句尾非預期效應對語句處理的影響:事件相關腦電位研究
No Thumbnail Available
Date
2013
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
在言談交際當中,人們常常會預測對方欲表達的概念或用詞。過去研究指出,當人們的預期和實際情況不相符時,句子理解起來會比較費力;然而,過去研究並未特別區分,在一個句子當中出現其他語言,或是出現非預期的概念,在理解歷程上是否造成不同的困難。此外,在雙語社會中,常常有語碼交雜(code-mixing)的現象,以台灣社會為例,生活中常常出現國語轉台語、台語轉國語的情況,使得語言轉換方向成為有趣的研究議題。本研究旨在以事件相關腦電位技術,探討句尾非預期效應對語句處理的影響,以及不同的語言轉換方向是否在理解上造成差異。實驗以克漏字機率(cloze probability)高低來操弄句尾關鍵字的預期程度。句子以整句國語、整句台語、國語轉台語、台語轉國語的方式呈現。受試者是以國台語雙語為母語的人士,其國台語皆相當流利,但是較常使用國語。正式實驗中,受試者必須專心聆聽並理解句子的意思。實驗結果發現,相較於符合預期的情況,低克漏字機率(low-cloze words)與語碼交雜的情況皆引發N400反應,顯示兩者皆在語義整合上發生困難。然而,就N400的分佈而言,低克漏字機率的N400反應主要在中間偏後的腦區,語碼交雜的N400則是全腦的反應,顯示處理兩者的機制並不相同。此外,低克漏字機率的情況在前額葉的地方可以觀察到晚期正向波(late frontal positivity),可能與受試者預期的字和實際出現的字有概念上的落差相關。在語言轉換方向上,只有國語轉台語的句子引發N400,顯示由較常使用的語言轉換到較不常使用的語言時,會有語義提取或是語義整合的問題。
People generate predictions when processing linguistic information. Although it was generally agreed that there were processing benefits for predicted words and costs for unpredicted words, the difference/similarity between an unexpected “concept” and an unexpected “language” was never systematically explored. The aims of the current study were twofold. We first investigated how people processed sentences with an unexpected ending, specifically sentences that were terminated by an unexpected concept and sentences that were completed in a different language. We then examined whether there were processing differences between sentences switched between languages of different dominance. The expectancy of the concept of a sentence terminal word was determined by its cloze probability. The sentence context was in Mandarin (the dominant language) or in Taiwanese (the less dominant language), with the terminal word either kept in the same language as its context or switched into the other language. The participants were 21 native Mandarin-Taiwanese bilinguals, who were proficient in both languages but more dominant in Mandarin. They were instructed to listen to sentences for comprehension. The ERP results revealed that relative to the expected conditions, both the low-cloze and language-switched conditions elicited an N400 effect, showing that in both situations, participants might have problems in lexical integration. However, the scalp distributions of N400 in the two conditions were fairly different. While low-cloze related N400 was larger over parietal sites, language-switching related N400 had a broad scalp distribution, suggesting that they were processed by different underlying mechanisms. In addition, only the low-cloze conditions elicited a late frontal positivity, which could reflect a conceptual mismatch. As for switching directionality, while switching from Mandarin to Taiwanese elicited an N400 effect, this effect was absent in the other way round. The asymmetric results demonstrated that the participants might encounter greater difficulties in lexical access/integration when switching into a less dominant language.
People generate predictions when processing linguistic information. Although it was generally agreed that there were processing benefits for predicted words and costs for unpredicted words, the difference/similarity between an unexpected “concept” and an unexpected “language” was never systematically explored. The aims of the current study were twofold. We first investigated how people processed sentences with an unexpected ending, specifically sentences that were terminated by an unexpected concept and sentences that were completed in a different language. We then examined whether there were processing differences between sentences switched between languages of different dominance. The expectancy of the concept of a sentence terminal word was determined by its cloze probability. The sentence context was in Mandarin (the dominant language) or in Taiwanese (the less dominant language), with the terminal word either kept in the same language as its context or switched into the other language. The participants were 21 native Mandarin-Taiwanese bilinguals, who were proficient in both languages but more dominant in Mandarin. They were instructed to listen to sentences for comprehension. The ERP results revealed that relative to the expected conditions, both the low-cloze and language-switched conditions elicited an N400 effect, showing that in both situations, participants might have problems in lexical integration. However, the scalp distributions of N400 in the two conditions were fairly different. While low-cloze related N400 was larger over parietal sites, language-switching related N400 had a broad scalp distribution, suggesting that they were processed by different underlying mechanisms. In addition, only the low-cloze conditions elicited a late frontal positivity, which could reflect a conceptual mismatch. As for switching directionality, while switching from Mandarin to Taiwanese elicited an N400 effect, this effect was absent in the other way round. The asymmetric results demonstrated that the participants might encounter greater difficulties in lexical access/integration when switching into a less dominant language.
Description
Keywords
事件相關腦電位, 語句處理歷程, 克漏字機率, 語碼交雜, 語言轉換方向, N400, late frontal positivity, ERPs, sentence processing, cloze probability, code-mixing, switching directionality, N400, late frontal positivity