| 摘要: | 先前的研究闡明，當雙語使用者在兩種語言之間做轉換時，會造成轉換上的耗損。其中，有兩種假說針對語言轉換時所造成的耗損提出了可能的解釋，其一是抑制控制假說（Green, 1998），另一個則是雙語相對激活假說 (Grainger & Dijkstra, 1992; van Heuven, Dijkstra & Grainger, 1998)。當前的研究目的主要是希望能夠探討當拼音文字 （即：英文）與象形文字（即：中文）此兩種語言在轉換的過程中，其語言轉換所造成的耗損是與雙語相對激活假說 (Grainger & Dijkstra, 1992; van Heuven, Dijkstra & Grainger, 1998) 所提出的與語言獨特的編碼有關，抑或是與抑制控制假說 (Green, 1998) 所提出的與一般性作業基模相關。在實驗中，採用了隱蔽促發刺激的範式(masked priming paradigm)，其觸發刺激(prime)的呈現時間在實驗一與實驗二分別為100 毫秒與30 毫秒，並且，在兩個實驗中各自有二十位的中英雙語使用者參與實驗。在實驗中的觸發刺激(prime)有三種不同的類型，分別為中文，英文以及韓文隨機出現。實驗中，受試者需要在中文與英文為主要刺激(target)的兩個組別中皆進行語意判別(semantic categorization judgment)作業。實驗當中，利用事件相關電位(eventrelatedpotential)分別蒐集在中文與英文為主要刺激(target)時，中文、英文以及韓文三種觸發刺激(prime)其腦電波變化。實驗一中，不論是在中文或是英文為主要刺激(target)的組別中，英文觸發刺激(prime)相較於中文與韓文而言，引發了較強的N/P150。接著，韓文在中文與英文為主要刺激(target)的組別中，相較於中文與英文的觸發刺激(prime)，引起了較強的N250。當主要刺激(target)出現後大約250 毫秒，我們發現了編碼轉換的效果(code-switching effect)。此效果在兩種不同的主要刺激(target)的組別ii中皆發現，當觸發刺激(prime)與主要刺激(target)為不同語言時，相較於當觸發刺激(prime)與主要刺激(target)為同種語言時，引發了較強的P325。再者，英文觸發刺激(English prime)在中文為主要刺激(Chinese target)的組別中引發了較大的N400。然而，在英文為主要刺激(English target)的組別中，中文觸發刺激(Chinese prime)相較於英文與韓文而言，則沒有引發較大的N400。在實驗二中，當觸發刺激(prime)的呈現時間在閾值(threshold)之下時，發現當觸發刺激(prime)與主要刺激(target)為兩種不同語言時(即：英文的觸發刺激配上中文的主要刺激，與中文的觸發刺激配上英文的主要刺激組別)，相較於當觸發刺激(prime)與主要刺激(target)為同種語言時(即：中文的觸發刺激配上中文的主要刺激，與英文的觸發刺激配上英文的主要刺激)，引發了較大的N1 編碼轉換效果(code-switching effect)。在主要刺激(target)出現後的170 至290 毫秒之間，在中文為主要刺激(target)的組別中，韓文相較於中文與英文，引發了較正極走向的波。另一方面，在英文為主要刺激(target)的組別中，英文相較於韓文與中文，產生了較正極走向的波。 在中文與英文為主要刺激(target)的組別中，皆沒有發現N250 編碼轉換效果(code-switching effect)。另一方面，在英文為主要刺激(target)的組別中，發現中文為觸發刺激(Chinese prime)時，相較於英文觸發刺激(English prime)，引發了較強的N400 編碼轉換效果(code-switching effect)。在實驗一與實驗二中發現的編碼轉換效果(code-switching effect)暗示了在文字辨識歷程中，語言獨特編碼的自動調節以及一般性作業基模的控制，在拼音文字與象形文字做語言轉換時，皆會造成語言轉換的耗損，而此效果發生在不同階段的語言處理歷程。Previous studies have demonstrated that changing between languages inbilinguals incurred switching costs. Two models, the inhibitory control (IC) model(Green, 1998) and the bilingual interactive activation (BIA) model (Grainger &Dijkstra, 1992; van Heuven, Dijkstra & Grainger, 1998), have been proposed toaccount for such costs. The aim of the present study was to investigate whether thelanguage switching costs between one alphabetic (English) and one logographic(Chinese) language are caused by a language-specific code as proposed in the BIAmodel or by the general task schema as proposed by the IC model, or by themechanisms proposed by both the hypotheses. A masked priming paradigm wasadopted in two experiments with the prime duration in Experiment 1 and 2 to be100 ms and 30 ms, respectively. Twenty unbalanced Chinese-English bilinguals wererecruited in each experiment. Following a prime in one of the three languages(Chinese, English, and Korean), participants performed a semantic categorizationjudgment on a Chinese or English target word. Event-related potentials (ERPs) thatwere elicited by the target following a between-language, a within-language, or a nonlexical(i.e., Korean) prime were recorded.In Experiment 1, the English prime elicited a stronger N/P150 component ofthe prime than the Chinese and Korean prime in both the Chinese and English targetblocks. Only the Korean prime that are unknown to participants elicited a strongerN250 component of the prime in both target blocks. The code-switching effect wasdetected in the ERP components after 250 ms from the target onset. Specifically, thebetween language and the non-lexical conditions elicited a larger P325 componentthan the within language conditions in both target blocks. Moreover, the Englishivprime induced a larger N400 component of the Chinese target word than the Chineseand the Korean prime, though the Chinese prime did not induce a larger N400component of the English target than the English and then Korean prime.In Experiment 2 when the prime was subliminal, the code-switching effectwas first revealed in the N1 component in showing more negative going waveforms inthe between-language conditions (i.e., English prime-Chinese target, Chinese prime-English target) than in the within-language conditions (i.e., Chinese prime-Chinesetarget, English prime-English target). In the time window of 170-290 ms after thetarget onset, the Korean prime induced more positive going waveforms than theEnglish and the Chinese prime in the block of Chinese target words. In the block ofEnglish target words, however, the English prime was more positive than the Koreanand the Chinese prime. The N250 code-switching effect was not found in both of thetarget blocks. The N400 code-switching effect was revealed only in the block ofEnglish target words in showing more negative ERP waveforms in the betweenlanguagecondition (i.e., Chinese prime-English target) than in the within-languagecondition (i.e., English prime-English target).Taken together, the code-switching effects found in Experiment 1 and 2 suggest that the language switching costs between one alphabetic and one logographic language might result from both the automatic modulation of the language-specific code during the visual word recognition process and the control of the general task schema. These effects occur at different stages of linguistic processing and are not mutually exclusive. |
