nonidentical processing mechanisms and with distinct linguistic functions. In addition, the results of the L2 Chinese learners of the present study also suggest that the long-distance reference in reflexives may not be emphasized enough in existing Chinese teaching textbooks. Linguistic and explicit instructions on the local and long-distance reference interpretations may enhance learners’ effectiveness of comprehension Chinese reflexives.