熟手與生手口譯員視譯理解過程差異之眼動研究

全文

(1)A Thesis Presented to the Graduate Institute of Translation and Interpretation National Taiwan Normal University 國立台灣師範大學翻譯研究所碩士論文. Thesis Advisor: Dr. Tze-wei Chen 指導教授：陳子瑋博士. Differences in comprehension process between experienced and novice interpreters – an eye movement study 熟手與生手口譯員視譯理解過程差異之眼動研究. Advisee: Deh-I Chen 研究生：陳德怡. July, 2013 中華民國一零二年七月.

(2) Table of Contents. Acknowledgements List of Tables………………………………………………………………………….iii List of Figures………………………………………………………………………..iv Abstract………………………………………………………………………………….v 摘要...............................................................................................vii Chapter 1. Introduction…………………………………………………………..1. 1.1. Research Background……………………………………………….1. 1.2. Research Questions………………………………………………….7. Chapter 2 2.1. 2.2. Literature Review…………………………………………………..9. Expertise………………………………………………………………….9 2.1.1. Definitions and criteria of experts............................9. 2.1.2. Developing expertise…………………………………………12. Interpreting…………………………………………………………….18. 2.2.1. Sight translation……………………………………………….19. 2.3 Expertise in Interpreting………………………………………….25 2.4. Eye Movements and Comprehension…………………………34. 2.5. Eye Tracking in Interpreting Studies…………………………38. Chapter 3 Methodology………………………………………………………..45 3.1. Experiment I: eye movement and comprehension……….46. 3.1.1 Participants……………………………………………………….46 3.1.2. Apparatus…………………………………………………………47. 3.1.3 Design………………………………………………………………48 3.1.4 Materials…………………………………………………………..50 3.1.5 Procedure………………………………………………………….52 i.

(3) 3.1.6. Data processing…………………………………………………55. 3.1.7 Results………………………………………………………………56 3.2. Audio Clip Evaluation (II) – Fluency and Accuracy……..68. 3.2.1. Fluency evaluation…………………………………………….68. 3.2.1.1 Participants………………………………………………….69 3.2.1.2. Design…………………………………………………………69. 3.2.1.3 Material………………………………………………………70 3.2.1.4 Procedure…………………………………………………….71 3.2.2. Accuracy evaluation…………………………………………..72. 3.2.2.1. Participants………………………………………………….72. 3.2.2.2. Design….……………………………………………………..72. 3.2.2.3 Material………………………………………………………73 3.2.2.4. Procedure……………………………………………………74. 3.2.3 Results……………………………………………………………..74 Chapter 4 Discussion…………………………………………………………..80 4.1 Experience and Expertise…………………………………………82 4.2. Multiple Levels during Reformulation Stage………………91. Chapter 5 General Discussion……………………………………………….96 5.1. Conclusion and Implications…………………………………….96. 5.2. Research Limitations and Future Perspectives…………..99. References…………………………………………………………………………..103 Appendix 1 Eye movement and comprehension experiment source texts (Huang, 2011)……………………………………………………108 Appendix 2. Fluency evaluation grading sheet………………………..115. Appendix 3 Accuracy evaluation grading sheet………………………116. ii.

(4) List of Tables. Table 3.1. Correlations of the three components of interpreting with the three. tasks in this experiment (Huang, 2011)……………………………………………………49 Table 3.2 Mean and standard deviation (in parenthesis) of first fixation duration (FFD), single fixation duration (SFD), gaze duration (GD), rereading time (RRT), and total viewing time (TVT) performed by experienced interpreters…………………………………………………………………………………………..60 Table 3.3 ANOVA results of time duration during sight translation performed by 18 experienced interpreters, and comparisons between different tasks……………………………………………………………………………………………………..61 Table 3.4 Comparison of means and standard deviations (in parenthesis) of the five indices between novice and experienced interpreters. (Novice interpreters data: Huang, 2011)………………………………………………………………63 Table 3.5. Comparison of time duration differences between Novice and. Experienced interpreters performing different tasks…………………………………65 Table 3.6. Comparison of time duration differences when the same group is. performing different tasks………………………………………………………………………68 Table 3.7 T-test analysis results on fluency between novice and experienced interpreters………………………………………………………………………………………..….75 Table 3.8. T-test analysis results on accuracy between novice and. experienced interpreters…………………………………………………………………………75. iii.

(5) List of Figures. Figure 3.1 Eye movement patterns of an experienced interpreter performing silent reading, read aloud, and sight translation……………………………………….58 Figure 3.2 Duration measures of experienced interpreters performing silent reading, read aloud, and sight translation………………………………………………..60 Figure 3.3 Comparison of time duration indices of novice interpreters and experienced interpreters performing silent reading, sight translation, and read aloud. (Novice interpreters data: Huang, 2011)…………………………………………64 Figure 4.1 The vertical perspective model (Macizo & Bajo, 2004)…………….84 Figure 4.2 The horizontal perspective model (Macizo & Bajo, 2004)………..86 Figure 4.3 Eye movement patterns of experienced interpreters……………….88 Figure 4.4 Eye movement patterns of novice interpreters……………………….89 Figure 4.5. Eye movement pattern of an experienced interpreter………………93. Figure 4.6. Eye movement pattern of a novice interpreter………………………..93. iv.

(6) Abstract. Student and novice interpreters have struggled to find an exact trait or processing model that allows more experienced interpreters to constantly deliver quality interpretations. However, although past research has attempted to observe and define objective criteria for an “expert interpreter” with various standards and from different perspectives, there has been no conclusive answer so far mainly due to the complexity of the interpreting task. This study observes the eye movements of experienced interpreters during sight translation and compares the results with eye movement data collected from novice interpreters, to see whether the former group shows any difference in terms of comprehension and processing procedures. The interpretations were also evaluated to confirm experienced interpreters do deliver better quality outputs.. 18 experienced interpreters took part in the experiment, and were required to complete silent reading, read aloud, and sight translation. All source texts were in Mandarin Chinese, and for the sight translation task, interpreted into English. Eye movements during all tasks were recorded, as well as the deliveries of sight translation. The results were later analyzed and compared with those of novices collected by Huang (2001).. Findings suggest that experienced interpreters begin to engage in other efforts besides comprehension during first pass reading when sight translating, as opposed to novices who seem to be only able to handle comprehension. However, even though they began to process input earlier, experienced v.

(7) interpreters still required rereading time. It is speculated that the reformulation stage during sight translation can be further divided into two levels, and while novices can only complete one (the basic) level of reformulation, experienced interpreters are able to finish both levels, therefore delivering better interpretations.. Keywords: sight translation, eye-tracking, expertise, expert, novice, interpretation quality, silent reading. vi.

(8) 摘要. 口譯學生及生手都希望能找出產出品質穩定的好譯文的關鍵因素。但到目前為止，雖已有不少研究透過不同方式角度探討，依舊無法找出熟手口譯員之所以有好表現的客觀因素。本研究將觀察熟手口譯員在視譯時的眼動軌跡，並將其與生手的眼動資料比較，探討兩者在理解處理過程中之異同。實驗的口譯產出亦會進行評分，以確認熟手口譯員的產出品質的確較佳。. 本實驗由十八名熟手口譯員為受試者，每位需完成閱讀、朗讀、及視譯。所有原文皆為中文，視譯則譯入英文。實驗中的眼動軌跡資料全程記錄，口譯產出亦錄音。收集的資料之後與黃致潔(2011)所收集的生手口譯員眼動資料做比較。. 結果顯示熟手口譯員在視譯過程中，首次閱讀時已經不單僅閱讀與理解，而同時還進行其他處理，生手口譯員則是僅能處理閱讀跟理解。再者，熟手口譯員雖在首次閱讀時已經開始處理，之後仍需要重覆閱讀處理的時間。由於熟手口譯員最終產出品質比生手口譯員佳，因而推論視譯過程中的重組階段(reformulation stage)可進一步分層為兩階層，而生手口譯員僅能完成第一（基本）階段，但熟手口譯員已可順利完成二階段，進而產出高品質譯文。. 關鍵詞：視譯、眼動法、專業、專家、生手、口譯品質、閱讀. vii.

(9) Chapter 1 Introduction. 1.1 Research Background. As interactions between different cultures and people increase in the modern world, so do the situations during which interpreters are required. When students train to become competent interpreters, they constantly observe the performance of their more experienced counterparts for distinctive strategies or traits that lead to better performances. However, so far there has only a little amount of research to identify specific strategies that set experienced interpreters apart. This gap between “experienced/expert interpreter” and “novice interpreter” is difficult to bridge, yet this is an important issue to be explored in terms of interpreter training. During interpreter training, instructors can only give suggestions and comments to the students based on the latters’ interpretation output. In other words, instructors analyze the delivery and offer advice on how to improve it in the future. Yet based on the author’s personal past experience as an interpreting student, this approach can be rather frustrating since the student, more often than not, feels that they 1.

(10) still have no clue exactly how to improve their performance. For the students, even with assistance and guidance from instructors, everything seems to happen so fast during interpreting that they are not sure how to properly conduct and finish the task. In order to solve this challenge, this study attempts to break down the sight translation process of experienced interpreters, and observe where and how experienced interpreters differ from novices, and is there actually a “model” that experienced interpreters follow but has not been further observed and studied yet. Hopefully the findings will offer student interpreters more substantial guidelines on how to improve their performance. Interpreting is indeed a complicated field to study, as it involves many components, and different modes of interpreting requires collaborations of different efforts (Gile, 1995). Sight translation is a mode of interpreting which input is in written form, and often employed in courtrooms or academic conferences (Weber, 1990). In classrooms, it is used as a preparatory stage for students before they begin simultaneous interpreting training (Agrifoglio, 2004), or to train students to read ahead of their notes for a smoother delivery (Weber, 1990). Although the pace and timing of intepretation is controlled by the interpreter, not the speaker, as opposed to simultaneous interpreting, this 2.

(11) does not imply sight translation is easier (Mikkelson, 1995; Agrifoglio, 2004). In fact, some studies actually state that visual influence is greater than audio influence (Agrifolio, 2004; Shreve, Lacruz and Angelone, 2010), thus sight translation requires more effort or the fluency of delivery may be compromised (Agrifolio, 2004). As mentioned before, observing the performance of more experienced interpreters is one of the means which interpreting students do in order to improve their own delivery. Experts have the ability to deliver accurate and complete performances fast and constantly, and form new solutions and strategies as required (Glaser, 1976). Many studies have pointed out that “experience” is crucial to become an expert (Moser-Mercer, 1997; Hoffmann, 1996; Dreyfus, 2004). Accumulated experience allows the expert to grasp the comprehensive situation faster (Glaser, 1976), and make quicker decisions on the action to take (Ericsson and Smith, 1993; Moser-Mercer B., 1997; Klein and Hoffman, 1993; Dreyfus, 2004). Some suggest that experts possess a mental “database” which stores and categorizes all past experience, so that the expert can draw solution references relatively faster when they encounter similar situations in the future (Klein & Hoffman, 1993). Despite different theories on how past experience improves efficiency, most studies agree that 3.

(12) experience is crucial to the making of an expert, and that more experience usually leads to shorter reaction time and consistent quality. However, past researches have shown much difficulty defining the level of “expertise” in the field of interpreting. Some suggested that students, upon completing their interpreting training, can say they are experts to a certain degree (Moser-Mercer, 1997). “Experience” is still stressed on in interpreting expertise studies (Glaser, 1976; Moser-Mercer, 1997), in fact, how long has the interpreter been in practice is often used to evaluate the level of expertise (AIIC, n.d.). Experience is also considered a major factor that causes limitation of skills and ability of novices (Lambert, 2004). More efficient strategy decision is also mentioned in past researches (Kuenzli & Moser-Mercer, 1995), and there are also studies that suggest different strategies implemented by novice and experienced interpreters, respectively; thus leading to different qualities of output (Moser-Mercer, 1997; Mayer, 1992; Moser-Mercer, Frauenfelder, Casado, & Künzli, 2000). These past studies present various differences between novice and experienced interpreters, yet they still cannot offer a clear suggestion on exactly what do the experienced ones do that lead to better performances. If this exact strategy or difference can be pin-pointed, then in addition to offering novices more concrete 4.

(13) guidelines, that may also be considered as a criteria to objectively define an “interpreting expert”. So far most interpreting researches still focus on studying the final output, and provide rather few observations on the entire process that produces the delivery. This direction of research only sheds very limited light on how interpreting students and novices can improve their skills and produce better interpretations. Unfortunately, this is a rather common challenge faced by interpreting instructors and students in the classroom (Huang, 2011). Recently, the evolvement of new equipment, such as the eye tracker, provides researchers new ways to observe cognitive behaviors such as reading (Rayner, 2009; Duchowski, 2002; Richardson & Spivey, 2004). Some researchers are also attempting to reveal the secrets of experienced interpreters by breaking down and studying the interpreting process. Related studies are still at the budding stage, nevertheless they have begun to provide some basic observations on the reading behavior of novice interpreters during sight translation (Huang, 2011). Huang (2011) studied the different eye movement indices of novice interpreters when they were performing sight translation. The indices revealed the actions of the participants during different stages of the entire process. This study will extend her research by replacing the 5.

(14) participants with experienced interpreters instead of novices, and compare the eye movement results of the two groups afterwards. The reading habits and methods of experienced interpreters would also be observed through eye movement indices, and see whether there are any differences to those of novices. Since the input for sight translation is in written form, eye movements of the participant can be recorded by an eye tracker for further analysis. Studies in the past have established the relevance between various eye movement indices and comprehension (Rayner, 2009; Just & Carpenter, 1980; Richardson and Spivey, 2004); other studies further compared the results of different tasks, such as silent reading and read aloud, with sight translation in attempt to observe the level of effort required to complete them (Shreve, Lacruz, & Angelone, 2010; Richardson, Dale, & Spivey, 2007; Huang, 2011). Various eye movement indices are further observed and compared to yield more observations on comprehension behaviors (Rayner, 2009; Just & Carpenter, 1980; Richardson and Spivey, 2004). Many studies have compared the performance of bilinguals or novices interpreters with that of experienced, or professional, interpreters (Moser-Mercer, Frauenfelder, Casado, & Künzli, 2000). However, in the past, the implementation of eye trackers in the T&I 6.

(15) field have mostly focused on translation (Huang, 2011); studies on interpreting, on the other hand, are relatively limited. Furthermore, most of the existing eye movement studies on T&I are mainly conducted on European languages (Moser-Mercer, Frauenfelder, Casado, & Künzli, 2000), and studies on Mandarin Chinese–English interpreting are even few and far between (Huang, 2011). As a result, this study will extend on existing study results and see whether experienced interpreters indeed, as experts in other fields, deliver better performances than novices because they have more experience. The study will also compare the comprehension processes of novice and experienced interpreters whose working languages are Mandarin Chinese and English, and observe if there are any differences between the two groups, and if there are, would it be the special strategy or trait that novices should be aiming for to acquire in order to improve their final delivery.. 1.2 Research Questions. This research attempts to observe the sight translation comprehension process of experienced interpreters and study the differences between theirs and the process of novices, in hope to recognize traits that make a delivery by 7.

(16) experienced interpreters better. In Huang’s (2011) study, she had already collected eye movement data from novice interpreters, and also provided some insight on the cognitive behavior of novices during sight translation. In order to be able to compare the collected data from both researches, this study will also conduct an eye movement experiment and collect eye movement records of experienced interpreters. Participants will be asked to complete three tasks: silent reading, read aloud, and sight translation. Eye movements and vocal output would be recorded for later studies. Output quality of both groups (novice and experienced interpreters) would also be evaluated in terms of accuracy and fluency. This study will observe and study the following areas: 1) Whether experienced interpreters actually in fact require less time in certain eye movement indices, which would support the hypothesis that experienced interpreters are more efficient; and 2) Observe if experienced interpreters really do employ specific strategies during sight translation which are not performed by novices.. 8.

(17) Chapter 2 Literature Review. 2.1. Expertise. 2.1.1 Definitions and criteria of experts. When people mention “an expert”, the image is usually an individual who is “very good at something”, and are able to provide performances with consistent good quality (Ericsson, 2000). Glaser (1976) offered a more detailed description of the general expectations of experts. He pointed out that when one has achieved expert status, they should be able to constantly deliver an accurate and complete performance relatively fast, integrate individual skills and choices, have a more comprehensive perception of the task, and form new solutions and strategies as required. Efficiency, flexibility, and consistency are traits that experts should possess. And many studies point out that experience is crucial to obtaining these traits. For Moser-Mercer (1997), an expert is, “someone who has attained a high level of performance in a given domain as a result of years of experience”. Hoffmann (1996) also suggests that time alone does not make an expert; rather, the key is to accumulate 9.

(18) experience and practice. As experts accumulate experience, they become more familiar and comfortable with their skillsets. They can manage, adjust, even develop new strategies on the spot as the tasks and situations demand, which allows experts to be able to response and/or react faster than novices when faced with the same task, as Ericsson and Smith (1993) suggested; and are more likely to achieve similar levels of quality each time. This “automation” strategy is also a means of showing one has mastered the skills of the particular field, and allows experts to utilize production capacities to full efficiency and shorten reaction time; as opposed to novices who have to consciously manage all strategies they employ, resulting in more required time and effort and probably a not very satisfactory outcome (Moser-Mercer, 1997). Past researchers have found experts or experienced individuals in certain fields, such as chess masters, would visualize the task in “chunks” as a more efficient way of understanding the comprehensive situation (de Groot, 1965, 1966; Chase & Simon, 1974). Klein and Hoffman (1993) also mentioned that experts are able to tell when things are amiss faster, visualize the entire process and anticipate the possible outcomes because of their experience. This skill increases the efficiency of experts, as it allows them to figure out the most 10.

(19) appropriate and efficient solution and strategy for the task as early as possible, decreasing the possibility of employing a wrong solution and failing the task. Similar judgment skills also assist in the experts’ problem solving and decision making stages (Klein & Hoffman, 1993). In short, experts have reached a point where they do not need to spend extra effort to consciously access their expertise (Anderson, 1995; Dreyfus, 2004), therefore are faster and more flexible when finding a solution to complete the task at hand. So what are the exact criteria to define an expert? Traditionally, the emphases are placed on different strategies and/or larger knowledge bases that experts possess that allow them to deliver better performances (Klein & Hoffman, 1993). However, so far there has been no consensus on how to scientifically or objectively evaluate or qualify one as an expert. As Ericsson (2000) points out, there is the possibility that the individual’s reputation and level of training is mistakenly used to evaluate their expertise, and these social perspectives of expertise are usually not consistent with the individual’s performance. He argued that experts should be able to repeatedly deliver performances of similar quality anytime; therefore researchers can collect a set of expertise qualities and evaluate expertise in a scientific way in laboratories (Ericsson, 2000). Ericsson, Prietula, and Cokely (2007) suggested three indicators to 11.

(20) evaluate real expertise: consistent superior performance, concrete results, and “can be replicated and measured in the lab”. Some may question that in certain areas, such as creative professions, measuring expertise would run into many challenges as it is impossible to recreate the performances in a lab, let alone scientifically measure the results. Nevertheless, according to Ericsson, Prietula, and Cokely (2007), modified testing methods for these areas, such as art and writing, still exist, and results accurately reflect the experts’ technical proficiencies. Therefore, regardless of area and profession, it is still possible to identify at least the technical strategies and traits of experts.. 2.1.2. Developing expertise. Some people may argue that experts are born with special talents, or “gifted”. Experts possess some kind of innate ability that allows a particular individual to perform better, to achieve a certain goal faster and easier. On the contrary, those born without talents have to work harder, and may even never be able to reach the level of expertise the former group have. Therefore for those who are not born “with the talent”, any form of expertise training or development would at most yield limited results. However, according to Ericsson, Prietula 12.

(21) and Cokely (2007), experts are made (trained), not born; in fact, it will take a person at least 10 years (or 10,000 hours) to become an expert. In their article, they also mentioned the findings of Bloom (1985). Bloom’s study showed that individuals who perform exceptionally well “had practiced intensively, had studied with devoted teachers, and had been supported enthusiastically by their families throughout their developing years”. Ericsson (2000) also remarked that so far there have not been any set standards for evaluating the effects of innate talents on adult achievement. If innate talents do bring certain advantages for individuals, there is no supporting evidence that others cannot compensate the gap with extensive training later on in life. The only exceptions would probably be body size and height, especially for athletes. But overall, given enough motivation and training, individuals who had already accumulated substantial experience in the area usually can accomplish different degrees of improvements later in life. In order to understand the progress of individuals on their journey to become experts, Dreyfus (2004) purposed a five-level model of expertise development, starting with Novice, to Advanced beginner, Competence, Proficiency, and finally, Expertise. The Dreyfus model has been used to study interpreter training (汝明麗，2010). According to the study done by 汝明麗 (2010), upon 13.

(22) finishing two years of interpreting training, student interpreters in Taiwan should have a Competence level skillset. However, the implementation of the Dreyfus model in interpreting studies is still at its beginning stage (汝明麗， 2010). In the Dreyfus model, both novices and advanced beginners are placed in controlled environments where they are given simple basic rules to follow. Individuals at the Novice level begin by simply follow basic rules given by the instructor, regardless of the environment and situation. Gradually, the novices begin to note different situations and start to integrate related context and information for a better performance. This is when they enter the Advanced beginner level. At this level, advanced beginners have more experience than novices, and more understanding of related context; yet the learning at the level are still based on examples and instructions, and distanced from actual environments and scenarios. Once advanced beginners enter the Competence level, they begin to make their own choices. Students at this level learn how to pick out relevant and important aspects and make decisions during tasks. Results of the tasks are now controlled by the student’s own choices, which, understandably, cause emotional stress for the student. Interestingly, if handled well, stress may actually assist the student to improve, as it pushes 14.

(23) them to improve their performance. Starting from the Proficient stage, the student is able to conceive goals (as in what they need to accomplish) and form various strategies to react. However, proficient individuals lack sufficient experience to anticipate possible outcomes of different strategies; therefore still require time and effort to decide which method of solution to use. As they do not have enough experience to assist their decision, proficient individuals would fall back on their basic trainings. Here experience becomes the biggest difference between Proficient and Expertise levels: proficient performers, who have relatively lesser experience, still need to fall back on basic rules and training to make decisions, consequently requiring more time and effort to complete the task. On the other hand, experts are able to form intuitive responses immediately by leveraging their past experiences, and decide on better and more effective strategies. Dreyfus (2004) pointed out that after the expert encounters a new situation, their brain will later categorize the final solution, and form a database-like storage for the expert to reuse or refer to in the future should their encounter similar problems. This echoes the findings of Klein and Hoffman (1993) mentioned earlier which suggested that since experts are more experienced, they are able to quickly predict possible outcomes of 15.

(24) different strategies; therefore more likely to choose and apply the most efficient solution within the shortest period of time. Again experience is the crucial factor that distinguishes experts from other levels of expertise. However, although experience is an important factor that allows experts to make decisions faster and more efficiently, Ericsson, Prietula and Cokely (2007) pointed out that some studies have shown that without continuous training, expertise actually declines with experience. The training here does not indicate aimless repetitive practice/exercises, as Klein and Hoffman (1993) points out; but rather a kind of deliberate practice, which is defined as “practice that focuses on tasks beyond your current level of competence and comfort” (Ericsson, Prietula, & Cokely, 2007). It includes two aspects: stepping out of one’s comfort zone and work on something new and unfamiliar, and to continuously improve on the skills one already possess. Deliberate practice also makes great difference in performance. As Ericsson, Prietula and Cokely (2007) found out in their study, senior experts who neglect to practice deliberately on a regular basis may be more prone to dealing with events automatically, or even relying on intuition; therefore are more likely to run into problems when faced with atypical situations they are not familiar with. 16.

(25) Dreyfus (2004) did not address the issue of deliberate practice in his model. Expertise in his model is defined by the ability to make immediate, intuitive responses according to the situation (Dreyfus & Dreyfus, 2005; Dreyfus, 2004). When faced with atypical situations, experts should be able to compensate with their vast experience “data-pool”, and still produce quality performances (Dreyfus, 2004). Past literature and studies offer general observations on experts in various fields, such as expertise and quality performance are the results of experience accumulation (Ericsson & Smith, 1993; Moser-Mercer, 1997; Hoffman, 1996), and experts employ “chunking” and other strategies during tasks (Klein & Hoffman, 1993). One phenomenon often used to decide whether an individual is an expert or not, is their ability to constantly deliver a quality performance (Glaser, 1976; Ericsson, Prietula, & Cokely, 2007). Although some may say it is difficult to objectively grade performances in certains fields, the creative arts, for example, Ericsson, Prietula and Cokely, (2007) point out there are still methods to grade leves of technical aspects and skills. Experts, according to Ericsson, Prietula and Cokely (2007), are deliberately trained and developed, not just individuals natually born with innate talent. In fact, deliberate practise is very crucial to the making of an expert (Bloom, 1985; Ericsson, 17.

(26) 2000; Hoffman, 1996). Since currently there is no other model established to distinguish different levels of expertise in the field of interpreting, the Dreyfus model is used in related interpreting studies as guidelines for grouping interpreters at various skill levels (汝明麗，2010).. 2.2. Interpreting. The history of interpreting may be traced back to a time when two people speaking different languages wanted to understand what the other person was talking about, so they sought out the help of a person who happened to speak and understand both languages. Methods of interpreting have evolved and changed with time and the advance of technology. Interpreting is generally divided into two major categories: consecutive interpreting and simultaneous interpreting (Agrifoglio, 2004). During consecutive interpreting, or CI, speaker and interpreter take turns to speak/interpret. The interpreter delivers the interpretation after the speaker has spoken a part of the speech (usually no more than three minutes). In simultaneous interpreting (SI), the interpreter listens to the original speaker through headphones, and interprets the context into the target language simultaneously. The delivery is spoken into a 18.

(27) microphone and delivered to the audience’s headphone. Sight translation (ST), during which the input is visual instead of audio, is often considered a type of SI; however, since the input during ST is readily available on paper all the time, the interpreter is able to follow their own pacing, rather than being controlled by the speaker (Agrifoglio, 2004).. 2.2.1. Sight translation. Sight translation (ST) is a task that requires the interpreter to read a text and simultaneously interpret it into another language. This task may be employed in judicial situations, such as courtrooms, or scientific and technical academic conferences, where the speaker may opt to directly read from their papers (Weber, 1990). In many translation and interpreting training institutes, sight translation is considered as a preparatory stage for simultaneous interpreting (Agrifoglio, 2004), and also used as a tool to train interpreting students to read ahead of their notes and deliver a smoother output. Prior to assignments, interpreters may also sight translate related text and articles as background knowledge preparation, and to familiarize themselves with the jargon and technical terms of the area (Weber, 1990). 19.

(28) During ST, the source text is always visually present, and the interpreter can process the input according to their own speed; as opposed to SI, during which the input is audio, and flow and speed is dictated by the speaker. However, ST should not be considered as easier to accomplish than SI on the basis that it has no time limitations. Many researches have pointed out that ST is as difficult as SI, as interpreters mentally process language during both tasks similarly (Mikkelson, 1995; Agrifoglio, 2004).To further illustrate the complexity of cognitive loading during various interpreting tasks, Gile (1995) broke down the components and established effort models to describe how cognitive efforts are distributed during different interpreting modes. According to the model, a SI task is composed of Listening and Analysis Effort (L), Memory Effort (M), Production Effort(P), and Coordination Effort (C), resulting in a model as follows: SI = L + M + P + C Spontaneously listening, comprehending and speaking is one of the main characteristics of SI, which also implies the importance of coordination during the task. Furthermore, since the interpreter is required to deliver the interpretation as the speaker is speaking, the interpreter must not only go through all efforts at the same time, but also adjust their own processing pace 20.

(29) according to the speaker’s speed. Students or novices of interpreting often find this task daunting, as the pace and speed of the input is not controlled by the interpreter themselves. In fact, studies have shown even experienced professionals may make mistakes during SI (Gile, 1997). Agrifolio (2004) also concluded that during SI and CI, memory saturation and note-taking problems could interfere with the final interpretation output, causing failures. For sight translation, Gile (1995) claimed that during this task, Listening and Analysis Effort is transformed into a Reading Effort (R), and Production Effort (P) is still present. But Memory Effort is not required during this task, as the source text is always present on paper and interpreters do not have to rely on short-term memory to memorize the beginning of sentences, or previous information input in order to produce a complete and coherent sentence. The model is displayed as follows: ST = R + P The absence of Memory Effort in Gile’s (1995) is debatable. Gile assumed that since the source text was always available, there would be no need for extra short-term memory. He did acknowledge that syntax differences between languages, and longer sentences with embedded clauses require more time and effort; however, the interpreter deals with these challenges by rereading 21.

(30) more Translation Units, not by employing any memory-related strategies. Actually, researchers disagree on whether the always-present visual source text is more of an interference which could compromise the fluency of delivery, instead of a visual aid. Lambert (2004) argued that students perform better during sight translation and simultaneous interpreting with text than during free simultaneous interpreting, mainly because of having the text which serves as visual assistance present during the processes (Lambert, 2004). Also, since the text is a visual input, when combined with audio inputs, it does not cause much interference to the interpreter (Shaffer, 1975; Viezzi, 1989; Lambert, 2004). On the other hand, some argue that visual interference could be stronger than that from audio, as mentioned by Agrifolio (2004) and Shreve, Lacruz and Angelone (2010). Since the source text remains available throughout the ST process, the attention and gaze of the interpreter may involuntarily be drawn back to the source text, causing diversions and increasing the cognitive load (Agrifoglio, 2004). Furthermore, Mikkelson (1995) also noted that with the text always present, interpreters find it more difficult to focus on the meaning of the message, instead of the words; therefore the constant presence of the written text may prove to be a drawback instead of assistance. 22.

(31) It should also be noted that Gile did not address the Coordination Effort when describing the model of sight translation; yet it can be assumed that the effort is still present so that the interpreter can smoothly employ both Reading and Production efforts. Agrifoglio (2004) also pointed out that one of the main difficulties of sight translation is the interpreter requires significant coordination to ensure a fluent, coherent delivery, especially when the two languages have different grammatical structures. The act of doing many things simultaneously is not a natural activity for humans, normally people can only concentrate on one single task (Lambert, 2004). Two tasks may be manageable if they are related to the same higher-order activity, as the attention switches rapidly between the different tasks, or at least one of the tasks can be carried out automatically (Lambert, 2004). However, Lambert (2004) had also noted that with proper training and practise, it is possible to learn how to divert attention, or “multi-task”. The participating novices and experienced interpreters in Huang’s (2011) study and this research were all trained in sight translation skills, therefore it is assumed that they should be able to multi-task and coordinate all efforts in order to complete the task. Effort distribution during sight translation can also be affected by different methods of input. For example, the different nature of written and oral source 23.

(32) texts may cause more difficulty in sight translation since written texts are often more complex than oral source texts in terms of syntax, rhetorical structure, and more complicated grammar usages. Written texts may also contain features that otherwise would be omitted in oral source texts. Therefore, some argue that the demands of comprehension during sight translation are much higher than that of other modes of interpreting (Shreve, Lacruz, & Angelone, 2010). In fact, interpreters performing sight translation have to exercise more effort on reading and processing the visual input to a point where the fluency of the interpretation may be compromised (Agrifoglio, 2004). Dillinger also pointed out that if more effort was spent on syntactic analysis and proposition generation when reading, the accuracy of the performance is likely to suffer/decrease (Dillinger, 1990). The aforementioned studies present sight translation as a task similar to SI but the pacing is controlled by the interpreter, not the speaker. The task still requires more than one effort, as seen in Gile’s (1995) model. And the ability to optimize effort distribution is crucial during this task since the fluency and accuracy of interpretations could be compromised due to extra effort spent on reading and comprehending the input, as suggested by both Agrifoglio (2004) and Dillinger (1990). 24.

(33) 2.3 Expertise in Interpreting. In order to figure out objective criteria for an expert interpreter, previous studies attempted to identify concrete strategies or traits found among experienced interpreters which allow them to continuously deliver quality performances, and consequently use the traits as criteria as the definition of expert interpreter. Currently, amount of experience is often used to evaluate an interpreter’s skills and competency (Ericsson, 2000; AIIC, n.d.). In terms of delivery quality, fluency and accuracy are two of the basic requirements and commonly used during output evaluation (藍順德、劉敏華、張嘉倩、陳子瑋、林慶隆、吳紹銓，2007). However, it should be noted that above mentioned factors are commonly used, but not definite. Some researches propose skill acquisition procedures (Anderson, 1995), and some further combine skill level models with interpreting pedagogy (汝明麗，2010), in an attempt to observe “expert traits” through the learning and maturing process of interpreting students. Other areas discussed in order to find the difference between novice and. experienced. interpreters. include. input. information. processing. (Moser-Mercer, 1997), knowledge structure (Kuenzli & Moser-Mercer, 1995), 25.

(34) strategy employment (Moser-Mercer, 1997; Mayer, 1992; Lambert, 2004; Moser-Mercer, Frauenfelder, Casado, & Künzli, 2000). According to Moser-Mercer (1997), an interpreting and translation student may call themselves an expert of certain degree once they had finished their training and received diplomas. However, years of field experience are still required for the individual to become an established professional. As mentioned before, it will take at least ten years for an individual to become an expert on average; yet in the field of interpreting, so far there is no consensus on the exact criteria of an expert interpreter, let alone the amount of time required to reach expert level. Evaluating expertise in interpreting can be quite challenging. Unless an interpreter mistranslates the message or makes obvious language mistakes, it is difficult to objectively judge an interpretation as “good” or “bad”, and currently the judging and scoring of interpretation deliveries mostly rely on the professionalism of seasoned professional interpreters (藍順德、劉敏華、張嘉倩、陳子瑋、林慶隆、吳紹銓，2007). This problem is similar in certain areas, such as creative professions, where measuring expertise would arouse many challenges. For example, the nature of some expertise makes it impossible to recreate performances in an experimental environment, let alone scientifically measure the results. 26.

(35) Nevertheless, Ericsson, Prietula, and Cokely (2007) stated that modified testing methods still exist, and results are in line with especially the experts’ technical proficiency. For most interpreting examinations and evaluations, accuracy and fluency are always among the grading criteria (藍順德、劉敏華、張嘉倩、陳子瑋、林慶隆、吳紹銓，2007). It shows that there two are the technical aspects of interpreting, and can be measured. In this study, the interpreting deliveries of both novices (from Huang’s research) and experienced interpreters would be graded in terms of accuracy and fluency in order to determine the quality of output. Besides the two criteria mentioned above, currently there have not been any other universal criteria for evaluating different levels of interpreting expertise, although studies attempting to find some criteria have been continuously conducted. Kurz (2001) studied interpreting delivery quality from a user’s point of view. The research revealed that although there has been plenty of data collected from previous studies, they still lacked coordination and interaction (Kurz, 2001). Also where users are judges of delivery quality are concerned, the various types of demands and expectations among these interpretation service users can easily affect one’s concept of “a good interpretation” (Kahane, 2000). Bühler (1986) tried to avoid this problem by 27.

(36) conducting his study among professional conference interpreters. However, a “good interpreter” may still have completely different profiles in the minds of other professional interpreters as they would in conference attendees (Cartellieri, 1983). However, how long the interpreter has been active, or “years of practise”, seems to be a common way of judging the abilities and skills of an interpreter: the longer this individual has been working, the better they ought to be. The International Association of Interpreters (AIIC), which is the only global association of professional conference interpreters, requires those who wish to apply for membership to have “at least 150 days of work overall according to AIIC's rules and regulations” (AIIC, n.d.); yet the reason for this number of days is not explained. Nevertheless, the 150-day criterion is included as one of the requirements for experienced interpreters participating in this study’s experiment. In terms of expertise development, Anderson (1995) suggested that interpreters have to pass through three stages to acquire necessary skills: a cognitive stage, during which the novice is taught the basic skills and facts of interpreting; an associative stage where the novice learns and experiments with the acquired skillset, and finally reaches the autonomous stage where the interpreter does not need to consciously choose and manage the various skills 28.

(37) to deliver good quality output, consequently leads to faster efficiency and accuracy. These three stages can be compared with Dreyfus’ (2004) five-stage model: the cognitive stage of Anderson (1995) would be similar to Dreyfus’ (2004) Novice and Advanced Beginner stage, the associative stage echoes Dreyfus’ (2004) Competent and Proficient stage, and the final autonomous stage agrees with Dreyfus’ (2004) Expert stage, in which experts are able to make immediate decisions for solutions to the task at hand. Both models indicate similar learning processes for an individual to become an expert, providing check points for interpreting instructors and students to refer to. In Taiwan, interpreting student who have completed two years of post graduate training in T&I schools are expected to reach at least Competent level skills on the Dreyfus model (汝明麗，2010). Researches comparing the performances of interpreting novices or bilinguals and professionals have been carried out in the past (Dillinger, 1990), yet it is still extremely difficult to isolate a specific skill or aspect that distinguishes an expert from a novice, let alone yield any results to benefit interpreting pedagogy (Moser-Mercer, 1997). Some early researches suggest that differences between expert and novice can be seen in reasoning knowledge structures and how information is processed at different levels, such as the 29.

(38) knowledge base and level of strategies (Moser-Mercer, 1997). In terms of knowledge base, experts and novices seem to organize their bases differently (Kuenzli & Moser-Mercer, 1995). Experts display better organization and connections, consequently shortening reaction time, which is consistent with Dreyfus’ (2004) model. Yet if the expert ventures into unfamiliar subjects, there is still a possibility the expert would deliver a novice-like performance. Experts are also more likely to able to adjust language register according to the context, and to draw meaning from the context and carry on with the interpretation, as opposed to being stuck on one unfamiliar word. Novices, on the other hand, have a higher possibility to be only able to produce individual sentences with no discourse links that fail to deliver the message clearly, and are more likely to be stuck or unable to carry on at one point during the task because of one unknown word (Moser-Mercer, 1997). In other words, experts use a more global approach, whereas novices may turn to “low-level – microcontextual” methods (Moser-Mercer, 1997; Mayer, 1992). During sight translation, professional interpreters are able to shift easily between the written and oral forms and deliver a smooth and overall more satisfactory interpretation; while beginners, or novices, are more constricted by the original written text, and have a higher possibility of misreading the original 30.

(39) text (Lambert, 2004). In addition to the later processing strategies, interpreters also engage “comprehension strategies” during the initial stage. After the initial comprehension stage, planning strategies are then employed for production. This stage would affect not only the output language per se, but also register of speech. It should be noted that research indicates that experts would employ planning strategies in the comprehension stage, instead of waiting until the production stage (Moser-Mercer, Frauenfelder, Casado, & Künzli, 2000), yet it is not clear whether novices make any attempt towards employing this stage or not. Past researches point out that experience, training and deliberate practice are crucial to expertise development (Dillinger, 1990), and according to Ericsson, Prietula and Cokely, experts are trained, not born. In other words, those who excel in respective fields achieved their success completely through hard work. However, Dillinger focused on the aspect “experience”, and offered another theory after conducting an experiment that involved eight experienced professional interpreters and eight bilinguals who had never done simultaneous interpreting. Through his experiment, he found the only difference due to experience were that experienced interpreters were able to 31.

(40) extract the information more quickly than the other group, and were more selective in the information they processed. Therefore, he concluded, the skill to translate and interpret seems to come naturally with bilingualism. Experienced (trained) interpreters do not acquire special skills, they just have a more flexible way of comprehending and processing information. Yet it should be noted that it is unclear whether the slightly better performance (“17% improvement of accuracy”) of experienced interpreters was due to training or experience. Macizo and Bajo studied if there were any differences between professional translators and untrained bilinguals when reading for repetition and reading for translation (Macizo & Bajo, 2006). Results showed that two groups dealt with the two tasks in a similar way, which supports Dillinger’s suggestion that professional interpreters comprehend input in a similar way as bilinguals. However, Macizo and Bajo further pointed out that these results cannot lead to. the. conclusion. that. the. skills. and. abilities. of. experienced. translators/interpreters are completely the same as untrained bilinguals, as they had observed a difference in overall reaction time between the two groups in their experiment (Macizo & Bajo, 2006). The studies mentioned above seem to suggest that experts react and employ 32.

(41) strategies faster due to their experience and accumulated “database”, and also are able to distribute their efforts more efficiently during a task. However, experiments that were designed to observe time durations mainly focused on total time required to complete the task, which cannot show whether experienced interpreters began any sort of strategy earlier than novices. In her research, Huang (2011) observed the eye movements of novices performing sight translation. Different eye movement indexes recorded during the experiment presented actions of the participants during different stages of the entire process. This study will extend her research by using experienced interpreters as participants instead of novices and comparing the eye movement results of the two groups afterwards. It should be noted that even though we lack clear standards on how to qualify an interpreting “expert”, the word is still used to describe an interpreter who is good at their job. It should be noted that the “experts” may not be actually officially qualified, and the word is just used to distinguish the relative skill levels of different individuals. Nevertheless, in this research the term “interpreting expert” will be avoided since the definition of such an individual is undetermined. Instead, the participants of this experiment, who have all worked in the interpreting field longer than the novices, will be identified as 33.

(42) “experienced interpreters”. However, since they still possess relatively more experience, the experienced interpreters in this study are expected to show the above mentioned traits of experts, including faster and more flexible reactions. When compared to the Dreyfus model, the novices in this research are likely to be at the Competence or Proficiency stage ( 汝明麗， 2010), and the experienced interpreters are expected to reach Expertise stage. With the support of eye movement data, this study will observe if experienced interpreters indeed show different strategies, for example, begin to process input earlier than novices, during sight translation. Delivery quality of the experienced interpreters will also be evaluated according to fluency and accuracy, two most commonly used criteria, to further support the contention that their strategies do lead to better output.. 2.4. Eye movements and comprehension. Initially, eye trackers were mainly used in areas including reading and other information processing activities as a type of diagnostic study (Rayner, 2009; Duchowski, 2002). In recent years, eye trackers have become quite popular in cognitive research, especially in areas such as reading, scene perception, and 34.

(43) visual search (Rayner, 2009; Richardson & Spivey, 2004), memory, imagery, complex tasks, and various interactive applications for the disabled (Richardson & Spivey, 2004). Correlations between eye movements and cognitive activity have also been gradually established, for example, the eye-mind assumption indicates that the gaze duration spent on the word is also the time required to process the input (Just & Carpenter, 1980). This type of observation is often used in cognitive research because it displays the end products of human behaviors, provides certain insights to the task processes, and most importantly, decreases interruption to the experiment (Richardson, Dale, & Spivey, 2007). More sophisticated technology allows eye trackers to provide real-time feedback, and the researcher is able to collect more accurate data than observing button-press reaction time used in traditional cognitive related experiments (Richardson & Spivey, 2004). When implemented in interpreting studies, eye-movement data provide not only final results, but also offer a glimpse to the cognitive behavior of the participant during the interpreting task (Richardson, Dale, & Spivey, 2007). Eye-movement index results may offer clues on participant behavior, especially cognitive behavior. In the eye-mind assumption proposed by Just and Carpenter (1980), the gaze duration spent on a word is the time required 35.

(44) to process it. The fixation will last as long as the participant is processing the input, and there is no lag of time between fixation and processing (Just & Carpenter, 1976, 1980). Since this study focuses on time duration, eye-movement indices observed include first fixation duration (FFD), single fixation duration (SFD), gaze duration (GD), rereading time (RRT) and total viewing time (TVT). The definitions of the indices are as follows (Rayner, 2009):  First fixation duration (FFD): time duration of the first fixation in region of interest.  Single fixation duration (SFD): time duration of the fixation that is only fixated once on a region of interest during first pass.  Gaze duration (GD): the sum of all fixation durations during the first pass.  Rereading time (RRT): the sum of all fixation durations in region of interest after first pass. Since fixations imply the participant is comprehending the word (within region of interest), rereading time indicates the need to further process or comprehend input information.  Total viewing time (TVT): total sum of all fixation durations in region 36.

(45) of interest, including fixations after regression. According to past eye movement studies, a reader fixates their gaze on the target word or term when they are processing the input, and the fixation will remain on the target, or region of interest (ROI) for the entire period of comprehension or process (Rayner, 2009; Just & Carpenter, 1980; Richardson and Spivey, 2004). The average duration of eye fixations when reading English lasts about 200-250 milliseconds (Pollatsek, Rayner, & Collins, 1984; Rayner, 2009), and about 220-230 milliseconds when reading Mandarin Chinese written in horizontal format (楊馨慧、陳子瑋、江健新、蔡介立、黃致潔，2011). Readers fixate longer on words they find difficult or unfamiliar, so as to lessen the memory load (Just & Carpenter, 1980), and regression percentage increases with the difficulty of the text. If a sentence is more complex, and the reader needs more effort to process it, the eyes will regress and fixate on words that had been read before (Richardson & Spivey, 2004; Rayner, 2009; Kemper, Crow & Kemtes, 2004). The concept is also mentioned by Kennedy and Muray (1987, 1988) who pointed out that when rereading, readers find and fixate on the words they did not understand accurately and process them again. Kemper, Crow and Kemtes (2004) also implied that more regressions and longer regression periods indicate difficulty 37.

(46) processing input, for example, sentence parsing. For experienced interpreters, it is expected that they are able to comprehend and process the input faster. Given that eye fixation stays on the word that the interpreter is processing (Just & Carpenter, 1980), it can be assumed that experienced interpreters should have a shorter total viewing duration. It should be noted that fixation durations may also vary due to other limitations of the task, and are not all necessarily affected by comprehension and processing reasons. For example, as Richardson and Spivey (2004) pointed out, fixation durations of silent reading and reading aloud are different in that the duration of reading aloud is longer than that of silent reading. This phenomenon was also found in Rayner’s (2009) research. He suggested that actually the comprehension stage had already been completed, yet the vocal output is not fast enough, therefore the average fixation for reading aloud would be longer than silent reading.. 2.5. Eye Tracking in Interpreting Studies. Studies on reading behavior often employ eye trackers. Just and Carpenter (1980) studied the correlation between outward eye movements and internal 38.

(47) cognitive action. They concluded that fixation durations were longer in parts that demanded heavier processing workloads, such as infrequent source words; and gaze duration reflects the time needed to process input (Just & Carpenter, 1980). Yet the utilization and studies of the eye tracker in the area of interpreting are relatively few, and eye movement researches focused on sight translation are even scarcer. McDonald and Carpenter (1981) observed the. eye. fixations. when. translating. ambiguous. idioms,. and. how. mistranslations were detected and corrected. The study showed that after the interpreter read and comprehended the source idiom, they proceeded to separate the input into meaningful units, or “chunks”, and reformulate it. This is followed by misinterpretation corrections (grammar, lexicon) and finally, the interpreter delivers the outcome (McDonald & Carpenter, 1981). The “chunking” procedure is consistent with the immediacy strategy proposed by Just and Carpenter (1980), and the horizontal perspective by Macizo and Bajo (2004). All of the aforementioned concepts suggest that when an interpreter is sight translating, they “chunk”, or break down the input into smaller meaning units and interpret the chunk simultaneously, instead of trying to comprehend the entire sentence or message. Eye fixations from this research reveal three different categories: the initial reading pass, which includes all 39.

(48) forward fixations before regression; the translation pass, which is all rereading fixations within the same phrase before fixating on further material; and the error recovery pass, which includes re-fixations that returned to previous phrases after further text had been read. When the interpreter detects an error, the eyes regress, and they may re-chunk previous source text for another version of delivery. In 2010, Shreve, Lacruz and Angelone conducted a research on the influence of visual input during interpreting. Participants were asked to sight translate four paragraphs from Spanish to English while their eye movements were tracked by a Tobii 1750 eye tracking system. The four paragraphs are manipulated so that two of them will contain complex syntax, while the other two are simpler. Prior to the actual sight translating task, participants read a “practice text” in Spanish. Translating sessions were not timed. The researcher observed eye movements such as fixations and regressions, as well as verbal data including pauses, repetition and fillers to study the challenges encountered by the participant, also the possible visual influence as proposed before by Gile and Agrifoglio (Shreve, Lacruz, & Angelone, 2010). Results show that when reading, there were fewer and shorter fixations, less regressions, and participants on average needed less time to complete the task, 40.

(49) compared with sight translation. It should be noted that in the former mentioned research, the definitions of the measuring standards were not clearly stated, thus may cause confusion due to the inability to accurately replicate the experiment. Also, the exact sequence of the task material would possibly cause a “learning effect”, which would affect the results of the experiment. If the participants go through the tasks in the exact same orders, it is possible that they, consciously or unconsciously, pick up skills as they finish each task during the entire progress, thus leading to a better performance in subsequent tasks, and consequently affect the accuracy of the results. Huang (2011) conducted another experiment whose procedure is somewhat similar to the one done by Shreve, Lacruz and Angelone. In Huang’s experiment, while sitting in front of an eye tracker, participants were asked to complete three tasks: silent reading, read aloud, and sight translation, and answer (true and false) comprehension questions after the tasks. Source texts are all in Mandarin Chinese, and will be interpreted into English during sight translation. The sequence of the three tasks is randomized, so as to avoid aforementioned “learning effect”. All eye movements were recorded and later compared. All participants were interpreting students who had one year of 41.

(50) sight translation training, with Mandarin Chinese as their native language, and English as working language. The three tasks were designed to let the researcher isolate and observe the respective components during different stages of interpreting, or in this case, sight translation. Sight translation requires the interpreter to comprehend the input (written text in Mandarin Chinese), reformulate it into English with the appropriate words, grammar and lexical structure, then produce the interpretation in English. Silent reading only requires comprehension of the written text, which is later confirmed by the results of comprehension questions; and reading aloud requires comprehension and production, since the participant needs to read out the written source text. By comparing eye movement records during different tasks afterwards, Huang (2011) concluded that when interpreting from Mandarin Chinese to English, the participants, or novice interpreters, usually employed a meaning-based strategy, also known as the vertical perspective as proposed by Macizo and Bajo (2004). This strategy indicates that the individual receive information chunks for comprehension, then reformulate and produce it in another language, or “deverbalize”, after the comprehension process is complete (Seleskovitch, 1976). Furthermore, the action “reading ahead” was also 42.

(51) observed during the experiment, proving sight translation is similar to simultaneous interpreting as there is an overlap of input (reading/listening) and production (Huang, 2011). Eye movement results from this study are compared with those from Huang (2011) to observe if experienced interpreters perform sight translation differently. In conclusion, past studies on expertise indicate that an experienced expert should be able efficiently deploy their efforts and show flexibility in order to complete a task, and they also consistently deliver quality performances. However, so far there is still no consensus across all fields of profession on how to quantize and objectively evaluate an individual as an “expert”. And the situation is even more so in the interpreting field. Currently, “years of practice” is often used as an important factor when determining if an interpreter is good or not; however, whether an experienced interpreter actually delivers a quality interpretation has yet to be proven. Furthermore, experts should be able to decide and execute immediate strategies when performing sight translation, as Dreyfus (2004) mentioned in his study. Therefore experienced interpreters should also show this trait during their sight translating process, which may be reflected in shorter fixation durations in some eye movement indices. In order to observe the sight translation process of experienced interpreters, 43.

(52) this study extends Huang’s (2011) eye movement study by observing eye movement duration indices of experienced interpreters, including first fixation duration, second fixation duration, gaze duration, rereading time, and total viewing time, and comparing those to Huang’s (2011) novice results. Based on eye-tracking data, this study attempts to prove that more experience in the interpreting field (experienced interpreters) does help interpreters produce better quality performances, and enables them to execute faster, more efficient strategies during sight translation.. 44.