• 沒有找到結果。

專家與生手口譯員的句子理解歷程: 事件相關腦電位研究

N/A
N/A
Protected

Academic year: 2021

Share "專家與生手口譯員的句子理解歷程: 事件相關腦電位研究"

Copied!
140
0
0

加載中.... (立即查看全文)

全文

(1)國立臺灣師範大學翻譯研究所 博士論文 Doctoral Dissertation Graduate Institute of Translation and Interpretation National Taiwan Normal University. 專家與生手口譯員的句子理解歷程: 事件相關腦電位研究 Sentence comprehension in expert and novice interpreters: An ERP study. 指導教授:詹曉蕙博士、劉敏華博士 研究生:范家銘 中華民國一○二年七月. Advisors Dr. Chan, Shiao-hui Dr. Liu, Minhua By Fan, Chiaming (Damien) July 2013.

(2) 摘要. 過去研究發現,專家口譯員相較生手口譯員更擅於利用各種語境線索預測講者將 表達的訊息。本研究旨在以事件相關腦電位技術,探討專技程度不同的口譯員在 聆聽具有脈絡限定性的句子時,預測句末詞語的能力是否有差異,並藉此神經相 關機制佐證過去行為研究之發現。實驗材料為具有脈絡限定性之全中文或第一句 為中文、第二句為英文之句組,句末為符合或違反語境脈絡之目標詞。三組專技 程度不同之口譯員(口譯研究所學生、資歷約 1 年之資淺口譯員及資歷約 11 年 之專家口譯員)聆聽句組後,須判斷該句組是否符合邏輯。實驗結果發現,在目 標詞出現後 300-600 毫秒間,專家口譯員在左腦之語意一致性效果(semantic congruity effect)顯著大於口譯學生,顯示專家口譯員較能運用脈絡訊息形成預 測。然而,究其原委,三組受試者在目標詞違反語境脈絡的情況下,N400 波形 並無差異;反而是專家口譯員在目標詞符合語境脈絡下,N400 較其他兩組受試 者為正。N400 振幅可反映大腦提取詞彙之難易程度。據此,專家口譯員振幅較 小之 N400 可能係因累積長期經驗,擁有較豐富之詞彙提取架構(retrieval structure),因而得以預測並輕易提取符合語境脈絡之目標詞。另外,在 600-900 毫秒時,專家口譯員的前腦產生較其他兩組受試者明顯之正向波。晚期正向波 (late positivity)可解釋為大腦更新心智表徵(mental representation)所費之力氣, 顯示專家口譯員在句組合理時(意即符合平常翻譯情況)花費較多力氣更新心智 表徵,而不會將力氣花費在處理不合理的句組。即使受試者僅須判斷句組是否合 理,專家口譯員似乎習於更進一步處理訊息以利後續翻譯。此外,資淺口譯員之 腦波模式介於專家與學生之間,顯示訓練與經驗或許能強化語意預測能力。. 關鍵詞:事件相關腦電位、口譯員、預測、N400、晚期正向波.

(3) Abstract. Past studies showed that expert interpreters are better than novice interpreters at using contextual cues to anticipate upcoming information. The present study aims to identify the neural correlates of anticipation by using event-related potentials (ERPs) to explore whether interpreters of different levels of expertise differ in their anticipation of sentence-final target words in contextually constraining sentence pairs. Sentence pairs that were entirely in Chinese or switched to English in the second sentence were aurally presented to expert interpreters, novice interpreters, and post-graduate interpreting students.. Results revealed that between 300 to 600-ms. post target onset, expert interpreters showed a significantly larger semantic congruity effect in the left hemisphere than interpreting students, indicating that the former used a more predictive process. However, the larger effect was due to an attenuated N400 for the congruent targets.. An attenuated N400 could reflect easier lexical access.. Due to training and work experiences, expert interpreters may have constructed rich retrieval structures to these lexical items; therefore they accessed the congruent items more easily.. In addition, in the 600 to 900-ms time window of congruent conditions,. a distinctive frontal positivity existed in the frontal regions of expert interpreters. Late positivities could reflect the updating of mental representations. Expert interpreters could be accustomed to deeper processing so as to prepare optimal mental representations for later translation. Therefore, even though the only task was to judge the logicality of the sentence pairs, expert interpreters still spent more efforts in updating the mental representations to congruent sentences instead of those of incongruent ones. This is because congruent sentences conform to their usual working condition.. In addition, the brainwave pattern of novice interpreters seemed. to be migrating from that of interpreting students to that of expert interpreters, ii.

(4) suggesting that training and experience might be potentially beneficial to improving anticipation capability.. Key words: ERP, interpreter, anticipation, N400, late positivity. iii.

(5) Acknowledgment. There is a reason why very few conference interpreters pursue a doctorate degree. Fewer fancy an academic career. I argue that the nature of scholarly research and a formal institutional setting are somewhat contradictory to the reasons we chose to become (freelance) conference interpreters in the first place. So I have surprised myself in embarking on this journey and actually obtaining the degree while continuing my teaching duties and being active in the market. Without my two advisors, Dr. Shiaohui Chan and Dr. Minhua Liu, I would have given up. Dr. Liu’s belief in rigorous empirical research has a profound influence on how I see my profession. Dr. Chan not only initiated me into the field of neurolinguistics, she also provided an unbelievable amount of guidance and support. Their encouragements are truly inspiring, and they gave me the courage to enter the field of cognitive neuroscience and do something that has probably never been done before. I would also like to express my gratitude to Dr. Chia-ying Lee, Dr. Chia-chien Chang, and Dr. Tze-wei Chen for offering valuable suggestions during my oral defense. My heartfelt gratitude goes to all the colleagues and students who took part in my experiment. Thank you for giving me your valuable time and lending your interesting brains. Without these 37 generous people, the experiment could not have been completed. I am also hugely indebted to the team in the neurolinguistics lab: Gracie, Helen, Ken, Matt, Vivi, Elvis, and Jeff. Thank you for your assistance, and thank you for your unwavering support. I would also like to thank my colleagues at school, my colleagues in the booths, and my students and their parents in the marching band that I have been teaching for more than a decade. Their understanding and blessings have been extremely reassuring. Finally, I would like to thank my family. I knew Dad and Mom were right when they said that I would regret if I did not pursue a PhD. I am glad that I did, and thank you for believing in me.. iv.

(6) Table of Contents List of Tables................................................................................................................vii List of Figures .............................................................................................................viii Chapter 1. Introduction ............................................................................................ 1. Chapter 2. Literature Review .................................................................................. 5. 2.1 Performance differences in expert and novice interpreters .................................. 5 2.2 Language anticipation and interpreting ............................................................... 8 2.3 Brain imaging in interpreting studies ................................................................. 12 2.3.1 High spatial resolution techniques ............................................................... 12 2.3.2 High temporal resolution technique: event-related potentials (ERP) ........... 14 2.4 Anticipation in ERP studies ............................................................................... 17 2.4.1 N400 and anticipation .................................................................................. 17 2.4.2 Late positivity and anticipation .................................................................... 21 2.5 Summary ............................................................................................................ 23 Chapter 3. Methods ............................................................................................... 25. 3.1 Subjects .............................................................................................................. 25 3.2 Materials ............................................................................................................ 26 3.2.1 Overview ...................................................................................................... 26 3.2.2 Stimulus construction ................................................................................... 28 3.2.3 Pre-test and cloze probability ....................................................................... 32 3.3 List generation ................................................................................................... 34 3.4 Recording ........................................................................................................... 35 3.5 Procedure ........................................................................................................... 38 3.6 Behavior and EEG recording ............................................................................. 39 3.7 Data Analysis ..................................................................................................... 40 Chapter 4. Results .................................................................................................. 44. v.

(7) 4.1 Behavioral data .................................................................................................. 44 4.1.1 Accuracy rate ................................................................................................ 44 4.1.2 Reaction time ................................................................................................ 45 4.2 ERP data ............................................................................................................. 47 4.2.1 N400 ............................................................................................................. 52 4.2.2 PNP ............................................................................................................... 63 Chapter 5. Discussion ............................................................................................ 69. 5.1 Behavioral Data ................................................................................................. 69 5.2 N400................................................................................................................... 71 5.2.1 Switch effect ................................................................................................. 71 5.2.2 Congruity effect versus absolute N400 amplitude ....................................... 72 5.2.3 Congruity effect ............................................................................................ 74 5.2.4 Absolute N400 amplitude ............................................................................. 75 5.3 PNP .................................................................................................................... 76 5.3.1 Pattern differences across groups ................................................................. 76 5.3.2 The hypothesis of updating mental representations ..................................... 77 5.4 Skilled memory in interpreting expertise ........................................................... 80 5.5 Mental representation in interpreting expertise ................................................. 83 5.6 Summary ............................................................................................................ 87 Chapter 6. Conclusion ........................................................................................... 89. References .................................................................................................................... 93 Appendix I. Experiment Materials ......................................................................... 103. Appendix II. Experiment Instruction .................................................................. 128. Appendix III. Subject Consent Form .................................................................... 129. vi.

(8) List of Tables Table 1 Upper and lower list segmentation.................................................................. 27 Table 2 Example of congruent stimuli construciton in upper set................................. 28 Table 3 Example of incongruent stimuli construction in upper set ............................. 30 Table 4 Example of congruent stimuli construction in lower set................................. 30 Table 5 Example of incongruent stimuli construction in lower set ............................. 31 Table 6 Arrangement of context and carrier sentences ................................................ 34 Table 7 Arrangment of experiment lists ....................................................................... 35 Table 8 Length of experiment lists............................................................................... 37 Table 9 Accuracy rate of each condition and total accuracy rate ................................. 44 Table 10 Accuracy rate of collapsed conditions .......................................................... 44 Table 11 Average reaction time in each experiment condition .................................... 45 Table 12 Average reaction time in collapsed conditions .............................................. 46 Table 13 Pairwise comparisons of the congruity and switch effects in N40 epoch ..... 53 Table 14 Lateratliy analysis of congruity effect ........................................................... 60 Table 15 Pairwise comparisons of anterior mean amplitudes in PNP epoch ............... 64 Table 16 Pairwise comparisons of posterior mean amplitudes in PNP epoch ............. 65. vii.

(9) List of Figures Figure 1. Procedure of stimuli presentation................................................................. 39 Figure 2. Congruent vs. incongruent grand average waveforms of all 30 subjects. .... 48 Figure 3. Non-switched vs. switched grand average waveforms of all 30 subjects. ... 48 Figure 4. Grand average waveforms of non-switched congruent vs. incongruent conditions (top) and switched congruent vs. incongruent conditions (bottom) in Students. ............................................................................................................... 49 Figure 5. Grand average waveforms of non-switched congruent vs. incongruent conditions (top) and switched congruent vs. incongruent conditions (bottom) in Graduates. ............................................................................................................ 50 Figure 6. Grand average waveforms of non-switched congruent vs. incongruent conditions (top) and switched congruent vs. incongruent conditions (bottom) in Professionals. ....................................................................................................... 51 Figure 7. Students’collapsed brainwaves of congruent vs incongruent conditions ..... 54 Figure 8. Graduates’ collapsed brainwaves of congruent vs incongruent conditions .. 54 Figure 9. Professionals’ collapsed brainwaves of congruent vs incongruent conditions .............................................................................................................................. 55 Figure 10. Representative EEG waveforms from three electrodes delineating the collapsed congruent vs collapsed incongruent conditions ................................... 55 Figure 11. Students’ collapsed brainwaves of non-switched vs switched conditions .. 56 Figure 12. Graduates’ collapsed brainwaves of non-switched vs switched conditions56 Figure 13. Professionals’ collapsed brainwaves of non-switched vs switched conditions ............................................................................................................. 57 Figure 14. Comparing congruity effect waveforms across the three groups of subjects. .............................................................................................................................. 58. viii.

(10) Figure 15. Representative EEG waveforms from three electrodes illustrating the congruity effect .................................................................................................... 59 Figure 16. Group comparison of the collapsed congruent condition waveforms. ....... 61 Figure 17. Group comparison of the collapsed incongruent condition waveforms. .... 62 Figure 18. Topographic maps of the congruent condition ........................................... 66 Figure 19. Topographic maps of the incongruent condition ........................................ 67. ix.

(11) Chapter 1. Introduction. One morning ten years ago, when I was still an interpreting student, my fellow classmates and I came out from the simultaneous interpreting (SI) booths after a grueling practice session.. The teacher, after pointing out all the faults we had made,. nonchalantly ended the class by saying, “If you can’t anticipate what the speaker is going to say, you won’t be able to do SI. Start anticipating!”. Ten years on, I often. find myself silently reciting this exclamation before turning on the microphone on the interpreter’s console in the booth.. I have also found myself relaying this mantra to. my own interpreting students for the past several years. However, how does an interpreter go about anticipating? processes involved?. What are the. Are there neural correlates that could serve as hard evidence to. show that interpreters are actually anticipating what the speaker is going to say? Without looking into questions like these, it is quite difficult to figure out how one could teach anticipation to students, let alone tell them what to anticipate. There is scant literature on anticipation in SI, and fewer instructions on how to teach anticipation or improve anticipation abilities. The biggest volume on anticipation in interpreting studies is probably Ghelly Chernov’s work entitled Inference and Anticipation in Simultaneous Interpreting: A Probability-Prediction Model, published in English in 2004.1. His main argument was that SI is only. possible because of an “objective redundancy” and a “subjective redundancy” in natural discourse.. Objective redundancy refers to “theme” (i.e., repetitive. information) and “rheme” (i.e., new information) that are juxtaposed and alternated to. 1. The original version was published in Russian in 1987 under the title Основы Синхронного Перевода. (Introduction to Simultaneous Interpretation). 1.

(12) form peaks and troughs in information density.. Subjective redundancy refers to the. interpreter’s familiarity with such parameters of the discourse situation as speaker, topical knowledge, and socio-pragmatic relationships.. Therefore, if interpreters. could analyze and discern between old and new segments of information, they could then focus their attention on dealing with the critical rheme while “compressing” (p. 113) the repetitive theme (Chernov, 2004). Most importantly, interpreters’ brains “[generate] hypotheses in anticipation of certain verbal and semantic developments of the discourse” (p. 93).. These hypotheses could be generated because there are. multiple levels of redundancy and a cyclical alternation between “theme” and “rheme” as discourse unfolds, therefore leading to predictability (Chernov, 2004). Chernov validated his theory by an experiment.. He asked professional. interpreters to interpret (simultaneously) two types of sentences: syntactically well-formed but semantically anomalous sentences such as “The round knife flew squarely inside the bottom of smoke,” and sentences with misleading anticipatory cues such as “Very often we are told that Rome was not built in a day. Of course we realize that Rome was not built in a dale.” The results validated his hypothesis: interpreters could barely interpret the first kind, and there were many pauses, errors, and omissions in their renditions.. In contrast, for the second type of sentences,. despite the smooth translations, interpreters followed their own anticipation in 75% of all analyzable cases, meaning that they made the wrong rendition of the last proposition because the sentences ended with unexpected words (Chernov, 2004). However, these behavioral results do not reveal the actual process of hypothesis generation in the interpreters’ brains. Chernov’s exemplary theory of “message development probability anticipation” provided a description of the psycholinguistic mechanism which, in his views, made SI possible.. Nonetheless, it would be. potentially meaningful to actually see neural evidence that interpreters’ brains are 2.

(13) indeed working to generate the said hypotheses, or simply put, that anticipation is taking place.. It would be meaningful because if such neural evidence exists and we. could compare and contrast the evidence of interpreting students with that of professional interpreters, we could potentially trace and analyze the underlying processes of anticipation.. This would give us a better understanding of whether. anticipation is teachable, and advance the endeavors of developing ways to teach anticipation. Seeking neural evidence inevitably implies that brain imaging techniques have to be employed. The use of event-related potentials (ERPs), an imaging technique that utilizes electroencephalography (EEG), is especially useful, because it has high temporal resolution, which would benefit studies focusing on language processing. Although there have been a few ERP studies that have recruited professional interpreters as subjects, none on the issue of anticipation per se has been carried out. Naturally, the methodological challenges and attitudinal issues mentioned by Gile (1995) and Pöchhacker (2004) are valid reasons why neurolinguistic experiments in interpreting studies are so rare. Cognitive neuroscientists and psycholinguists are not necessarily interested in the kinds of phenomena interpreter-cum-researchers are excited about, while interpreters rarely possess the scientific and methodological know-how to perform experimental research that meet the criteria of the brain imaging community. Therefore, the motivation of the present study is to lay down one additional brick to shorten the gap between interpreting studies and cognitive neuroscience.. In the. course of this undertaking, there are three objectives to meet. Firstly, the present study hopes to find the neural correlates of language anticipation in interpreters. Secondly, it has been suggested that expert interpreters are capable of predicting 3.

(14) upcoming information so as to allocate mental resources to other processes (Liu, 2008). The present study hopes to explore whether there is neural evidence to support this conjecture, and whether novice interpreters perform differently. Thirdly, the present study hopes to elucidate the sentence comprehension process of interpreters. Should neural evidence permit, it could provide a hint on how anticipation should be perceived or could be taught in the future. To attain these goals, the present study will adopt and modify experiment paradigms used in previous ERP anticipation studies and recruit as subjects interpreters of different levels of expertise.. It is hypothesized that expert interpreters. would show better anticipation abilities, and such advantage could be identified in neural correlates. Chapter 2 provides a synopsis of past literature on interpreting expertise and anticipation studies using ERP. Chapter 3 describes the methods used in the present experiment.. Chapter 4 summarizes the results of the experiment.. discusses the findings, and Chapter 6 is the conclusion.. 4. Chapter 5.

(15) Chapter 2. Literature Review. This chapter will begin with an overview of how expert and novice interpreters differ in interpreting performance. Section 2.2 will point out how language anticipation has been studied.. Section 2.3 will introduce brain imaging techniques. that might contribute to studying sentence comprehension and anticipation in interpreters, and the few interpreting studies that adopted such methods, focusing especially on the ERP technique. Section 2.4 will introduce how anticipation has been studied using the ERP technique. The chapter will end with hypotheses for the experiment and a summary.. 2.1 Performance differences in expert and novice interpreters Interpreter trainers are naturally concerned about the developmental progress of their students.. They would like to know how students can acquire interpreting skills. efficiently and effectively2. A prerequisite to this goal is to first identify what these skills are, and whether they are of interest in interpreter training. To identify the skills that are pertinent to successful language interpreting, it would be useful to view interpreting as a form of human information processing (Gerver, 1971). Specifically, simultaneous interpreting3 (SI) has been studied from the perspective of cognitive psychology, and has been rendered into various mental structures and procedures that as a whole describe SI as an information processing activity (Pöchhacker, 2004).. Scientific research has endeavored to deconstruct or. 2. The mode of interpreting referred to in this study is spoken-language interpreting. Simultaneous interpreting (SI) is the act of listening to the speaker’s speech in one language (source language) and reformulating the message into another language (target language) almost simultaneously, all the while continuing to pay attention to the speaker’s incoming flow of messages and maintaining an acceptable level of fluency and accuracy in the output. 3. 5.

(16) decompose the act of interpreting into different segments, or subskills (Moser-Mercer, Lambert, Daró, & Williams, 1997).. The breaking down of a complex set of skills. into various components is beneficial to the studying of interpreting, because it allows researchers to examine the acquisition of interpreting skills from an ecological perspective and as a developmental process (Moser-Mercer, 1997a). From the perspective of expertise studies, these processes could be then converted into operationalized tasks that can be deliberately practiced in order to progress through the various developmental stages before attaining expertise (Ericsson, 2010; Hoffman, 1997). Expertise studies suggest that there is a developmental progression in which novices attain incremental qualitative shifts through effortful deliberate practice before becoming an expert (Ericsson, 2010; Feltovich, Prietula, & Ericsson, 2006; Hoffman, 1997). Specifically, during the development of expertise, considerable skills, knowledge, and mechanisms are amassed to monitor and control cognitive processes, resulting in a more efficient and effective performance of a delimited set of tasks (Feltovich et al., 2006).. According. to Feltovich et al. (2006), automation is an important step in the development of expertise, because it releases valuable cognitive resources to deal with other higher functions, such as reasoning, comprehension, inference, monitoring, integration, and accessing available knowledge. Automation can be seen as performing a set of tasks more effectively and efficiently, and this phenomenon is not only observed in natural settings, but also reflected in the behavioral data collected in experimental settings. For example, Spelke, Hirst, and Neisser (1976) asked college students to practice reading unfamiliar text while noting down words read by an experimenter for one hour every day for six weeks. By the end of the experiment, the students could do this without sacrificing reading speed or comprehension (as cited in Feltovich et al., 2006, p. 53).. Ericsson (2010) believed that these observable changes in task 6.

(17) performance could be due to changes in cognitive mechanisms that mediate how the brain and nervous system control performance.. How much the physiological. systems of the body have adapted could also be reflected in these performance changes. Automation has not been researched extensively in interpreting studies.. Bajo,. Padilla, and Padilla (2000) found that interpreters seemed to be faster in accessing lexical and semantic information from long-term memory, but it is difficult to conclude that the results pointed to an automation of lexical access in interpreters. However, by studying how expert and novice interpreters perform the same task and comparing their performance (i.e., the expert-novice paradigm), the nature of these procedural changes could be better understood. In fact, expertise in conference interpreting has long been an area of much attention in interpreting studies. Researchers have attempted to identify the factors underlying the better performances of expert interpreters as compared against those of interpreting students or novice interpreters. Some have investigated whether the differences lie in linguistic abilities. Christoffels et al. (2006) found that interpreters had better language proficiency than equally proficient bilinguals.. On the other hand,. Dillinger (1990) did not find interpreters to possess a different set of listening comprehension skills, but the results from Yudes, Macizo, Morales and Bajo (2012) indicated that they actually have better reading strategies.. Others have looked at. whether there were differences in cognitive abilities such as working memory, of which results have been mixed. For example, some studies found that expert interpreters had better working memory than novice interpreters, (Bajo et al., 2000; Christoffels, de Groot, & Kroll, 2006; Christoffels, de Groot, & Waldorp, 2003), while others did not find significant differences between the two groups (Köpke & Nespoulous, 2006; Liu, Schallert, & Carroll, 2004; Padilla, Bajo, & Macizo, 2005). 7.

(18) In terms of executive control, some studies suggested that interpreters possessed better mental flexibility to allow them to shift between mental sets, but not necessarily stronger inhibition control (e.g. Yudes, Macizo, & Bajo, 2011), while others indicated that interpreters had better inhibition control because they were capable of overcoming the articulatory suppression effect and coordinate comprehension and production smoothly (e.g. Yudes, Macizo, & Bajo, 2012).. Therefore, findings. regarding whether expert interpreters were better at executive control were also mixed. Liu (2008) surveyed past empirical studies and summarized that expert interpreters produced more accurate output, committed fewer errors and omissions, and sounded less literal and more coherent and natural.. This was probably because expert. interpreters selected more important information to interpret, comprehended at a more conceptual level, recognized patterns of equivalence between two languages, took a more semantic approach in translation, processed larger chunks of information, and anticipated upcoming information. That expert interpreters are better anticipators is an assumption that has rarely been examined. One possible reason is the difficulty of identifying the exact indicators of the anticipation mechanism. This aspect will be further discussed in the following section.. 2.2 Language anticipation and interpreting Anticipation is a routine in our daily lives.. We expect to find food in a. restaurant and know that we will have to pay for it. weather forecaster says a typhoon is coming.. We expect heavy rain when the. We also use anticipation to play tennis. and drive cars. Such ability is also demonstrated through language. following example:. 8. Consider the.

(19) Mary: Don’t you think the soup is rather tasteless? John: Here’s the pepper. Mary: Thanks.. Without the ability to anticipate, John would have given “yes” or “no” as an answer. Instead, he knew that Mary’s question was a rhetorical one and inferred that she wanted more flavor to the soup, so upon seeing a pepper shaker on the dining table he offered it to Mary.. Linguistic and pragmatic knowledge triggered this chain of. mental calculation, which probably happened within a couple of seconds, so fast that Mary was probably impressed by how gentlemanly John was.. However, it is also. possible that John knew what was expected of him, because he was probably too slow to pass the pepper shaker in their last dinner together and was thus reproached by Mary.. To avoid embarrassment again, he learned from the prior experience and. acted swiftly. The interlocution above is a typical illustration of anticipation, without which daily conversations cannot happen smoothly.. Turn-taking is essential in natural. conversations, and the normative rule is for listeners to respond to speakers as soon as they have finished.. Anticipation helps the listener prepare for articulation right after. the speaker has ended so as to prevent any gap between turns.. An off-line study has. shown that speakers and listeners are capable of switching their roles in natural conversation without temporal gaps because they are able to anticipate the end of a turn based on the predictions made about the content and word forms of the turn (Magyari & de Ruiter 2008).. An on-line EEG study has shown that when people. could predict the words in the turn, they could anticipate the turn-end as early as 1.7 seconds before the actual ending (Magyari, Bastiaansen, de Ruiter, & Levinson, 2011). 9.

(20) People not only know when it is their turn to speak in a conversation, they can also predict upcoming words while reading by using their knowledge from the wider discourse.. Past studies reviewed by Otten and Van Berkum (2008) showed that. people anticipate upcoming syntactic structure, the grammatical role of upcoming words, upcoming meaning (e.g. inferring that a porcelain vase will break after reading that it fell from a 20-story building), words from a specific field (e.g. expecting a word that denotes a type of meat when reading the unfinished sentence “the vegetarian never ate…”), and specific additional information about particular discourse entities (e.g. after reading “David praised Linda because…” people would expect something about Linda being mentioned). If anticipation is so natural, it should be safe to assume that interpreters also anticipate during language comprehension.. In fact, anticipation has been regarded as. an important strategy by practitioners, trainers, and researchers (Chernov, 2004; Gile, 2009; Seeber, 2001; Setton, 1999; Van Besien, 1999; Vandepitte, 2001).. Gile (2009). pointed out that when interpreting simultaneously, interpreters anticipate the incoming information in order to reduce the level of uncertainty.. This relieves the cognitive. load interpreters experience and allows them to allocate more cognitive resources to attend to other tasks (e.g. comprehension, production, or monitoring), which in turn should help achieve a higher quality of interpretation.. Chernov (2004) went further. in suggesting that simultaneous interpreting is only possible when interpreters could anticipate the speaker’s statement in real time.. He believed that anticipation could. facilitate comprehension, without which comprehension would always be fragmented, faulty, and inadequate. Despite its recognized importance, scarce literature directly addresses the acquisition of anticipation skills.. Kalina (1992) briefly mentioned anticipation and. cloze exercises as useful practices in the SI classroom. 10. Findings from empirical.

(21) evidence have suggested that performance in cloze tests is indicative of interpreting performance (Gerver et al. 1989, Moser-Mercer 1985). Therefore, Pöchhacker (2011) developed a SynCloze test that modeled on Gerver’s cloze exercises and Moser-Mercer’s oral message completion, as they predicted interpreting performance. The test scores correlated moderately with students’ end-of-course consecutive interpreting exams. Although anticipation has been identified as an important aspect of interpreting, there has been little empirical research on the topic. This could be because anticipation was defined and examined rather narrowly in earlier days.. Van Beisen’s. (1999) definition of anticipation succinctly captures how it was defined in earlier research: anticipation is “the production of a constituent (a word or a group of words) in the target language before the speaker has uttered the corresponding constituent in the source language” (p.250).. Evidence of such anticipations often comes from. corpus analysis, analyzing the temporal aspects of interpreters’ output against that of speakers.. For example, Van Besien (1999) examined the interpretation of. professional interpreters working between German and French, as the two languages differed in the position of the verb: French verbs usually followed the subject, while German verbs often came at the sentence-final position, after the object. found that anticipation (defined above) occurred every 85 seconds.. Van Besien. Among the 78. instances of anticipation, 60 were verb anticipations that involved uttering the French verb before the German speaker said the sentence-final verb. However, empirical on-line evidence showing that interpreters were actually anticipating is rare. This is probably because, according to Van Besien’s definition above, anticipation per se is only observable as an end product, and could be observed only through behavioral studies. Nevertheless, it is difficult to examine the various cognitive mechanisms that might be at work that eventually contributed to the 11.

(22) behavioral results.. It is also difficult to pinpoint whether or not the cognitive. mechanism that is being examined is actually anticipation.. For example, it is. possible that interpreters have faster processing speed or have better linguistic knowledge. This was why the narrower sense of anticipation (i.e. interpreters uttering words in the target language before the speaker said them) has been criticized as being too product-oriented and unhelpful in unveiling the actual cognitive processes that lie behind anticipation (Vandepitte, 2001). Researchers have thus taken into consideration the recent developments in psycholinguistics and neurosciences. Vandepitte (2001) reviewed and adopted the framework of relevance theory and posited that anticipation “is the interpreter’s mental generation of (parts of) assumptions that correspond to those that have not yet been expressed by the speaker” (p. 329), emphasizing that anticipation is a mental process. If anticipation were part of the interpreter’s mental process, it would be useful to look at how this process could be examined.. One way is to use neuroimaging. techniques, which was rarely adopted by interpreting researchers, probably due to technical and knowledge barriers.. In the following sections, a brief summary will. first be given on how brain imaging techniques, which have proven to be quite useful in understanding how the human brain makes predictions during language comprehension, have been used in interpreting studies. Then, a review will be given on how a particular type of technique, the event-related potential (ERP), could shed light on how interpreters anticipate.. 2.3 Brain imaging in interpreting studies 2.3.1 High spatial resolution techniques The improvement and greater accessibility of brain imaging technologies have. 12.

(23) allowed researchers to employ such techniques as functional magnetic resonance imaging (fMRI), positron emission tomography (PET) and anatomical magnetic resonance imaging (aMRI) to study language interpreting.. fMRI and PET are. spatially-sensitive metabolic functioning techniques that exploit the fact that brain regions engaging in higher local activity need more glucose and demand higher rates of cerebral blood flow (Ingram, 2007). Therefore, when there is an increase in blood flow or the proportion of oxyhaemoblogin in local blood vessels, the magnetic field changes over seconds to minutes, and radiographic or magnetic detection techniques can pick up these changes. Anatomical MRI is a kind of structural MRI that supplements fMRI by providing anatomical reference for visualization of activation patterns. Some researchers have used these techniques to study how professional interpreters perform their tasks and whether there are neural correlates of the observed performances; and if there are, what they are. For example, Rinne et al. (2000) used PET to measure brain activation while native Finnish professional interpreters performed simultaneous interpreting into English (L2) and Finnish (L1). They found that the left prefrontal dorsolateral cortex, which includes areas related to lexical search, semantic processing, and verbal working memory, were activated during the task.. More interestingly, there were. more areas activated when the interpreters had to interpret into the non-native English. This phenomenon reflects the claim by many professional interpreters that it is more difficult to interpret into the non-native languages. Elmer, Meyer, Marrama, and Jäncke (2011) used fMRI to measure whether professional interpreters and control subjects responded differently when they were required to attentively discriminate and categorize non-verbal auditory stimuli. Although in-scanner behavioral results showed no difference between the two groups, there were significant differences in terms of blood oxygen level-dependent responses 13.

(24) in frontal-parietal regions. The authors concluded that the long-term language training interpreters have received modulated the functional architecture that supports the discrimination of non-verbal stimulus attributes. Hervais-Adelman, Moser-Mercer, and Golestani (2011) used the aMRI technique and scanned eleven interpreting students before, during, and after a 15-month training period. They compared the high-resolution structural MRI data with that of thirteen students whose age, sex, and multilingualism were matched against the interpreting students, but were studying in other fields. were similar.. Scanning time points of both groups. After comparing the three sets of data collected at the three time points,. the researchers found that there is evidence for brain structural plasticity in interpreting students in the left middle frontal gyrus, the left supramarginal gyrus, the left pars orbitalis, the left middle temporal gyrus, and the rostral anterior cingulate, which are involved in semantic processing, semantic memory, cognitive control of memory, and verbal working memory. These regions were significantly larger than the size measured at baseline. However, such plasticity in the regions examined was not seen in the non-interpreting students, whose data showed no significant difference from baseline. The study provides further evidence of brain plasticity due to intensive training. 2.3.2 High temporal resolution technique: event-related potentials (ERP) PET, fMRI, and aMRI are brain imaging techniques that have high spatial resolution.. In contrast, event-related potential recording (ERP) is a time-sensitive. technique that measures changes in brain electrical activity which enables researchers to study the neural events in on-line processing (Ingram, 2007). ERPs are measured by using electroencephalography (EEG), a procedure that measures voltage fluctuations over time by placing electrodes on the scalp.. The EEG is time-locked to. the stimuli of interest, but the signals are weak against the backdrop of numerous 14.

(25) ongoing neural activities.. In order to examine the brain’s response to the stimuli,. multiple samples of the same stimulus event with time-locked signal averaging are used to extract the ERPs of interest.. The time resolution of the ERP technique is in. the milliseconds, thereby making it particularly useful in examining time-sensitive cognitive skills such as online language comprehension. Among the many ERP features, the N400 component is of particular interest.. It. is a negative-going potential, that peaks around 400-ms after stimulus onset, and is indicative of, for example, the semantic relationship between a meaningful stimulus (usually a word, but could be anything potentially meaningful, such as sign-language signs, pictures, and gestures) and the context in which it occurs (Kutas & Federmeier, 2011; Kutas & Hillyard, 1984). The N400 amplitude is modulated by several factors that influence the ease of accessing information from semantic memory, such as word frequency (smaller for higher frequency) and semantic context (smaller for higher relatedness) (Kutas & Federmeier, 2009; Kutas, Moreno, & Wicha, 2009). For example, in a sentence like “I couldn’t get into my house because I forgot to bring my keys/dog,” the N400 amplitude to the word “dog” would be larger (i.e. more negative) than that of “keys”, because “dog” in this instance was far less associated with the previous sentential context, thus becoming a sort of semantic violation. In addition, there is a graded relation between the N400 amplitude and the cloze probability4 of the target word (Fedemeier & Kutas, 1999).. For example, in a. sentence like “It occurred to Henry that today is Valentine’s Day, so before he went home, he decided to buy some roses/tulips/magazines,” the word “roses” would be the most likely completion of the sentence, as it is strongly associated with Valentine’s Day. On the other hand, “tulips” is plausible but less likely; thus, its N400 would be. 4. Cloze probability is operationally defined as the proportion of people who give a particular word as the most likely completion of a sentence fragment (Taylor, 1953). 15.

(26) larger than that of “roses”. However, the word “tulips” shares semantic features with “roses” so it is more readily accessible in the long term memory, or it requires less effort to integrate than “magazines”, which barely shares any semantic feature with “roses”. Therefore, “tulips” will elicit a smaller N400 than “magazines”. N400 effects5 could be examined by using a priming paradigm, in which subjects are presented a pair of items, of which the first one is usually the prime and the second word the target.. Such a paradigm was used in the study of Elmer, Meyer,. and Jancke (2010) to investigate the impact of long-term language training on single word processing. They recruited eleven German professional interpreters and the same number of controls who were matched in proficiency and age of L2 (English) acquisition and compared their N400 effect in semantically congruent and incongruent noun pairs.. The subjects were asked to decide whether. visually-presented noun pairs that appeared within the same language (English and German) or across different languages (English to German and German to English) were congruent or not.. Results showed enhanced N400 responses in the interpreter. group when the incongruent pairs appeared within the same language and when both congruent and incongruent pairs appeared in the German to English (L1 to L2) condition.. The authors suggested that this might reflect training-related altered. sensitivity to lexical-semantic processing within and across L1 and L2.. It was. suggested that a follow-up study examining auditory sentence comprehension, which more closely resembles the task interpreters are trained to perform, be conducted. N400 could also be used to study the brain’s reaction to switching between two languages.. Code switching between L1 and L2 is often associated with a cost, as. reflected in longer reaction times and larger N400 amplitudes (Kutas et al., 2009). 5. The “N400 effect” in language studies refers to the neural reflection of a linguistic anomaly. It is calculated by the subtraction between two conditions to reveal N400 amplitude differences (Kutas & Federmeier, 2011). 16.

(27) Proverbio, Leoni, and Zani (2004) recruited professional interpreters (to control the factor of language proficiency in L1 & L2) and asked them to read short sentences that were either unmixed or mixed. Unmixed sentences were entirely in English or Italian, and the body of the mixed sentences was in either English or Italian, but the final word in the other language (Italian or English). Half of the sentences had an unexpected final word that caused a semantic incongruence.. The results of this four. (language condition) by two (congruity) design showed that at the N400 level, ERPs to L2 words were significantly larger to L1 words, collapsed across congruity. They termed this the “switch effect”, meaning that the N400 is larger in mixed language conditions than in unmixed ones. The discussion above showed that the ERP technique is an insightful way of examining real-time processing of language. Researchers have asked interpreters to judge the congruity of a particular target against its prime or sentential context. However, to examine whether or not expert interpreters are better anticipators, it would be more beneficial to use stimuli at the discourse level instead of at levels of word pairs or a single sentence, as interpreters naturally work in a context-rich discourse environment.. The following section will discuss how certain ERP. components are sensitive to discourse-level language comprehension and how they reveal that listeners do anticipate in an unfolding message.. 2.4 Anticipation in ERP studies 2.4.1 N400 and anticipation The sentence violation paradigm (i.e., using words that are either congruent or incongruent at the ending of sentences) was used in the very first study that discovered the N400 component (Kutas & Fedemeier, 2011; Kutas & Hillyard, 1980). 17.

(28) and has since been applied to the discourse level to examine whether listeners processed larger pieces of message in the same way as single sentences.. Van. Berkum, Zwitserlood, Hagoort, and Brown (2003) designed two experiments that directly compared the impact of discourse-level information on an unfolding sentence to one without a preceding context.. Subjects listened to sentences that ended in. equally plausible words (“Jane told her brother that he was exceptionally quick/slow.”), but the sentences were preceded by contextual discourse in the first experiment (“As agreed upon, Jane was to wake her sister and her brother at five o’clock in the morning.. But the sister had already washed herself, and the brother. had even got dressed.”), while being presented alone in the second experiment.. In. the first experiment, the discourse-anomalous words elicited a typical N400 effect that sometimes began before the words ended (e.g. “slow” in this particular example), but the same word did not elicit an N400 component in the second experiment.. The. experiments showed that listeners could relate the unfolding words rapidly to the preceding context, and discourse-dependent N400 effects were identical with sentence-dependent N400 effects. Both the studies on cloze probabilities and contextual discourse revealed that for expected target words, there was an inverse relationship between the amplitude of the N400 of the target words and the level of contextual constraint. This means that, usually, the higher the level of contextual constraint, the smaller the N400 amplitude for expected target words. However, unexpected words could also follow highly constraining context (e.g. “He painted the walls in the wrong HUES.”).. Federmeier,. Wlotko, De Ochoa-Dewald, and Kutas (2007) demonstrated that N400 amplitudes are graded by expectancy (indexed by the cloze probability) but unaffected by contextual constraint, therefore control methods other than cloze probability of the target word is needed to determine and control the level of contextual constraint. Previous studies 18.

(29) used the Latent Semantic Analysis (LSA) methodology to control the contextual constraint by looking into the semantic distance from the context sentence to the target words (Davenport & Coulson, 2011; Kuperberg, Paczynski, & Ditman, 2011). LSA, according to its inventor Thomas Landauer, is “a fully automatic mathematical/statistical technique for extracting and inferring relations of expected contextual usage of words in passages of discourse” (Landauer, Foltz, & Laham, 1998, p.263).. It applies a form of factor analysis called singular value decomposition. (SVD) to a matrix that represents the text, and through various mathematical steps of dimension reduction, a cosine between vectors in the reduced dimensional space is produced to present the similarity between words (Landauer et al., 1998).. Empirical. studies have demonstrated that LSA “produces measures of word-word, word-passage, and passage-passage relations that are well correlated with several human cognitive phenomena involving association or semantic similarity…[and] closely approximate[s] human judgments of meaning similarity between words” (Landauer et al., 1998, p. 260). For example, it has successfully simulated a lexical priming study by Till, Mross, and Kintsch (Landauer & Dumais, 1997). Conference interpreters are routinely exposed to the processing of meaningful and contextualized discourse. Therefore, the present study adopted a discourse level sentence violation (congruent/incongruent) paradigm to investigate how whether interpreters of different levels of expertise processed sentences differently.. In. addition, the LSA was used to control the level of constraint between the context and target word. Despite the sensitivity of the N400 to offline semantic expectancy (as indexed by cloze probability), there was a debate about whether the N400 could actually index “anticipation” rather than just “integration”. Some argued that listeners did not anticipate the target word; they merely found the word easier to integrate into the 19.

(30) prior context when it was congruent, and difficult when it was not (Hagoort, Baggio, & Willems, 2009). Hagoort and colleagues posited that the N400 “reflect[s] processes involved in the integration of the meaning of a word into the overall semantic representation constructed for the preceding language input” (Hagoort et al., p. 820). On the other hand, Federmeier and Kutas (1999a) believed that some degree of anticipation has occurred, because they showed that in highly constraining contexts (e.g. “He caught the pass and scored another touchdown.. There was. nothing he enjoyed more than a good game of…”), even though both “baseball” and “monopoly” had low cloze probabilities when evaluated offline, subjects in an ERP experiment would generate a smaller N400 effect for “baseball” than for “monopoly”. They argue that this was because the language processing system was strongly predicting the word “football”, thus pre-activating conceptual features of that word, which shared more features with within category anomalous words (“baseball”) than with between category violations (“monopoly”). An ingenious design to provide more evidence that listeners are actually anticipating specific words came from studies looking at the EEG signals of words preceding the target word that was presumably being anticipated. Otten, Nieuwland, and van Berkum (2007) asked subjects to listen to short stories in Dutch that were highly constraining for a specific noun, or stories that were less predictive but contained the same prime words as the predictive stories. The gender marking (neuter or common) on prenominal adjectives was manipulated so that it was either congruent or incongruent to the anticipated noun.. Results showed that when the. stories were predictive, adjectives with an unexpected gender inflection would evoke a negative potential around 300 to 600-ms.. When stories were not predictive, no. such effect existed. De Long, Urbach, and Kutas (2005) manipulated the English indefinite article a/an in a similar design and saw the same results. 20.

(31) One way to reconcile the debate of whether the N400 indexed anticipation or integration is by taking into consideration the roles that the left and right hemispheres of the brain might play.. Federmeier and Kutas (1999b) devised a visual-half-field. study to explore whether the two hemispheres played different roles. Subjects read pairs of sentences that ended in either expected words, unexpected words from the expected semantic category (within-category semantic violation), or unexpected words from an unexpected category (between-category semantic violation).. The. critical sentence-final target words either appeared in the left or right visual field, which corresponds to the right and left hemispheres respectively. The results showed that within-category anomalous words presented to the right visual field (left hemisphere) elicited smaller N400s than between-category anomalous words, while both kinds of unexpected words elicited similar N400 effects in when presented to the left visual field (right hemisphere), suggesting that the brain is lateralized in the sense that the right hemisphere is more “integrative”, while the left is more “predictive”. This is because the unexpected words that shared semantic features with the expected word (e.g. football vs. baseball in the previous example) elicited a smaller N400 than the unexpected words that belonged to a totally different word category (e.g. football vs. monopoly) in the left hemisphere, showing that the left brain has probably activated the semantic features of a particular word due to anticipation.. Wlotko and. Federmeier (2007) conducted a similar experiment and replicated the findings. 2.4.2 Late positivity and anticipation Despite the benefits that prediction brings to language comprehension, recent studies have shown that it might come with a cost, indexed by a late positivity appearing around 600-ms after the critical stimuli (reviewed in Van Petten & Luka, 2012).. It has been referred to as the “semantic P600” (Kuperberg, 2007), the late. positivity complex (LPC) (Federmeier et al., 2007), and the post-N400 positivity 21.

(32) (PNP) (Van Petten & Luka, 2006).. In the past, this characteristically parietal. positivity that was observed 600-ms post-stimulus onset and lasting until 900-ms (often referred to as the P600) has been reported when there were syntactic manipulations such as grammaticality violations (e.g. subject verb agreement), sentence ambiguity (e.g. garden path sentences), and higher degrees of sentence complexity (e.g. complex clauses) (Kolk & Chwilla, 2007; Osterhout & Holcomb, 1992). However, since grammatically well-formed sentences sometimes also elicited PNPs, and some were observed in the frontal region, scholars have suggested different hypotheses to account for this phenomenon. Federmeier et al. (2007) observed a frontal positivity between 500 and 900-ms for unexpected words completing strongly constraining sentences. Since a PNP was not observed for expected words completing strongly constraining sentences nor unexpected words completing weakly constraining sentences, they suggested that the PNP indexed the need to “override or suppress a strong prediction for a different word or concept” (p. 91).. Delong, Urbach, Groppe, and Kutas (2011) reanalyzed data. from Delong et al. (2005) that used the English indefinite article a/an to investigate whether readers anticipated particular nouns, and reported similar frontal positivities between 500 to 1200-ms when readers encountered less probably noun continuations. They hypothesized that the frontal positivity indexed the consequences of disconfirming predictions. Kolk and Chwilla (2007) suggested that the PNP, or more specifically the P600, “could be an aspect of cognitive or executive control of language” (p. 258), implying that the brain is constantly monitoring how language is understood, so that in cases of conflict between different possible sentence interpretations, a re-analysis could be done in time to prevent comprehension errors. Kuperberg (2007) built upon the monitoring theory and suggested that the language processing system has two streams to comprehension: “one that links incoming 22.

(33) semantic information with existing information stores in semantic memory, and another that combines relationships between people, objects, and actions to construct new meaning” (p. 45).. The PNP thus reflected a combinatorial analysis when. conflicts between the two systems happen. Van Petten and Luka (2012) reviewed 45 ERP studies that used a congruent versus incongruent paradigm, and 13 studies that adopted a high versus low cloze probability (both congruent) paradigm.. They found. that a parietal PNP was more commonly revealed in the former paradigm, indicating a reprocessing cost associated with verification and repair. On the other hand, a frontal PNP was identified in the latter paradigm, signaling a failed prediction being made. Whether or not this frontal PNP reflected inhibition (as suggested in Kutas 1993) or not was reserved for further exploration. Therefore, in addition to the N400 component, whether or not a PNP arises from the present study will also be examined.. If the amplitude of incongruent endings is. more positive than that of congruent ones, then there is a PNP. Since the LSA and cloze probability will be used to ensure high contextual constraint, of particular interest will be the absence or appearance of a frontal positivity.. 2.5 Summary Past literature pointed to the usefulness of leveraging the high temporal resolution of the ERP technique in observing language comprehension processes. Many studies have shown that listeners predict upcoming information in an unfolding sentence. Prediction of specific words was also observed for sentences that were contextually constraining. Since anticipation is elusive to identify from behavioral studies but theoretically important for interpreters, it would be helpful to adopt this brain imaging methodology to investigate whether or not interpreters with different. 23.

(34) levels of expertise are different in terms of anticipation.. To extend the study of. Elmer et al. (2010), aurally presented sentence pairs that were entirely in Chinese or switched to English in the second sentence were used as stimuli. These contextual constraining sentences ended in either congruent or incongruent words. The N400 component and PNP were examined and compared across different groups of interpreters. Previous literature has suggested that expert interpreters probably were more capable than novice interpreters of using contextual information and semantic cues to anticipate upcoming information (Liu, 2008).. Therefore, it was hypothesized. that as interpreting expertise improves, the congruity effect would become larger, because expert interpreters’ stronger anticipation of the (congruent) words would in effect make the incongruent words more difficult to access or integrate.. It was also. hypothesized that the switch effect would reduce as expertise improves, because expert interpreters are more accustomed to the handling of two languages at the same time. The present study investigated whether there was neural evidence to support such postulations.. 24.

(35) Chapter 3. Methods. The present study used event-related potentials (ERP) to investigate the neural correlates of anticipation in interpreters of different expertise levels.. 3.1 Subjects Three groups of subjects were recruited to participate in the experiment: Professionals, Graduates and Students. The Professionals group (Professionals) consisted of 12 active conference interpreters in Taiwan (6 males, mean age = 39.6 years, range = 33-48 years). They have been working for an average of 12.7 years (range: 7-20 years).. In addition, they were all professionally trained and hold a. master’s degree in conference interpreting. The Graduates group (Graduates) consisted of 12 recent graduates or senior students (3 males, mean age = 28.8 years, range: 25-34 years) of two graduate-level conference interpreting programs in Taiwan. They have all taken the professional conference interpreting exams jointly organized by the two interpreting schools between 2010 and 2012 after completing all required courses.. All of them practice. conference interpreting on a part-time or full-time basis, and they have been working for an average of 2.1 years (range: 1-4 years). The Students group (Students) consisted of 13 students of conference interpreting (7 males, mean age 25.6 years, range: 22-33 years). They are all currently in their first year of postgraduate conference interpreting programs in Taiwan. These programs take at least two years to complete; therefore, at the time the students participated in the experiment, they have only received training in introductory courses such as sight translation and consecutive interpreting for one semester (about four months). 25.

(36) All 37 subjects in the three groups are native speakers of Mandarin Chinese and use English as their working language.. As age of second language (L2) acquisition. and level of proficiency have been identified as factors that influence language processing (Elmer et al. 2010; Kutas & Federmeier, 2009), information regarding age of L2 acquisition and scores on standardized English tests such (e.g. TOEIC, TOEFL, and IETLS) were obtained. Age of L2 acquisition did not differ significantly, H (2) = 3.84, p = .147. Ten subjects could not recall their scores from standardized English tests (6 Professionals, 1 Graduate, and 3 Students). However, since Professionals have been interpreting professionally on the market for an average of 12 years, and Graduates and Students would necessarily have proficient English to be admitted into the interpreting program, it was assumed that they are highly proficient in English. For those who reported, all had a comparable score of at least 107 on the TOEFL iBT test.. All subjects are right-handed according to a simplified version of. the Edinburg handedness inventory (Oldfield, 1971). None of the subjects reported having any known neurological or psychiatric disorders, nor were they administered with medication to treat neurological or psychiatric disorders around the time of the experiment.. Ten of the subjects reported having a left-handed family member. All. subjects signed informed consent and received cash payment for their participation.. 3.2 Materials 3.2.1 Overview The material consisted of 200 pairs of sentences. Each pair included a Chinese context sentence followed by a carrier sentence which was consisted of a carrier body and a sentence-final target item. Carrier sentences were in one of the four following conditions: a Chinese carrier body with a congruent target (the non-switched,. 26.

(37) congruent condition), a Chinese carrier body with an incongruent target (the non-switched, incongruent condition), an English carrier body with a congruent target (the switched, congruent condition), and an English carrier body with an incongruent target (the switched, incongruent condition). The English carrier sentences and target items were translated versions of the Chinese ones. The 200 pairs of sentences were separated into two sets, the upper (the first 100 sentences) and lower (sentences 101 to 200). The Chinese context sentence established the contextual constraint and was unique across the 200 pairs.. The 200. carrier bodies mirrored each other across the two sets across subjects, i.e. the same carrier body would appear once in the upper set and once in the lower set, each ending in a different target item, which effectively separates the target items into the upper and lower sets as well (see Table 1). As a result, the carrier sentences were identical up until the target item. When the context sentence and the target word were chosen from the same set (i.e. both from the upper set), the target would be a congruent ending to the sentence pair; on the other hand, when the context and the target were chosen from different sets (i.e. one from the upper while the other from the lower set), the target would be an incongruent ending to the sentence pair. More details are given below with regard to how the stimuli were constructed.. Table 1 Upper and lower list segmentation. Upper list Lower list. Chinese Context sentence. Carrier body. Target. Carrier body. Target. #1-100. #1-100. #1-100. #1-100. #101-200. #101-200. #1-100. #101-200. #1-100. #1-100. Congruent carrier sentence. 27. Incongruent carrier sentence.

(38) 3.2.2 Stimulus construction For this experiment, the contextual constraint of the context sentence on the target item was first examined by using the Latent Semantic Analysis (LSA) methodology. The Chinese LSA Website6 was used to measure the cosine of the target item and the context sentence (“term to document” method7). The Academia Sinica Balanced Corpus of Modern Chinese8 was used as the referenced semantic space. The process of constructing a sentence pair stimulus is described as follows. First, a context sentence and its carrier body, along with the target item, are constructed for the upper set, as illustrated in Table 2.. Table 2 Example of congruent stimuli construction in upper set Context sentence. Carrier body. (upper set) 熱帶小島的天氣變化很快。 你下午出門的時候要記得帶你的 (The weather of tropical islands changes very fast.). (When you go out in the afternoon, remember to bring your). Congruent Target item 雨傘。 (umbrella.). The first sentence (“The weather of tropical islands changes very fast”) is the context sentence, the second sentence (“When you go out in the afternoon, remember to bring your umbrella”) is the carrier body, while the sentence-final noun (“umbrella”) is the target item. The cosine between the target item and the entire context sentence (“term to document” method) was examined by using the “pairwise comparison” function on the Chinese LSA Website to ensure the relatedness of the target item. 6. http://www.lsa.url.tw/modules/lsa/ The “term to document” method is a function in the Chinese LSA Website that calculates the cosine between a term and a sentence/paragraph. Other functions include “term to term”, in which the cosine between two terms are computed, and “document to document”, in which sentences or paragraphs can be compared against one another. 8 http://app.sinica.edu.tw/cgi-bin/kiwi/mkiwi/scorp.pl 7. 28.

(39) For the example given above, the cosine between “umbrella” and the context sentence (“The weather of tropical islands changes very fast”) was 0.161. The next step is to construct an incongruent version of the carrier sentence.. The. “near neighbor” function9 of the Chinese LSA Website was used to generate a list of terms that were related to the target item. These terms were not necessarily “semantically” related to the target item, but simply appeared in the list because it was contained in the Balanced Corpus. The threshold of the cosine was set at 0.01 to generate terms that were more distant from the target item.. From the list, a term that. had a much lower cosine than and intuitively different from the original target item would be chosen.. If an appropriate term cannot be found from the returned list, a. new target item would be devised independently.. In the example above, 護照. (passport) was chosen as the new target item, whose cosine with 雨傘 (umbrella) is 0.022. This new term would be used to construct the incongruent version of the second sentence:. 你下午出門的時候要記得帶你的護照。 (When you go out in the afternoon, remember to bring your passport.). As seen from the example above, the same carrier body will end with the new target item and the second sentence will become incongruent when preceded by the original context sentence, as illustrated in Table 3:. 9. The “near neighbor” is a function in the Chinese LSA Website that allows users to check other terms that are in a user-defined semantic space. The terms in the list generated by this function have varying cosine values to the original term and may be semantically close or distant from it. 29.

(40) Table 3 Example of incongruent stimuli construction in upper set Context sentence. Carrier body. (upper set) 熱帶小島的天氣變化很快。 你下午出門的時候要記得帶你的 (The weather of tropical islands changes very fast.). (When you go out in the afternoon, remember to bring your). Incongruent Target item 護照。 (passport.). This new incongruent carrier sentence is actually identical to its congruent counterpart except for the sentence–final target item. The cosine between the incongruent target item (“passport”) and context sentence (“The weather of tropical islands changes very fast”) was then calculated.. In this instance, it was -0.001.. In the case that the “term. to document” cosine is high, opinions from independent raters would be sought to confirm that the stimuli would not be mistaken as congruent.. This completes the. construction of a sentence pair in the upper set. The incongruent version of the carrier sentence in the upper set was then used to construct a new context sentence in the lower set.. This new context sentence. would render the target item of the originally incongruent carrier sentence congruent. See Table 4 for an example:. Table 4 Example of congruent stimuli construction in lower set Context sentence. Carrier body. (lower set) 我們去曼谷的班機是晚上起飛,機票 你下午出門的時候要記 和旅行箱我都處理好了。 得帶你的 (Our flight to Bangkok takes off in the evening, and I’ve already taken care of the plane ticket and luggage.). (When you go out in the afternoon, remember to bring your). 30. Congruent Target item. 護照。 (passport.).

(41) The cosine between the (congruent) target item and the newly devised context sentence would be checked for relatedness.. In the above example, it was 0.249.. The incongruent version of the carrier sentence that would follow this new context sentence in the lower set is the corresponding carrier sentence in the upper set:. Table 5 Example of incongruent stimuli construction in lower set Context sentence. Carrier body. (lower set) 我們去曼谷的班機是晚上起飛,機票 你下午出門的時候要記 和旅行箱我都處理好了。 得帶你的 (Our flight to Bangkok takes off in the. (When you go out in the. evening, and I’ve already taken care of the plane ticket and luggage.). afternoon, remember to bring your). Incongruent Target item. 雨傘。 (umbrella.). The cosine between the (incongruent) target item and the context sentence would also be calculated.. In this case, it was 0.077.. In short, a context sentence and its carrier sentence containing the congruent target item were first constructed for the upper set of materials.. Then an incongruent. target item was determined to form the incongruent version of the carrier sentence. This new carrier sentence, which was identical to the previous one except for the target item, served as the basis for constructing a new context sentence for the lower set of materials that would in this case render the target item congruent.. This. process was repeated to create 120 pairs of carrier sentences that ended with 240 unique target items and would be preceded by 240 unique context sentences. The congruent and incongruent versions of the carrier sentence that followed the same context sentence were thus constructed simultaneously, ensuring consistency and counterbalance in sentence length and word frequency across conditions.. In 5% of. the cases, the cosine in the incongruent version would be higher than that of the 31.

參考文獻

相關文件

Milk and cream, in powder, granule or other solid form, of a fat content, by weight, exceeding 1.5%, not containing added sugar or other sweetening matter.

The present paper attempts to clarify whether the title of the Dunhuang manuscript Φ142 corresponds to its content. In the process of research the reliability of the

However, due to the multi-disciplinary nature of this subject, schools may consider assigning teachers with different expertise to teach this subject at different levels (S4, 5

Reading Task 6: Genre Structure and Language Features. • Now let’s look at how language features (e.g. sentence patterns) are connected to the structure

™ ™ When ready to eat a bite of your bread, place the spoon on the When ready to eat a bite of your bread, place the spoon on the under plate, then use the same hand to take the

This study aimed to explore the effectiveness of the classroom management of the homeroom teacher by analyzing the process of the formation of the classroom management and

This study was conducted using the key factor from Technology Acceptance Model (TAM), Theory of Reasoned Action, Diffusion of Innovation, and Involve Theory to explore the

This paper aims to study three questions (1) whether there is interaction between stock selection and timing, (2) to explore the performance of "timing and stock