語言轉換與語言混合之功能性磁振造影研究

全文

(1)國立臺灣師範大學英語學系碩. 士. 論. 文. Master’s Thesis Department of English National Taiwan Normal University. 語言轉換與語言混合之功能性磁振造影研究 The Processing of Language Switching and Language Mixing: An fMRI study. 指導教授：詹曉蕙博士 Advisor: Dr. Shiao-hui Chan 研究生：江欣粦 Student: Shin-Lin Kong. 中華民國一百零六年六月 June 2017.

(2) 摘要對於雙語及多語人士可以輕鬆地進行語言轉換(language switching)及語言混合 (language mixing)的現象一直是語言學家感興趣的議題，其背後的神經機制更是神經語言學家所感興趣的。本研究旨在探討健康成人在語言轉換及語言混合上的神經機制是否有差異，以補足先前文獻對此議題的探討。此外，本研究也進一步探討語言轉換及語言混合的方向性(從優勢的語言(L1)至非優勢語言(L2)或相反)在處理歷程上的差異。此實驗採用功能性磁振造影(fMRI)技術的區塊設計(block design)，操弄三個因子：語言(中文及英文)、語言轉換或語言混合以及語言轉換的方向。23 位中文為母語、英文流利的受試者參與此實驗。正式實驗中，受試者必須在 MRI 機器內仔細聆聽所播出的音檔。實驗結果發現語言轉換及語言混合活化的腦區非常類似，尤其在顳葉顳上迴(superior temporal gyrus)及額葉腦迴區(inferior frontal gyrus)上的活化最強，研究結果與先前探討病人上所得到的結論似乎相反。實驗結果進一步顯示有語言轉換及混合的操弄相較於無任何操弄的情況下的腦區更為活化，這表示受試者在處理語言轉換及混合的句子下需要更多的認知功能處理。在語言轉換方向上，大腦活化從中文轉至英文比起英文轉至中文來得強。此結果顯示從優勢的語言（L1）轉換/混和至非優勢語言（L2）在處理歷程上是比較困難的。總結，此研究顯示語言轉換及語言混合在大腦活化上並無明顯差異，但在語言轉換的方向上具有顯著差異。關鍵字: 語言轉換，語言混合，語言轉換方向，語句處理歷程，功能性磁振造影. i.

(3) Abstract Bilinguals and multilinguals often switch and mix languages with automaticity, but was it really that easy? The aims of the current study were to investigate the underlying neuronal mechanisms for switching and mixing in healthy individuals as few study attempts to differentiate between these two phenomena. We then examined the processing cost of language switching and mixing in both directions (L1 to L2 or L2 to L1). A block fMRI design was adopted and the experimental materials were in Mandarin or in English, with the manipulation of switching or mixing and also the switching/mixing directions. Twenty-three proficient Mandarin-English bilinguals were recruited. They were instructed to listen to the auditory stimuli inside the MRI machine. The results revealed that the comprehension of language switching and mixing sentences activated very similar brain regions, with fairly strong activations in superior temporal gyrus (STG) and inferior frontal gyrus (IFG), which was drastically different from their clear distinction in language production in patient research. However, higher activations were found in the switched and mixed conditions compared to the non-switched/mixed conditions, indicating that higher cognitive effort was required in processing the switching and mixing conditions. As for switching direction, switching/mixing from Mandarin to English induced stronger brain activation than switching/mixing from the other way around, revealing that higher processing cost was required to process a switch/mix from a more dominant to a less dominant language. In sum, ii.

(4) the study showed that switching and mixing developed similar activity in the brain, while switching/mixing directions elicited different processing cost.. Key words: language switching, language mixing, switching direction, sentence processing, fMRI. iii.

(5) Acknowledgments 感謝在我論文生涯中協助過我的各位。感謝啟蒙老師，何德華教授，在我就讀於國立中正大學時因上了一堂社會語言學，進而得到了很多的鼓勵及啟發，讓我對語言學產生了濃厚的興趣，並且決定推甄念語言學研究所。感恩可以成為詹曉蕙教授的指導學生，並且在您的細心指導下，在這本論文上提供了許多方向及建議，讓我對神經語言學這門學問更加的認識。另外，也非常謝謝曉蕙老師在我很沮喪及焦慮時，總會溫柔地給予給我正面能量把我從低谷中拉起，讓我重拾信心繼續往上衝！非常謝謝曉蕙老師！此外，感謝口委老師們，蘇席瑤教授及郭文瑞教授。謝謝老師們提供了很多的建議，讓此論文可以更完整！另外，也感謝所有在我課堂上協助及鼓勵過我的教授們，陳純音教授、李臻儀教授、謝妙玲教授、吳曉虹教授、林蕙珊教授、颯楊教授，謝謝您們的教導，讓我認識語言學裡的不同領域。另外，也謝謝陳浩然教授在我碩士這段時期所提供的工讀機會，讓我學習到了很多寶貴的經驗。感謝實驗室的夥伴們，林耿育，謝謝你在我進行實驗時的任勞任怨，從實驗設計、到 MRI 儀器上的設定及經驗分享、SPM 上的分析、瑜珈好夥伴、一起經常為非常冷靜的夥伴們暴怒 and many more…謝謝你在我每次很焦慮地質疑自己能不能畢業時都很堅定地對我說“妳可以的，妳一定可以完成的！＂謝謝你 Ken！親愛的 Lilian，每次和妳聊天都覺得心暖暖，讓我得到了滿滿的能量！感恩妳在最忙碌的時候還幫我錄音、來實驗室一起吃飯聊天、一來實驗室就要準備會笑到練腹肌…現在要到畢業了，想到這些場景也要隨我而遠離，就有種默默的感傷…但！這些美好回憶，都記在腦海裡了吧? 以後 iv.

(6) 有空要常常拿出來溫習喔！Dear Matt，謝謝你很神奇的氣場，一次又一次幫忙解除危機，桌電都會因為有你而變得正常運作！以後不需要放“乖乖”餅乾，麻煩 Matt 過來就危機解除！謝謝你一直在論文上協助我解決一些統計問題還有陪我聊天，打擊壞心情！你也是我的瑜珈好夥伴，去美國後要繼續做瑜珈喔！親愛的雙胞胎姊姊 Lucy，想不到你在我論文口考當天竟然可以出現！我覺得超級無敵 super 感動，很開心可以和你一起分享當天的喜悅，一起吃熱炒及被蚊子叮。另外，特別感謝你在先前的句法學課程上的幫助，沒有你的 guidance，我可能還依舊很迷茫…還有！你也是我瑜珈的好夥伴，一直翻白眼翻到後腦的好棒友！在美國要好好照顧好自己喔！等我去找你～Dear Gracie，還記得與你相遇的第一次 meeting，妳一直對我說 “不管如何都要說是西瓜就對了，西瓜族很團結”。那時候聽得霧煞煞，後來知道其歷史後，很開心自己 officially 也是西瓜族一份子！而且很懷念妳常常出現在實驗室的歡樂時光！很期待妳學成歸來的日子，當個很棒的教授！親愛的 Julia，我終於明白致謝詞為何可以寫這麼多頁，因為要感謝的人太多了！寫著就浮現了好多需要道謝的人！對於你，我也不例外，或許先來個大擁抱才能表達我對你的感謝。感謝妳再挑選 materials 上，真的非常 super 仔細地幫我檢查及審閱，甚至還給予很多其他的例子讓我參考。在這非常 tricky 的步驟，謝謝妳成為了我雪亮的第二雙眼睛！而且，妳經常會很窩心的來問我狀況如何，這都讓我心暖呼呼到不行～還有，妳也非常的用心及積極的幫忙分享我實驗的連結，從自己臉書到英語系學會等等的平台，而且還會在幾天後幫我在貼文底下回覆，讓此貼文“浮”出水面增加曝光率。這部分多虧有妳這科技達人，我才能順利的在短時間內收集到足夠的人數，真的是謝謝有妳！我畢 v.

(7) 業後有空還要一起吃飯聊天聽妳講故事～親愛的 Helen 女神(我會努力不被損友誤導加個“姐”)，感恩妳在我提論文大綱前，仔細的教我要如何 present，讓我擁有了 100 分的自信。也非常謝謝妳忙碌中還會發個短訊詢問大家的狀況，及預祝我口考順利，謝謝像太陽一樣暖的 Helen！Dear Terry，你真的太忙了啦！好像快 2 年沒看到你了，雖然近期內看到的時候是我帶著 freshly wounded 的手趕著去輔大考中文檢定考試，真的很扯的偶遇(笑)。謝謝你之前帶我去體驗田野調查，解決我在句法學上困難，分享了很多人生經驗，和你當朋友真的是受益(損?)良多。要繼續一起加油！工作真的不要累壞了！親愛的 Vivi，謝謝妳在我論文期間也幫忙我填寫問卷，還有幫忙 PO 文，也很開心我口試結束後也特地抽空過來吃熱炒小慶祝！每次聽你講話都得到滿滿的歡笑，還有也會知道哪裡有好吃的！以後有機會要繼續跟著妳超強美食雷達到處吃好吃的喔！也感謝實驗室的 Ronald，博班加油！期待參加你的畢業聚餐！你一定可以的！理克，恭喜你畢業！也歡迎你來到師大念博班！還有 Natasha，謝謝你幫我分享招募受試者的連結，也歡迎加入神經語言室大家庭！實驗室有滿滿的正能量，你會很開心的！謝謝系辦的慕涵助教，還記得有次我為論文愁眉苦臉的對你說“完蛋了”，妳當時的反應讓我哭笑不得，還反問我是老師說我完蛋了嗎？這一問有點醒我，在大家還沒放棄妳時，請給自己多點信心相信自己！也謝謝妳盡責的協助每位研究生，不厭其煩地一直解釋及想辦法解決問題。謝謝您，慕涵！謝謝羽立助教，看到你都非常的忙碌於工作上，即使多忙也好遇到你還是面帶笑容跟我打招呼小聊天！多次從助教 FB 分享的資訊得到滿滿的啟發與共鳴，還有與役男的互動總是覺得很歡樂！希望助教辛苦工作之餘也要好 vi.

(8) 好的休息放鬆！另外，也謝謝仲茵助教、卉喬助教、姿儀助教、豪谷助教、易珊助教、及淑慧助教在我系辦工讀時期的幫忙，讓我工讀得很開心！感謝碩士一起熬過來的同學們，Sally、Amber、Pris、Aries 及 Treak。謝謝你們一直以來的陪伴及互相扶持鼓勵，在我課業上及論文上幫了很多，如今碩班熬過去了，真的很感謝這一路都有你們！因為一場“座位糾紛”的事情而認識的同學，林揚傑，感謝你在我缺少受試者時很阿莎力的來參加實驗。去我家鄉，檳城時也不忘記寄送給我卡片！你以後一定會是 100 分的教師，台灣學生的教育拜託你了！感謝陳怡廷及妳的好麻吉廖姝安，謝謝你們來參加我的實驗！也恭喜妳畢業了！有機會要一起去找翊茜玩！感謝學弟妹，Andrew 和 Eliza，成為我第一位及第二位受試者！謝謝你們願意進行 fMRI 實驗及鼓勵朋友們來參加實驗！也要感謝 Vivien、Amy 幫我 PO 文邀請朋友來參加！感謝所有幫忙進行實驗及轉 PO 文的朋友、學長姐、學弟妹及熱心的陌生人！謝謝你們！感謝 Christine 及 Ian 在我兼職的工作上，給予了我很多的協助及指導。讓我可以從中學習到很多，也可正面看待及磨合自己的工作能力。謝謝你們的所有幫助及支持！謝謝廖筱慧、胡瑄、黃子嘉、馮娪貞在我兼職的工作底下，非常盡心盡力的幫忙！如果沒有妳們迅速的協助，很多事情或許就不會那麼的圓滿了。謝謝妳們！感謝女一分舍寢室 ex-好室友們，Miss Liew、文麗、陳瑩容&容媽 boy (Alan)、Hui Ling、Summer。謝謝你們，讓我碩士生活過得多采多姿，一起吃飯聊天歡呼尖叫哀哀叫。偶而也會一起解鄉愁，吃馬來西亞家鄉飯菜，當然也有台灣美食，好想再去苗栗容媽家吃石頭燒烤！很珍惜與妳們的緣分，以後不管在哪裡，都一定要保持聯繫！不准突然消 vii.

(9) 失！感謝馬來西亞的朋友們，Yi Ting 姐、Pei Hoon、Kai Lin、Eunice Wong、Yammy Ng、 Mindy Lee、Soo Khim、Sok Fen、Yeong Huey、You Qian、Bae Huey、Xin Yuan、Yean Sin，謝謝你們一直在馬來西亞給予的鼓勵與支持！感謝梁育绮學姊、梁詩詩學姊，一開始在人生地不熟的台北，謝謝有妳們帶我到處趴趴走，去品嘗了很多美食和景點！如今碩士生涯即將結束的此刻，一定要感謝妳們！相信我們以後會常常聯絡見面的！感謝廖翊茜，廖溫立叔叔、廖媽媽、廖姊姊、廖弟弟。謝謝廖翊茜在我碩士生涯中，帶我去很多地方玩，還有與我爸媽一起去歐洲！這一切回想起來都有點不可思議，很開心與妳共創了許多美好回憶。還有謝謝你不時傳來加油打氣的訊息及分享照片，即感動又舒壓！謝謝廖溫立叔叔很熱情的煮好吃的及送我水果！有機會一定會再去叔叔家玩！感謝馮倚俊，在我許多的人生低潮中陪伴並且支持我。還有帶我去吃很多美食、運動、散心，真的很感恩有你。感謝馮清峰叔叔、劉文鈴阿姨，馮家及陳淑卿阿姨，在我撰寫論文這段時間，帶了很多好吃的水果及請我吃好吃的，並且關心我的論文進度。感恩在天上的阿公、阿嬤、外公、外婆，謝謝你們一直以來的照顧。哥哥們，謝謝你們如此的努力，給我這小妹立下了好榜樣及無懼的開拓了自己的人生藍圖，這些都讓我也鼓起了勇氣也跟隨您們的腳步，一起努力及進步，為社會貢獻。最後，感謝爸爸、媽媽，我至今成功拿到碩士學位，所有的功勞都要給您們，謝謝您們的辛苦付出讓我可以安心地唸完大學及碩士。謝謝您們都願意聆聽我的心聲，讓我自主的跟隨自己的夢想並且給予我無限的支持與愛。這論文，獻給您們。 viii.

(10) Table of Contents 摘要................................................................................................................................. i Abstract .........................................................................................................................ii Acknowledgments ....................................................................................................... iv Table of Contents ........................................................................................................ ix List of Tables ............................................................................................................... xi List of Figures .............................................................................................................xii Chapter One: Introduction ......................................................................................... 1 1.1. Motivation ............................................................................................................... 1. 1.2. Research Questions ................................................................................................ 4. Chapter Two: Literature Review ............................................................................... 5 2.1. Language Switching and Mixing: a Linguistic Perspective ................................ 5. 2.2. Language Switching and Mixing: a Cognitive Neuroscience Perspective ......... 9 2.2.1 Language Switching .................................................................................... 12 2.2.2 Language Mixing ......................................................................................... 18 2.2.3 Coexistence of Pathological Switching and Mixing .................................. 24. 2.3. Switching Direction and Cognitive Control ....................................................... 33. Chapter Three: Methods ........................................................................................... 41 3.1. Participants ........................................................................................................... 41. 3.2. Materials................................................................................................................ 41. 3.3. Procedure .............................................................................................................. 45. 3.4. Data Acquisition ................................................................................................... 46. 3.5. Data Analysis ........................................................................................................ 47. Chapter Four: Results ............................................................................................... 49 4.1 Behavioral data ............................................................................................................ 49 4.2 Neuroimaging Data ...................................................................................................... 50 4.2.1 Switching and Mixing .................................................................................. 50 4.2.2 Switching/Mixing Direction ........................................................................ 54. Chapter Five: Discussion and Conclusion ............................................................... 62 5.1 Language Switching and Language Mixing .............................................................. 62 ix.

(11) 5.2 Switching and Mixing Direction ................................................................................. 67 5.3 General Discussion ....................................................................................................... 69. References ................................................................................................................... 71 Appendix I: The online frequency-rating questionnaire for the experimental stimuli ... 90 Appendix II: A list of experimental stimuli with 6 conditions (Non-switched-MM, Non-switchedEE, Switching-ME, Switching-EM, Mixing-ME, Mixing-EM) ...................................... 91. x.

(12) List of Tables Table 1: Types of Pathological Mixing Phenomena............................................................ 18 Table 2: Neuroimaging Studies Investigating Pathological Switching and Pathological Mixing in Bilinguals and Multilinguals ............................................................................... 28 Table 3: Neuroimaging Studies Investigating Language Switching in Bilinguals and Multilinguals........................................................................................................................... 29 Table 4: Neuroimaging Studies Investigating Language Mixing in Bilinguals and Multilinguals........................................................................................................................... 31 Table 5: A set of Example Stimulus (MM: Mandarin-Mandarin; EE: English-English; ME: Mandarin-English; EM: English-Mandarin) ............................................................. 42 Table 6: Summary of the Accuracy Rates(ARs) and Reaction Times (RTs) in Each Experimental Condition ........................................................................................................ 50 Table 7: Brain areas activated in Non-switched/mixed, Switching and Mixing conditions .................................................................................................................................................. 53 Table 8: Language effect ....................................................................................................... 55 Table 9: The activation regions comparing the language switching in both directions with non-switching conditions .............................................................................................. 59 Table 10: Directionality effect in mixing vs. non-mixing comparisons ............................. 60. xi.

(13) List of Figures Figure 1: Brain areas that are responsible for pathological switching and mixing ......... 10 Figure 2: The Inhibitory Control Model (IC), proposed by Green (1998) ....................... 34 Figure 3: The Bilingual Interactive Activation Plus model (BIA+), proposed by Dijkstra and Van Heuven (2002) ......................................................................................................... 36 Figure 4: The Presentation of Two Trials in a Block (the Mandarin to English Switching Condition) ............................................................................................................................... 46 Figure 5: An illustration of the activation pattern while perceiving non-switched/mixed, switching and mixing sentences ............................................................................................ 51 Figure 6: The activation patterns in the comparisons of Switching vs. no switching and Mixing vs. no mixing conditions ........................................................................................... 52 Figure 7: The language effect of Mandarin and English conditions ................................. 54 Figure 8: The activated regions in the non-switched/mixed MM, non-switched/mixed EE, Switching EM, Switching ME, Mixing EM and Mixing ME conditions ................... 56 Figure 9: The first contrast (colored in red), Switching EM vs EE, overlay with the second contrast (colored in yellow), Switching ME vs MM ............................................... 58 Figure 10: Mixing direction: a figure portrayed the activation pattern in the comparison between ME vs. MM sentences ............................................................................................. 60. xii.

(14) Chapter One Introduction 1.1 Motivation Bilinguals and multilinguals possess amazing abilities to switch or mix languages to convey their messages to their target audience. Language switching can be defined as the alternation of two or more languages across sentences or clause boundaries, while language mixing is the use of two languages within one sentence. How the bilinguals’ and multilinguals’ brains work by allowing them to switch and mix languages automatically and interchangeably has been an interesting issue. Studies on bilingualism have been intensively conducted and lead to some major findings, such as the parallel activation of the bilingual’s two languages (Kroll et al., 2014), enhancement of cognitive abilities of bilinguals while code switching and language mixing (Green & Abutalebi, 2013; Green & Wei, 2014) and delayed onset of Alzheimer’s disease on bilinguals compared with monolinguals (Bialystok et al., 2007). The above studies all point to the benefits that learning more languages and applying them in daily conversations bring more good than harm to the speaker. However, what if a person has involuntary and uncontrolled language switching and mixing while communicating with an interlocutor who doesn’t know and understand any of those languages? 1.

(15) The increase of clinical cases for pathological switching and mixing has caused an upsurge of attention to medical authorities and Computed Tomography (CT), Magnetic Resonance Imaging (MRI), and functional Magnetic Resonance Imaging (fMRI) techniques are frequently used to identify the lesion(s) in the brain region and causes for the pathological outbreaks. Although it is not always possible to draw a clear distinction between mixing and switching, Fabbro et al. (2000) reported that pathological switching might reflect impairment centered in the pragmatic system independent of language aspects, whereas pathological mixing tends to be related to the linguistic system. Moreover, lesions in specific brain regions might cause pathological switching and mixing in brain-damaged individuals. As suggested by Fabbro’s clinical studies, it was found that pathological switching between languages mainly involved the lesions of the frontal lobe (left and right) (Fabbro, 2001a) and left anterior cingulate cortex (ACC) (Fabbro, 2001a), while persistent mixing of elements from languages were due to left parieto-temporal and left postrolandic lesions (Fabbro, 1999, 2001a). A few fMRI studies with pathological patients have revealed that similar brain regions (left frontal lobe, left ACC, and left basal ganglia) are activated when they comprehend language switching and mixing sentences (Mariën et al., 2005; Abutalebi et al., 2007; Adrover-Roig et al., 2011). Based on these studies, the current study adopted the fMRI technique to further explore the underlying neuronal mechanisms for switching and mixing in healthy individuals as few study attempts to differentiate between these two phenomena. In addition to 2.

(16) differentiating between language switching and mixing, the current study also looked into switching direction. Studies have shown that the cost of processing was asymmetric in switching directions (L1 to L2 or L2 to L1) (Meuter & Allport, 1999; Alvarez et al., 2003; Liao & Chan, 2016). The current study discussed two models, the Inhibitory Control Model (IC model) as proposed by Green (1998) and the Bilingual Interactive Activation Plus model (BIA+ model) (Dijkstra & Van Heuven, 2002), which provide two different proposals on the asymmetric cross-language influence in the language switching or mixing direction. In IC, the inhibition of L1 while performing a task in L2 is stronger as the dominant language has greater lemma activation and thus needs more effort in inhibiting; therefore, when switching back to L1 (i.e. an L2-L1 switch), it is more difficult to reactivate the inhibited L1 representations. In contrast, BIA+ proposes that the frequently used language (L1) would have higher resting-level activation and greater cross-linguistic influence from L1 to L2 exists. Therefore, when comprehending a sentence, an L1-L2 switch is harder since activating a representation in L2 needs more effort. This study hopes to unravel the cost of processing through the manipulation of the direction of language switching or mixing (from a more dominant L1 to a less dominant L2 or the other way round) by using the fMRI technique.. 3.

(17) 1.2 Research Questions The purpose of this study is to investigate the biological foundations of language switching and mixing sentences in bilinguals and multilinguals. By using the fMRI technique, this study aims to compare and determine the brain regions that are responsible for language switching and mixing in healthy individuals. Below summarizes the research questions: 1) Can language switching and mixing be distinguished at the biological level? In other words, can we observe linguistically induced involvement of the bilateral superior temporal gyrus (STG) and inferior frontal gyrus (IFG) for both language switching and mixing while a more pragmatically induced activations in subcortical-cortical networks in language switching? 2) Since switching or mixing implicates cognitive control, does the direction of language switching or mixing (L1 to L2 or L2 to L1) create any differences in processing cost?. 4.

(18) Chapter Two Literature Review 2.1 Language Switching and Mixing: a Linguistic Perspective Language switching and mixing are two distinct phenomena, which correlate to different expressive behaviors. It is in general agreement that language switching is inter-sentential, the alternation of languages across sentences or clause boundaries, while language mixing is intrasentential, the incorporation of linguistic units (e.g. word, syllable, morphemes, phonemes) from one language (e.g. L2) into a sentence or a sequence of sentences (e.g. narrative discourse or conversation) in another language (e.g. L1) (Sridhar & Sridhar, 1980; Bokamba, 1989; Kamwangamalu, 1992; Myers-Scotton, 1993; Fabbro, 2001b). The examples for the intersentential and intra-sentential switching are collected from Lyu et al. (2015) on MandarinEnglish bilingual are illustrated as follows (to differentiate between two languages, one of the languages is represented in italic format) 1. Inter-sentential switching Mandarin-English bilingual: Lyu et al., 2015 很容易罷了。Put the pumpkin in, fried it then add the chicken stock. It’s very easy. Put the pumpkin in, fried it then add the chicken stock. 2. Intra-sentential switching Mandarin-English bilingual: Lyu et al., 2015 我今天想去 Starbucks 買一杯 ice-coffee. I am thinking of going to Starbucks to buy a cup of ice-coffee. 5.

(19) However, there is a difference of opinions as to whether the distinction between language switching and mixing is truly necessary. Hatch (1976) argued that these two phenomena do not have sharp distinction and should treat them as the same. Some scholars also suggest to focus the language switching and mixing performance of speakers on the discourse and interactional functions and thus refer them as situational shifting, a tendency to switch by a change in situation (Gumperz, 1982; Pakir, 1989; Tay, 1989). On the other hand, others felt a need to distinguish them especially on the details of the grammatical structure (Kachru, 1978; McLaughlin, 1984; Bokamba, 1989). Therefore, this study hopes to unravel this confusion through exploring the underlying brain mechanisms of bilinguals in comprehending the language switching and mixing sentences. Based on linguistic corpora, linguistic theories are generated into predictive hypotheses to unravel the underlying linguistic and cognitive mechanisms in language switching and mixing. The current study will discuss the triggering hypothesis proposed by Clyne (1967, 2003). Clyne found out that the immigrants in Australia (German, Dutch, Hungrian, Italian, Spanish, Croatian, and Vietnamese) tend to code switch when a sentence contains one or more cognates. More specifically, cognate, including proper nouns and homographs, will trigger code switching to another language. Cognates are words that are related in origin to another word, and usually have similar meanings and spellings in different languages (e.g. tennis share the similar meaning and spelling in English and Croatian). In addition, Clyne (1967) differentiated 6.

(20) three forms of triggering loci: the sequential facilitation (where the code switched takes place after the trigger word), the anticipational facilitation (where the code switch precedes the trigger word) and the combination of these two (where the code switch is in the middle of these two trigger words). Therefore, the triggering mechanism suggested that the existence of cognates will first co-activate both languages in a bilingual and thus facilitate the code switch to another language. However, this hypothesis raised much debate as it is not put into statistical test and might be tagged as pure coincidence upon the arrival of these predictions from his corpus data. Broersma and De Bot (2006) tested the triggering hypothesis in their corpus study by analyzing several cognates and code switch conversations in Dutch-Moroccan Arabic bilinguals. Their statistical test of the code switching patterns found support for part of the original triggering hypothesis, which states that the cognate acts as a trigger to facilitate the code switch. However, the data were best explained in the adjusted triggering hypothesis, which suggested that the triggering takes place in the same clause and occurs before the surface structure is formed, in the lemma level. A trigger word in language X can cause a shift of activation in another language (the unintended language), for instance, the language Y, thereby causing the lemmas in language Y to be activated and thus increased the chances to be selected. Subsequent research (Broersma, 2009; Broersma et al., 2009) on the corpus analysis of Dutch. 7.

(21) immigrants in New Zealand and Australia reported that higher frequency of code switching patterns was observed in clauses containing a cognate. To reveal the underlying processing mechanisms in triggering code switching, Kootstra et al. (2012) investigated the triggering hypothesis under natural dialogue situation by means of confederate-scripted priming technique. This technique involved two participants that were instructed to describe pictures that were presented on a laptop. One of the participants was a confederate, who followed the instructions assigned by the experimenter and whose linguistic behavior was scripted. The other was the “real” participant, unknowingly aware of the identity of the other participant that the confederate’s dialogue to code switch or not was scripted. The main purpose was to examine whether the manipulation of switched or non-switched utterances of the confederate would influence the “real” participants to code switch. The overall result concluded that the immersion in a code switching discourse situation would consequently influence the lexical triggering of code switches. In addition, cognate words would co-activate bilingual’s syntactic representations in both languages and facilitate the syntactic integration of multiple languages into one sentence. Finally, several experimental studies with nouns as cognate triggers discovered that cognate lexical (noun) functioned as triggers in promoting the code switching process (Broersma, 2011; Kootstra et al., 2012). It is therefore questioned whether verb cognate would pose similar trigger as the noun cognate. Studies have reported that cognate verbs do not facilitate the 8.

(22) triggering effect in switching languages and that, cognate verbs rarely have similar orthography and phonology forms Bultena et al., 2012. In sum, the above studies suggest that the involvement of cognate words would facilitate the triggering of language switching. In addition, if provided with a discourse situation where high frequency of code mixing occurred, it is plausible to have the lexical triggering effect to code switch into another language. Finally, the facilitation process of code switching to another language is restricted to noun cognates.. 2.2 Language Switching and Mixing: a Cognitive Neuroscience Perspective Although the terms “code-mixing” and “code-switching” are sometimes used interchangeably, different disciplines or fields might apply different or similar definitions for the same terminology. Therefore, it is noteworthy to mention that in psycholinguistics and neurolinguistics research, including the literature review of the current study, either the “language switching/mixing” or “code switching/mixing” can collectively represent the language switching and mixing phenomena. Language switching and mixing might be different in essence. Patients with pathological switching tend to alternate their languages across different utterances, whereas in pathological mixing, patients intermingle different languages within a single utterance even though h/she is 9.

(23) fully aware that the interlocutor does not speak that language (Fabbro & Daro, 1995a; Fabbro, 1999, 2001b, 2001a). Fabbro (2001b) defined that language switching was in compliance to pragmatic constraint which concerns the ability to master the rules for social language use with the semantic aspect of language (the meaning) conveyed in an appropriate social context. In contrast, language mixing was constrained by the linguistic factors which deal with the structural aspect. As reported by Fabbro, pathological switching between languages mainly involved the lesions of the frontal lobe (left and right) and left anterior cingulate cortex (ACC) (Fabbro, 2001a), while persistent mixing of elements from languages were due to left parietotemporal and left postrolandic lesions (Fabbro, 1999, 2001a), as shown in Figure 1.. Figure 1: Brain areas that are responsible for pathological switching and mixing.. 10.

(24) It is important to note that, for a person to freely switch between or mix languages, cognitive control also plays a role. Kroll et al. (2014) discovered that bilinguals experience parallel activation in both of the languages while conversing. For that to happen, an effective control system must have been developed to suppress or inhibit the non-target language by bilinguals (Prior & Gollan, 2011). According to Green (1986), the bilingual language system is composed of two distinct devices: the linguistic and control units. Specifically, linguistic units deal with lexico-semantic system at the semantic, syntactical, phonological, and morphological levels, whereas control units involve the control of the output in the language selection, language translation and interpretation. It is suggested that high cognitive control is required to identify the relevant language, and to inhibit the non-target language in order to appropriately switch between languages. Therefore, if the speaker alternates languages without control and fails to convey the intended meaning to the interlocutors, it is a pathological condition that requires special attention by the medical authorities. Although many studies tend to treat language switching and language mixing as a similar phenomenon, the discoveries of lesions in specific brain regions have brought upon the attention towards the differences between pathological language switching and mixing conditions. Therefore, to further look into these phenomena, language switching and mixing are discussed in more detail in the following session.. 11.

(25) 2.2.1 Language Switching The frontal lobes (FLs) are cited as responsible for mediating activation and executive control of other cognitive systems, which include language system as well (Luria, 1973; Stuss & Benson, 1986). Indeed, clinical studies have shown that individuals with the FL damage, while not aphasic, would experience disoriented and disturbances in their language functions. The impairment in FL would make individuals unable to appreciate and respond to sentences with inferential or implicit meanings under certain social context as they are only able to focus on the concrete aspects of the information given (McDonald & Pearce, 1996). This finding is in accordance with Fabbro’s proposal (2000) that pathological switching might deal with the pragmatic aspects of the language as patients are unable to code switch effectively in an appropriate social context. Within the FL, the dorsolateral prefrontal cortex (DLPFC) had been correlated with nonverbal task switching (Meyer et al., 1997). Hernandez and colleagues’ studies (2000, 2001) reported the DLPFC as a key region for controlling language switching and inhibiting the currently non-targeted language. Their studies were conducted by using single- and duallanguage picture naming tasks. Spanish-English proficient bilinguals were instructed to name the picture covertly in accordance to the pre-stimulus cue indicating the language to use. The result demonstrated an increased activation of the DLPFC for the switched language condition relative to the non-switching condition, which reflects the executive processing induced by 12.

(26) language switching. Holtzheimer et al. (2005) provided further support for the involvement of the left DLPFC in language switching. Two bilingual patients suffered from major depression undergo repetitive transcranial magnetic stimulation (rTMS) treatment which applied to the left DLPFC. After the rTMS treatment, although being able to converse with others fluently in their L1, the patients felt more natural and experienced stronger urge to speak in their less frequently used language (L2) even knowing the interlocutor could not speak that language. The first patient had this strong impulse for at least two hours after the rTMS session, while the second patient only experienced this feeling for two minutes. Therefore, it was proposed that the role of language switching in the brain may have been disrupted when left DLPFC was stimulated repeatedly. Thus, this finding provides further support to previous studies (Fabbro et al., 2000; Fabbro, 2001b), suggesting that DLPFC may be involved in language switching. However, it is plausible that the rTMS disrupted the suppression of the non-target language rather than the stimulation of the language switching phenomenon (Rodriguez-Fornells et al., 2002). To tackle this question, more direct evidence was required to examine the competition between languages and to determine how bilinguals inhibit the non-target language interference. RodriguezFornells et al. (2002) investigated the underlying neural mechanism of language selection in early bilinguals (Catalan-Spanish) and monolinguals with fMRI. Participants had to inhibit the non-target language (Catalan), while responding only to the words in the target language 13.

(27) (Spanish). It was found that only bilinguals, but not monolinguals, exhibited activation in the left anterior prefrontal region (Brodmann areas 45 and 9) when executing the language inhibition process. This study provided support for the fact that even highly proficient bilinguals require the inhibition mechanisms to allow for the language selection process. Therefore, it might be possible that the rTMS disrupt the suppression of non-target interference rather than engaging in the language switching process. The PET (positron emission tomography) technique was used to investigate the underlying neural mechanisms in language switching. Price (1998) conducted a PET test on German-English bilinguals with high proficiency, participants were instructed to read alternated L1 and L2 words that were visually presented. Increased activations were found in the left inferior frontal gyrus and the supramarginal gyri (SMG) in the switched trials than nonswitched trials. The supramarginal gyri were suggested by Price (1998) to function as the mapping of orthography into phonology. Thus, while alternating between L1 and L2, higher demands would be placed on the phonological recoding operations. The result of this study is in contrast with Hernandez’s discovery as the DLPFC was not detected in this task. Price (1998) reasoned that the non-activation in the DLPFC might be due to the highly predictability pattern and cue given by the L1 and L2 input in this task. Therefore, it is important to know that the types of language task being assigned would affect the involvement of the systems in modulating language switching Price et al., 1999. 14.

(28) Chee et al. (2003) investigated the effect of word repetition within and across languages in high proficient Mandarin-English bilinguals. The within-languages condition comprises a single language (English-only), whereas the across-languages condition contains word pairs of mixed languages (e.g. English-Mandarin word pairs). Participants were requested to read each word silently and figure out its meaning. The result indicated a greater activation in the left prefrontal, lateral and inferior temporal regions under mixed-language condition. It was further suggested that greater cognitive resources were required while processing the mixed-language condition as more extended left prefrontal activity were observed. In addition to the cortical areas, subcortical regions also play a role in language switching, including the ACC and the basal ganglia. The involvement of the ACC was reported in Fabbro et al. (2000). The authors demonstrated a case study on a 56-year-old bilingual patient, who was fluent in both Friulian (L1) and Slovenian (L2). The patient had a lesion in the prefrontal lobe (both left and right) and left ACC and experienced pathological compulsive switching between languages. Various studies have also shown that the ACC plays an essential role in error detection and conflict monitoring (Ide & Chiang-shan, 2011), where it receives incoming information, provides responses, monitors the outgoing responses, and implements further actions if there is a violation of expectancy (Luu & Pederson, 2004). When the ACC receives signals, it would further allocate the signals to internal processing or external stimulation to take a prompt action (Seeley et al., 2007). 15.

(29) Crinion et al. (2006) conducted PET and fMRI studies to compare the highly proficient German-English bilinguals and Japanese-English bilinguals. The inclusion of entirely different linguistic families (i.e. German and Japanese) hopes to provide support for the universality of the language mechanism. Participants were instructed to perform relatedness semantic decision task based on the targeted words (either in the same or different language). The result revealed an increased activation in the left caudate for the unrelated semantic words pairs or word pairs that were presented in different languages. It was proposed that the left caudate was responsible for the identification of different languages and a universal role in controlling and monitoring the language in use. Many neuroimaging studies have shown that the dorsolateral prefrontal cortex, inferior parietal cortex, caudate nucleus (posited in the subcortical structure of the basal ganglia), anterior cingulate cortex (ACC) and bilateral supramarginal gyri were involved in the cognitive control and information processing (Graybiel, 1997; Botvinick et al., 1999; Duncan & Owen, 2000; Middleton & Strick, 2000; Botvinick et al., 2001; Braver et al., 2001; Bunge et al., 2002; Kerns et al., 2004; McCormick et al., 2006), implicating the involvement of subcortical-cortical network in the voluntary and involuntary language switching. Based on the previous neuroimaging studies on language switching in bilinguals, Abutalebi and Green (2008) proposed a neurological pathway which may explain the complex network at work during language switching. The prefrontal cortex (concerned with executive functions, decision16.

(30) making, response selection, response inhibition and working memory) was linked with the ACC (involved in attention, conflict monitoring and error detection) and basal ganglia (dealing with language selection, set switching, language planning and lexical selection) for the function of response inhibition, specifically, to suppress the interference from the non-target language. It was further proposed that the supramarginal gyrus (SMG) in the inferior parietal cortex (concerned with the maintenance of representation and working memory) dealt with cases of unpredictable language switches. The left SMG biased selection away from the language which is not in use, whereas the right SMG biased selection towards the language in use. Abutalebi and Green (2008) suggested that either a left basal ganglia-left prefrontal cortex network was activated to subserve language planning and/ or the inhibition of a prepotent response by basal ganglia towards the supplementary motor area (SMA) (Sumner et al., 2007). Finally, research has also shown that the right hemisphere might be involved in pathological switching. Lebrun (1990) investigated the spontaneous switching phenomenon, in which the patients experienced constant uncontrolled switching of languages within several sentences such as using L1 in the first sentence, switching to L2 in the second sentence, and switching to L3 in the third sentence, and so on. A multilingual patient (Italian-French-German) who was diagnosed with Alzheimer’s disease at the age of 57 after exhibiting memory disturbances was found to attain language switching disorder at the same time: he often switched between languages without reasons De Vreese et al., 1988. Lebrun (1990) suggested 17.

(31) that the lesion might localize in the right hemisphere which are also responsible for the regulation of the verbal output in languages. In addition, Martin et al. (1994) reported a case study of a polyglot patient (Indian-English), who had epilepsy in the right temporal lobe and experienced a pathological switching disorder. After anti-epileptic therapy was conducted, the patient attained full recovery from the language switching disorder (Martin et al., 1994). These two cases suggested that the right hemisphere was involved in the language selection and switching between languages.. 2.2.2 Language Mixing Language mixing was defined as the use of two languages within one sentence in conversations (Bokamba, 1989). Perecman (1984, 1989) provided one of the earliest detailed analysis and classification of various types of pathological mixing phenomena in aphasic multilingual patients which are listed in Table 1 below. Table 1: Types of Pathological Mixing Phenomena Types. Examples. Explanations. Word mixing. “…to think for my boys from England you know und hab ich immer so gemacht und one day I said to the boy” (…to think for my boys from England you know and I have always done this way and one day I said to the boy). When they cannot find a word owing to anomia, bilingual aphasics are likely to substitute it with a corresponding word in another language. In this case, patients are usually aware of the mixing phenomenon. Sometimes they unconsciously mix words from different languages within a sentence.. 18.

(32) Root and suffix mixing. Per andre all’ospedale ho preso la carra [to go to the hospital I took the carra] (from the English “car” plus the Italian suffix “a”). Patient would utter words in an English root and added the Italian suffix. Blending of syllables. The translation of the English butterfly into French la vontre fly. Syllables were blended from different languages within the same word. Use of the syntax of one language. Instead of reading the English sentence: “I got home from. English vocabulary with German syntax. and concurrent use of the lexicon of another language. work”, a patient said: “I will home coming,” which is syntactically wrong, because it uses the German syntax where the verb takes the final position.. Pronouncing the phonemes in another languages. A patient pronounced the English word door like the French word dur.; heard (English) was produces as hund (German). Utterance of a word in one language but pronouncing the phonemes in another language.. Responding in a language different from the language of address. Examiner: “What was your job in Canada?” Patient: “I was working with ce faccio coi…”(English-Italian alternation). The tendency to answer in a language different from that spoken by the interlocutor is to be considered a mixing phenomenon.. The above types of pathological mixing were analyzed from different linguistic levels (phonological, morphological, lexical-semantic and syntactic). Perecman (1989) suggested that the poorly delineated language boundaries in polyglot aphasic’s mental grammar and the phenomena in language mixing and spontaneous translation were pathological conditions commonly found in bilingual aphasics. However, Grosjean (1985) denied this proposal and suggested that the language mixing is not an abnormal behavior in bilingual aphasics; rather, he viewed the mixing of languages as a communicative strategy to construct an effective 19.

(33) conversation with others. Furthermore, studies have shown that morphological and lexical semantic level mixing are a common occurrence in normal bilinguals (Matras, 2000; Bhat & Chengappa, 2003; Bhat & Shyamala, 2005). Fabbro (1999) proposed three rules that normal multilingual individuals would have to adhere to in order to produce acceptable and appropriate mixed sentences. 1. In a sentence the subject (if it is a pronoun) and the predicate (verb) generally belong to the same language. 2. Hardly ever are prepositions alone expressed in a language other than the one used for the whole sentence. 3. In mixing, there is a clear tendency to express function words or proverbs in the mother tongue. Normal bilinguals or multilinguals respect the rules of mixing in order to have a constructive conversation with others; however, if an individual violates the above rules of language mixing and is unable to make an effective conversation with others, the situation is likely to be treated as a pathological condition. Numerous case studies on brain-damaged patients have reported that pathological mixing is mainly due to the lesion in the left postrolandic area (Fabbro, 1999 & 2001a), which was also commonly found to be associated with Wernicke’s aphasia, (Fabbro, 1999; Fabbro et al., 2000). It is also worth reviewing Muysken (2000) typology, who classified code-mixing into 3 20.

(34) distinct processes: a. Insertion. The incorporation of lexical items or entire constituents into a structure of another language; b. Alternation. The process of alternation occurred between different structures of languages within the same conversation. c. Congruent Lexicalization. The convergence of both languages’ grammatical and lexical structures to form a new utterances. The definition of insertional process is best fit to describe the language mixing sentences of the current study where the lexical items (nouns) are inserted into the base languages. We will explain more about the experimental materials in Chapter 3. In addition, the Matrix Language Frame model (MLF) by Myers-Scotton (1993) can be used to further explain the insertional code-mixing with the incorporation of the terms Matrix language and Embedded language, which means the assimilation of the lexical items from the embedded language into a more dominant language, also being referred to as the matrix language. However, here arises the debate of lexical borrowing (especially nouns) akin to insertional code-mixing. Sapir (1921) claims that borrowing is the transference between languages where one brings influence to the other language. In order to distinguish between lexical borrowing and code-mixing, we take consideration of Thomason (2003)’s quote: “A code-switched word (also known as code-mixing) or other morpheme becomes a borrowing if it is used more and 21.

(35) more frequently – with or without phonological adaptation – until it is a regular part of the recipient language, learned as such by new learners”(p.696). In other words, when the frequency use of a code-mixing word from a non-dominant language is elevated, there’s likely it will be borrowed by the matrix language. It should be noted that our experiment materials avoided using the borrowing words. Therefore, taken together with the above rules that were stated by Fabbro (1999) on language mixing sentences, the current study filtered out the highest rated nouns when we collected participants’ responses via questionnaire as we want to exclude borrowing words in our materials. After reviewing the linguistic characteristics of language mixing, we now will discuss the neural substrates for language mixing. Case studies on brain-damaged patients have shown that pathological mixing mainly centered in the parieto-temporal structures of the left hemisphere. Neurologist Leischner (1943, 1983) reported that mixing disorder was found in a deaf-mute polyglot aphasic. He acquired sign language as his first language, Czech as his second language and English as his third language. After experiencing strokes, he involuntarily mixed Czech and German syllables within a word, unable to express a word in different mediums such as writing, speaking and sign languages. Later autopsy revealed that the lesion mostly centered in the superior temporal gyrus and the inferior parietal lobe of the left hemisphere. In addition, neurologist Perecman (1984) found lesion in the left temporal lobe of a Slovene-Italian-Friulian-English multilingual patient who severely mixed all of her languages 22.

(36) and also suffered from Wernicke aphasia. For example, patients would respond to a question by mixing English and Italian languages: “I was working with ce faccio coi…del… fare, I signori la che I faceva…” (I was working with English I do with… do… men there who did…). Abutalebi et al. (2000) examined a case of pathological language mixing after the Armenian-English-Italian polyglot patient experienced lesion centering on the head of the left caudate nucleus. The patient developed a non-fluent output in oral productions tasks, which suggested that the damaged in the subcortical structure may cause the impairment in the mechanisms involving language selection. In addition to neuropsychological findings, numerous studies have been initiated to investigate code mixed sentence processing (Moreno et al., 2002; Liao & Chan, 2016). Most behavioral studies have shown that significantly shorter time will be required by bilinguals in response to non-mixed sentences as compared to mixed-sentences. It was thus suggested that the processing cost in perceiving the stimuli were higher in mixed-sentences. Moreno et al. (2002) conducted an ERP study to investigate the processing cost of English-Spanish bilinguals in language switching. Participants were requested to comprehend English sentences and idioms ending either with lexical switch (English synonyms) or code switch (Spanish translations). The result showed that the code switch condition elicited a left negative-going potential that occurred between 250 to 450 ms, the left anterior negativity (LAN). The LAN effect has often been associated with syntactic structural processing complexity, especially 23.

(37) related to the high demand of working memory (Kluender & Kutas, 1993; King & Kutas, 1995). In addition, at the parietal and occipital region, a large positive complex (LPC, also known as P600) was also observed in the code switch condition. LPC/P600 has been suggested to reflect the reanalysis of complicated sentences, such as the semantically unrelated sentences and garden-path sentences (Osterhout et al., 1994; Kim & Osterhout, 2005). Therefore, the processing cost for language mixing condition was higher than that for the non-mixed condition, as participants required more efforts for stimulus evaluation. In sum, the above studies demonstrated that language switching and language mixing are two distinct phenomena which correspond to lesions in different brain regions. Specifically, the bilateral frontal lobes and the left anterior cingulate cortex (ACC), left basal ganglia were found in pathological switching, whereas in pathological mixing conditions, the responsible brain regions mainly centered in the left superior temporal gyrus, left inferior parietal lobe, left temporal lobe, left postrolandic, parieto-temporal and basal ganglia. In addition, research has also shown that language mixing sentences, as compared to non-mixing ones, demanded greater effort and longer process time to bilingual speakers.. 2.2.3 Coexistence of Pathological Switching and Mixing It is sometimes difficult to have a clear cut distinction between pathological switching and pathological mixing. Therefore, it is plausible to have the coexistence of both pathologies in patients. 24.

(38) The coexistence of pathological switching and mixing might reflect impairment in the executive control and true language deficits, where patients have trouble in constructing phrases/sentences with appropriate syntactic structure or the inability to search for relevant words/vocabulary of a language (Ijalba et al., 2008). Patients who experienced brain damage may alter their language control and general cognitive capabilities in order to manage dual-language use (Hernandez et al., 2000). In this regard, Fabbro (2000) stated that both pathological switching and mixing might be correlated to the capacity to withdraw attention and resources from L2, where the patients experience inability to focus and extract L2 information from the L2 resources. In addition, pathological switching and mixing might reflect the weakening of the inhibition control of unwanted L2 in cognitive capacity as they are unable to avoid the L2 interference when using the native language. Moreover, the ability to use one language instead of another by bilinguals and multilinguals are mainly regulated by one’s proficiency level (Green, 1998; Price et al., 1999). Therefore, when the proficiency of L2 in bilinguals and multilinguals increases, the language users have the ability to transfer the common structures from L2 and incorporate with L1 structures, which is somehow lacked in pathological patients to correctly transfer and inhibit pathological switching and mixing of undesirable target language. Mariën et al. (2005) reported a case study on a 10 year-old English-Dutch bilingual (EM) 25.

(39) who had two strokes, which caused him to spontaneously switch from L1 to L2 and mix L1 and L2 linguistic units in a conversation, even though he was instructed to use only one language. The first stroke caused him to have disrupted L1 and L2 with poor comprehension and speech construction. He eventually experienced remission of those symptoms after three weeks, and gradually acquired recovery in his L1. However, the second stroke after seven weeks caused his L1 to deteriorate back to the initial state, the oral and comprehension reading for L2 was worse than the first stroke and he even experienced more frequent pathological switching and mixing conditions. After months of treatment, EM’s condition recovered gradually with remission of global aphasia in L1 and L2, and regained his ability to switch and mix languages appropriately. At the early stages of the strokes, computerized tomography (CT), MRI, and single-photon emission computed tomography (SPECT) revealed that the lesions were found in the left frontal cortex, left temporo-parietal areas, left caudate nucleus, and left thalamus. Improved blood flow (perfusion) were found in the left frontal lobe and left caudate nucleus on the later stage when EM recovered from his aphasia and pathological switching and mixing conditions. The parallel remission of frontal lobe and caudate nucleus with the pathological symptoms of language switching and mixing suggested that the anterior loop of the subcortical-cortical circuit (e.g., prefrontal cortex, caudate nucleus, globus pallidus, and the ventral anterior thalamic nucleus) as proposed by Alexander and Crutcher (1990) was proven to be the neural counterpart for the language processing of language switching and mixing. 26.

(40) Thus, the circuit which includes the left frontal lobe and basal ganglia might cause the pathological switching and mixing in EM. In fact, the anterior loop of the subcortical-cortical pathway was found to be involved in a more general process, not just specific to language domain, but associated in multiple domains such as executive, decision-making and attentional cognitive control (Alexander & Crutcher, 1990; Aglioti et al., 1997). It was further supported by studies on multilingual language processing, stating that cortical-subcortical loop is not unique to multilinguals, or specific to language, but crucial for language switching (Paradis & Goldblum, 1989; Zatorre, 1989; Aglioti et al., 1997; Crosson, 1999). To conclude Sections 2.2.1-2.2.3, we provide Tables 2, 3 and 4 to summarize neuropsychological and neuroimaging studies about language switching and mixing, respectively.. 27.

(41) Table 2: Neuroimaging Studies Investigating Pathological Switching and Pathological Mixing in Bilinguals and Multilinguals Authors. Pathological Switching. Herschmann and Pötzl (1920); Pötzl (1925, 1930); A Leischner (1948). Parietal lobe. Leischner (1943,1983a). Pathological Mixing. Left superior temporal gyrus and the left inferior parietal lobe. Stengel and Zelmanowicz (1934); Zatorre (1989). Frontal lobe. Perecman (1984). left temporal lobe. Aglioti et al. (1993, 1996). Left basal ganglia. Lebrun (1991). Right Hemisphere. Martin et al. (1994). Right temporal lobe (epilepsy). Fabbro and Paradis (1995b) Anterior structure of frontal lobe. -. Fabbro et al. (1997). Left thalamic. Fabbro (2001a). Frontal lobe (left and right) and Left ACC. Abutalebi et al. (2000). Left postrolandic and parieto-temporal (left hemisphere) Head of the left caudate nucleus (subcortical). Abutalebi & Green (1986; 2008). Subcortical (basal ganglia) and frontal lobe. Mariën et. al. (2005). Left frontal lobe and basal ganglia (caudate nucleus). Abutalebi and Green (2007) Ansaldo et al. (2008). Basal ganglia, ACC, prefrontal cortex Left basal ganglia. Adrover-Roig et al. (2011). Left basal ganglia. 28.

(42) Table 3: Neuroimaging Studies Investigating Language Switching in Bilinguals and Multilinguals Authors. Method. Subjects. Main Findings. Price et al. (1999). PET study in word naming task. Six late bilinguals (GermanEnglish). Higher activation in supramarginal gyri, and left posterior inferior frontal gyrus (Broca’s area) in switched condition. Hernandez et al.. fMRI investigation of picture. Early bilinguals (Spanish–. Increased activation of the dorsolateral prefrontal cortex. (2000, 2001). naming and language switching. English), more dominant in their L2 (English). (DLPFC) for the switched language condition. Jackson et al. (2001). ERP experiment of language switching in a digit naming task. Twenty-six native English (L1) speaker and capable of naming the digits 1 to 8 fluently in a second language (L2). Higher inhibition is required to suppress the more dominant language (L1). Language switching induced the parietal and frontal activity.. Rodriguez-Fornells et al. (2002). fMRI study of language selection between visually presented words. Seven high-proficiency early bilinguals compared to a group of seven monolinguals. Selective activation of the left anterior prefrontal region (Brodmann areas 45 and 9) presented only in bilinguals. Chee et al. (2003). fMRI study of word repetition within and across languages. Twelve early and high proﬁcient English–Mandarin bilinguals. Activation in left prefrontal, lateral and inferior temporal regions in mixed-language conditions (EnglishMandarin) compared to the single language (Englishonly) conditions. Greater cognitive resources in mixedlanguage conditions.. Holtzheimer et al. (2005). Two bilingual patients who suffered from major depression. Repetitive transcranial magnetic stimulation (rTMS) to the left DLPFC as 29. Urge to speak in their less frequently used language (L2) when left DLPFC is stimulated..

(43) treatment Crinion et al. (2006). PET and fMRI study of semantic decisions on target words preceded by prime words related or unrelated in meaning and either in the. PET 11 late German–English increased activation in left head of caudate for targetbilinguals primes pairs unrelated in meaning or related in meaning fMRI 14 late German– but differing in language English bilinguals fMRI 9 late Japanese–English. same language or in a different language. bilinguals fMRI 9 late Japanese–English bilinguals. 30.

(44) Table 4: Neuroimaging Studies Investigating Language Mixing in Bilinguals and Multilinguals Authors. Method. Subjects. Main Findings. Moreno et al. (2002). ERP experiment in comprehending unexpected language mixing sentences and idioms ending either. Thirty-four high-proficient English-Spanish bilinguals. Unexpected switching conditions induced the N400, LAN effects (relate to the complexity of syntactic structural processing) and LPC/ P600 (reflect the reanalysis of complicated sentences. The switched. with lexical switch (English synonyms) or code-switches (Spanish translations). condition incur greater processing cost as compared to non-switched condition.. Wang, Xue, Chen, Xue & Dong (2007). ER-fMRI in picture-naming tasks. Twelve native Chinese (L1) speaker and low-to-middle level of English(L2) bilinguals. The mixing conditions elicited greater activation in the right superior prefrontal cortex, left middle, superior frontal cortex, and also the right middle cingulum and caudate.. Liao & Chan (2016). ERP experiment in processing the unexpected switch of language at the end of the sentence. Twenty-one native MandarinTaiwanese bilinguals with more frequent use in Mandarin. An unexpected language switch generated a widespread N400; shorter time will be required by participants in response to non-mixed sentences as compared to mixedsentences. The language switch condition recruited higher cognitive control and longer processing cost while comprehending unexpected switch of language.. 31.

(45) In sum, through the investigation of pathological patients, electrophysiological and neuroimaging studies, most reviews have suggested that the underlying mechanisms of language switching and mixing varied (Herschmann & Pötzl, 1920; Pötzl, 1925, 1930; A Leischner, 1948; Green, 1986; Lebrun, 1991; Aglioti & Fabbro, 1993; Martin et al., 1994; Fabbro & Paradis, 1995b; Aglioti et al., 1996; Fabbro et al., 1997; Fabbro, 2001a; Abutalebi & Green, 2008; Ansaldo et al., 2008), while some reviews are unable to distinguish between these two phenomenon due to similar brain activations (Mariën et al., 2005; Abutalebi & Green, 2007; Adrover-Roig et al., 2011). Therefore, this brought us back to our first research question as to whether we can differentiate language switching and mixing at the biological level. Furthermore, many studies on healthy individuals did not examine language switching and mixing with a complete sentence or meaningful utterances, but solely with a series of unrelated words, digits or pictures (Price et al., 1999; Hernandez et al., 2000; Hernandez et al., 2001; Jackson et al., 2001; Wang et al., 2007). Our study added to the literature and used switching and mixing sentences as materials.. 32.

(46) 2.3 Switching Direction and Cognitive Control The above reviews on language switching and mixing indicated a measurable processing cost occurred when individuals switch or mix languages, but did not specify in which directions. Thus, this section discusses processing cost in terms of the direction of language switching and mixing. Many studies considered mostly unidirectional switching (mostly switches from L1 into L2), and not the opposite (from L2 to L1). However, numerous studies have shown that the cost of processing was asymmetric in these two switching directions and that such asymmetry might be able to be accounted for with the idea of cognitive control (Meuter & Allport, 1999; Alvarez et al., 2003; Liao & Chan, 2016), as demonstrated in the two models that will be described below: the Inhibitory Control model (IC) and the Bilingual Interactive Activation Plus model (BIA+). From a psycholinguistic perspective, Green (1998) proposed a production model, the Inhibitory Control Model (IC), to explain the modulation of the selective activation of the target language and inhibition of the non-target language (see Figure 2 for the diagram). In this model, the communication goal motivates the conceptualizer to construct the conceptual information during a linguistic task. The lexico-semantic system consists of lemmas (entries in the lexicon that contain information on the morphology, syntax, and phonology for each lexical item (Levelt, 1989) that were tagged for specific language. Thus, it was expected to find more highly 33.

(47) activated lemmas for a more dominant language. The language task schemas are networks that individuals may construct or alter in order to achieve a specific task. In addition, the SAS (Supervisory Attentional System) controlled the language task schema through inhibiting or activating the lemmas of unintended or intended language in the bilingual lexico-semantic system. According to Green, while executing a task in L2, stronger inhibition for the L1 was required as L1 lexical nodes were usually more activated. As a consequence, a longer reaction time and more efforts were required to switch into a more dominant and more suppressed language.. Figure 2: The Inhibitory Control Model (IC), proposed by Green (1998) Bilingual Interactive Activation Plus model (BIA+) (Dijkstra & Van Heuven, 2002), on the other hand, was built to explain the process of bilingual language comprehension and contained two interactive subsystems, respectively a word identification system and a 34.

(48) task/decision system (see Figure 3 for the diagram). The processing of the information flow will be directed from a bottom-up manner starting from the word identification system (the linguistic context). The input will first activate the orthographic, phonological and semantics representations and consequently builds an interactive network with the language nodes, which reflect the membership to particular language. All the information collected at the word identification system will thus be transmitted to the task/decision system to carry out further executive actions, which include all the non-linguistic information. In addition, this model proposes that the resting-level activation of words in a language depends on their recency of use. In most cases of bilinguals, L1 attains higher frequency of usage as compared to L2, and so higher L1 representation is expected, and the cross-linguistic influence will be greater from L1 to L2. Therefore, L1 to L2 language switches will be more energy consuming than the other way round.. 35.

(49) Figure 3: The Bilingual Interactive Activation Plus model (BIA+), proposed by Dijkstra and Van Heuven (2002) The language control in the IC model was achieved through the implementation of language task schemas. It was hypothesized that the language selection was expected to occur at an early stage and cognitive control was required to inhibit non-target language (Ye & Zhou, 2009). Whereas, in the language comprehension model, BIA+, the processing of language selection occurred at a late stage and the function of its cognitive control is to reanalyze and resolve ambiguous sentences (Ye & Zhou, 2009). In terms of processing costs in language switching, IC stressed the effect of inhibition. It was proposed that the dominant language will attain greater lemma activation and thus stronger efforts were required to inhibit the switching to L1 while performing a task in L2. However, BIA+ provided an alternative proposal. It. 36.

(50) suggested that the frequently used language (L1) will have higher resting-level activation and greater cross-linguistic influence from L1 to L2. Since the current study required bilinguals to comprehend switching and mixing sentences, studies concerned with the comprehension model, BIA+ and the language switching/mixing direction will be reviewed below. Numerous studies have shown temporal delay of the infrequently used language in processing(in most cases is the L2), which supports the hypothesis that the frequency use of a language would determine the activation of the language representation and thus affect the reaction time asserted (Ardal et al., 1990; Moreno & Kutas, 2005). Proverbio et al.’s (2004) behavioral data provided support towards the above studies. It was found that switching from L1 to L2 required longer response time, but not the other way around. Thus, it is possible to infer that the representations in the more dominant language were more active and thus easier to retrieve from the bilinguals’ mind. Liao and Chan’s (2016) study has also shown that switching from a more dominant language (Mandarin) to less dominant language (Taiwanese) required extra time, which indicates more effort and strength were exerted to access a word in Taiwanese, or to incorporate a word in the Mandarin context. This result supported the BIA+ model (Dijkstra & Van Heuven, 2002), which hypothesized that switching from a more dominant language to a weaker language would be more difficult. The. 37.