外國英文口音對專業及學生口譯員辨識度及理解力之影響

全文

(1)國立臺灣師範大學翻譯研究所碩士論文. 外國英文口音對專業及學生口譯員辨識度及理解力之影響. A Comparison of Expert and Trainee Interpreters’ Intelligibility and Comprehension of Foreign-accented Speech. 指導教授：陳子瑋研究生：高毓鎂中華民國一○三年七月.

(2) 致. 謝. 辭. 自開始著手論文寫作以來，一直期盼著寫謝辭，很感動這一刻終於到了。要兼顧工作與論文，確實有辛苦之處，但一直驅使我要完成的動力，來自於對翻譯所的感謝。就讀翻譯所一年級時，面臨許多的挫折與對自我的懷疑，非常感謝翻譯所給我機會重新認識自己，自我成長，也很感謝所上專業的老師，優秀的同學給我鼓勵與指導，大家一起練習、互相打氣的時光，一直會是我珍惜的美好回憶。本論文得以完成，首先要感謝論文指導教授陳子瑋教授。謝謝老師對我的包容與耐心，忍受我緩慢的步調，亦感謝老師一路上給我的教導與鼓勵，一年級的上課情景仍歷歷在目，能當老師的學生實感幸運與榮幸。此外，感謝口試委員汝明麗教授與張嘉倩教授，兩位老師口試時仔細地給予我評論與指正，並提供寶貴的建議，使論文品質更加提升。亦感謝翻譯所共同打拼的同學們，跟你們一起上課、練習、吃飯、聊天、徬徨不安的回憶真是美好。特別感謝我的好同學好朋友兼小老師郭恬君，Gina 不僅在本論文寫作過程中不斷地給予我協助與建議，在翻譯所求學的過程中，也是我努力學習的對象。亦感謝秋慧助教、容嫣不停地給我打氣，一直期盼趕快把我送出去。謝謝工作上的長官、同事、朋友對我的關心與包容，鼓勵我將論文早日完成。最後要感謝家人，尤其是我的父母，他們須要包容女兒平日加班，週末窩在圖書館看不到人，但仍不斷地支持我並用各種方法從旁給予協助，非常感謝他們。感謝身邊所有我關心及關心我的貴人朋友們，有你們真好! 高毓鎂 2014 於台北.

(3) Abstract This study investigates professional and trainee interpreters' intelligibility and comprehension of foreign-accented English speech in a simultaneous-interpreting like mode. Five professional interpreters (Group P) and ten trainee interpreters (Group T) who are Chinese A and English B were invited to perform shadowing tasks of three different source texts delivered by a native English (American) and nonnative English (Spanish) speaker. After the shadowing tasks, comprehension tests and retrospective interviews were conducted to verify participants' comprehension of the speech and elicit their perceptions of the process during the shadowing tasks and general comments on shadowing/interpreting for foreign-accented English speakers. The results of quantitative and qualitative analysis reveal that: a) Statistically speaking, trainees and professionals do not differ significantly in their intelligibility of a foreign-accented input. b) However, Group P performs significantly better than Group T in terms of comprehension and production under a foreign-accented speaker. c) According to retrospective interview results, Group P is more at ease with their effort management and can still efficiently utilize their interpreting skills despite encountering challenges. d) This difference may be mainly due to the accumulation of expert interpreters’ years of real work experiences. The results in this study suggest that trainee interpreters should engage in deliberate practice of coordinating their interpreting skills to meet different interpreting challenges, with foreign-accentedness being one of the problem triggers. Furthermore, schools of interpreting can facilitate more expert-novice dialogues to enable students to gain more knowledge and understanding of the actual working situation. This can also give them a better idea of which skills and competences they should focus on during their interpreting training.. Keywords: Foreign Accent, Intelligibility, Comprehension, Expert and Trainee Interpreters.

(4) 摘要. 本研究旨在探討及比較專業與學生譯者對於有外國口音的英語談話之辨識度（intelligibility）及理解力(comprehensibility)異同。研究對象為 5 名專業譯者與 10 名學生譯者，針對美國口音與西班牙口音之講者發言進行跟述，並於跟述後接受理解力測驗及回溯訪談。數量與質性分析結果發現：(1) 兩組譯者因外國口音而造成的談話辨識度下降程度並無顯著差異。(2) 專業譯者在有外國口音時之理解力及產出均明顯表現優於學生譯者。(3) 據訪談資料顯示，專業譯者較能因應不同情況有效調整氣力分配及運用口譯技巧。(4) 此一差異性應係主要源自於專業譯者豐富實務經驗之累積。根據研究結果，在口譯技能習得上，學生譯者應有系統的因應不同口譯情境挑戰，有效整合及運用所學之口譯技巧。此外，口譯學校可促進專業及學生譯者之對話交流，以增進學生譯者對口譯工作實務之瞭解及所應具備及強化之技能。. 關鍵字：外國口音、辨識度、理解力、專業及學生口譯員.

(5) Table of Contents Chapter 1 Introduction............................................................................................... 1 Chapter 2 Literature Review ..................................................................................... 4 2.1 Defining the constructs of this study .................................................................... 4 2.1.1 Accent and intelligibility .......................................................................................... 4 2.1.1.1 Accent ............................................................................................................... 4 2.1.1.2 Intelligibility ..................................................................................................... 6 2.1.2 Listening comprehension.......................................................................................... 6 2.1.2.1 Comprehension in Simultaneous Interpreting .................................................. 8 2.1.2.2 Working Memory Capacity .............................................................................. 8. 2.2 The relationship between accent, intelligibility and comprehension ................... 9 2.2.1 The inter-relations of intelligibility and comprehension in the context of accented speech ................................................................................................................................ 9 2.2.2 Factors influencing intelligibility and comprehension in accented speech ............ 10. 2.3 Differences between Expert and Novice Performance ...................................... 11 2.3.1Expertise acquisition ............................................................................................... 11 2.3.2 Expertise performance on comprehension and output in interpreting .................... 12. 2.4 Research Question .............................................................................................. 14 Chapter 3 Research methodology............................................................................. 17 3.1 Participants ......................................................................................................... 17 3.2 Materials ............................................................................................................. 18 3.2.1 Speakers.................................................................................................................. 18 3.2.2 Speech text Preparation .......................................................................................... 19 3.2.2.1 Single Words .................................................................................................. 19 3.2.2.2 Strings of Words ............................................................................................. 20 3.2.2.3 Speech Transcripts ......................................................................................... 21 3.2.3 Recordings of the Stimulus Materials .................................................................... 21. 3.3 Procedures .......................................................................................................... 22 3.3.1 Research Methods .................................................................................................. 22 3.3.1.1 “Smart” Shadowing ........................................................................................ 22 3.3.1.2 Comprehension tests....................................................................................... 23 3.3.1.3 Retrospective interviews ................................................................................ 24 3.3.2 Instructions ............................................................................................................. 25 3.3.2.1 Shadowing ...................................................................................................... 25.

(6) 3.3.2.1.1 Single words ............................................................................................ 26 3.3.2.1.2 Strings of Words ..................................................................................... 26 3.3.2.1.3 Speeches .................................................................................................. 27 3.3.2.2 Retrospective Interview .................................................................................. 27 3.3.3 Data Entry............................................................................................................... 28. Chapter 4 Findings .................................................................................................... 30 4.1 Quantitative findings .......................................................................................... 30 4.1.1 Intelligibility and Meaning Unit scores .................................................................. 30 4.1.2 Comprehension test ................................................................................................ 36. 4.2 Retrospective interviews .................................................................................... 39 4.2.1 Findings of Interview Results ................................................................................. 40 4.2.1.1 Intelligibility, comprehension and processing output..................................... 40 4.2.1.2 Phonological variations that interfered with intelligibility ............................. 42 4.2.1.3 Survival Skills ................................................................................................ 44 4.2.1.3.1 Coping Strategies .................................................................................... 44 4.2.1.3.2 Comprehension skills: Anticipation and contextual clues ...................... 45 4.2.1.4 Effort management ......................................................................................... 47 4.2.1.5 Attitudes toward the Task ............................................................................... 48. Chapter 5 Discussion ................................................................................................. 51 5.1 Discussion on Shadowing and Comprehension Test results .............................. 51 5.3 Discussion of Retrospective Interview Results .................................................. 52 5.3.1 The implications of “smart shadowing” ................................................................. 52 5.3.2. Comprehension Skills: Anticipation and Contextual Clues .............................. 56. 5.3.3 Effort Management ................................................................................................. 59 5.3.4 Attitudes toward the task ........................................................................................ 61. 5.4 Pedagogical implications of this study ............................................................... 62 Chapter 6 Conclusions ............................................................................................... 65 6.1 Summary of the Study ........................................................................................ 65 6.2 Limits of Current Study...................................................................................... 67 6.3 Conclusion .......................................................................................................... 69 Bibliography ............................................................................................................... 70 Appendix ..................................................................................................................... 76 Appendix i: Materials ............................................................................................... 76.

(7) Task 1. Single Words ...................................................................................................... 76 Task 2: Strings of Words ................................................................................................. 79 Task 3 Speeches .............................................................................................................. 81 Speech 1: What should government do to make people happier? .............................. 81 Speech 2: The link between creativity and mental illnesses ...................................... 82. Appendix ii: Sample of Speech Shadowing Transcript ........................................... 83 Appendix iii: Descriptive results of Task 3-Speeches ............................................. 87 Appendix IV: Descriptive results of Comprehension Tests ..................................... 89.

(8) Chapter 1 Introduction The global spread of English as a lingua franca (ELF) is one of the most significant developments of this century (Albl-Mikasa, 2010: 126). English is not only an official language in more than seventy countries, it is also the most widely taught foreign language in over 100 countries (Crystal, 1997 in Kuo, 2012:1). Consequently, it is more and more prevalent in international arenas to have English as the main working language, with speakers from various first language backgrounds. With native and non-native English as the most widely used language in international conferences (cf. Pöchhacker 1994: 154; Basel 2002: 16-19; Neff 2007 in Albl-Mikasa, 2010: 126-127), conference interpreters nowadays cannot ignore the challenges involved and have to integrate appropriate strategies into their toolbox of skills and into interpreting training curriculums (Jenkins 2006). Non-native English accents can make speeches more difficult to understand for any listener. Such comprehension difficulties pose an even greater challenge to simultaneous interpreters than they do to listeners with other purposes, because interpreters can only dedicate part of their cognitive capacities to listening and comprehension, while some resources are used to reproduce the content in another language (Gile, 1995/2009; Padilla, 1995). As successful comprehension is an indispensable prerequisite for correct reproduction in a second language (Yudes, Macizo, Morales, and Bajo, 2013: 2), understanding the challenges of foreign accents and developing techniques to cope with them is crucial to maintaining simultaneous interpretation quality in the current environment of diverse international accents.. 1.

(9) Kuo (2012) provides empirical evidence to investigate and compare professional and trainee interpreters' decoding and processing strategies of foreign-accented English input. The study’s main finding showed that professional and trainee interpreters alike suffer statistically significant intelligibility loss when listening to foreign-accented input. Interestingly though, in retrospective interviews participants generally claimed that this loss off intelligibility did not greatly impair their comprehension of the input text. Although professional interpreters showed better command of certain interpreting skills based on the interviews, the lack of an objective assessment on the level of comprehension the participants had of the source texts was identified as a limitation of that study. Also, the study did not explain how successful comprehension could be achieved without complete intelligibility, should the participants' claims hold true. Therefore Kuo (2012) suggested a follow-up study to answer some of the questions raised by her findings. Based on Kuo (2012), this study aims to further investigate the relationship between accent, intelligibility and comprehension and provide experimental evidence to answer the following pair of questions: 1. How are trainee and expert interpreters’ intelligibility and comprehension affected by foreign-accented input in a simultaneous interpreting-like setting? 2. Are accent, intelligibility and comprehension interrelated, or are they of separate, independent dimensions? Furthermore, following Kuo's suggestion that professional interpreters employ comprehension skills more efficiently than trainee interpreters (2012:92) and Seguinot's (2000: 96) understanding of interpreting as a “toolbox skill”, retrospective interviews are conducted to provide initial thoughts on the strategies interpreters can. 2.

(10) employ to achieve comprehension despite a lack of intelligibility, and on how experts and trainees differ in the application of these strategies. In order to separately investigate how intelligibility and comprehension are affected by foreign-accented input, this study conducts experiments with different kinds of input materials. The first two tasks consists of source texts that deprives the interpreter of any context, structure or other aspects that could serve as clues to overcome a loss in intelligibility. This way participants' output should provide a fairly accurate representation of his/her intelligibility of the source text. The third task aims to evaluate participants’ comprehension of foreign-accented speech. This is done by two methods, which include allowing participants’ to adopt “smart shadowing,” followed by Q&A comprehension tests. The study first reviews relevant literature to define the constructs of this study, and to look into the relationship between accent, intelligibility and comprehension and expert-novice differences. Next, the Methodology chapter explains the experimental designs and execution of the data collection procedures. The following chapters present the findings of the quantitative and qualitative results, and the discussions and analyses of the results. Finally, the study concludes with a summary of the main findings and suggestions for pedagogical applications.. 3.

(11) Chapter 2 Literature Review As an extended study of Kuo (2012), this study focuses on the differences between expert and novice interpreters’ intelligibility and comprehension of foreign-accented speech in a simultaneous interpreting-like setting. This chapter seeks to define the main constructs of this study and review relevant literatures on the relationship between accent, intelligibility and comprehension, and the comparison between expert and novice interpreters’ performance in comprehension. These studies serve as useful guidelines for the methodology and general discussions of this study.. 2.1 Defining the constructs of this study 2.1.1 Accent and intelligibility. The two main constructs explored in Kuo (2012) are accent and intelligibility. As an extended research, this study follows the definitions set out in her research. The following two paragraphs summarize and justify these definitions.. 2.1.1.1 Accent. Kuo (2012, 9-12) refers to (Wells, 1982:1) in defining accent as “someone's structural and systematic articulations of the segmental and prosodic properties of the English language.” Therefore, as is also largely agreed amongst researchers on the topic of foreign-accent (Hsiao, 2011), “grammatical or lexical differences in the use of the English language does not apply when discussing the issue of accents of English, be it native or nonnative.”. 4.

(12) Non-native speakers of English demonstrate many different accents, mostly deriving from their native language's phonology. But also native speakers' accents vary significantly according to factors such as region and social economic class. In practice, both native and non-native accents can cause comprehension difficulties for interpreters. Still, this study follows Kuo's (2012) argument for the differentiation between native and non-native variations in accent, and her focus on the latter. Firstly, native accents are mostly coherent, while non-native accented speech may contain various realizations of the same phonemes. A second additional challenge only found in non-native accents consists in the inclusion of phonological variations that do not exist in a native English speaker's language (Floccia et al. 2006). This second challenge might be less relevant for non-native speakers of English if these variations exist in their native language. Still, listeners with an advanced or near-native level knowledge of English – such as interpreters – will perceive these variations as a deviation from the norm. In order to study the effect of non-native accents on interpreters' simultaneous interpreting performance, this study uses a control native speaker who speaks in a General American accent. This accent "comprises that majority of American accents which do not show marked eastern or southern characteristics" (Wells, 1982b, p. 470). This accent is also chosen because it is what Taiwanese students are most frequently exposed to, be it for political reasons (Tse 1987, 86), or their parents' educational choice (Chang, 2008; Liou, 2010).. 5.

(13) 2.1.1.2 Intelligibility. Smith and Nelson (1985) were the first to establish a differentiation between the terms intelligibility and comprehensibility. They define intelligibility as the degree of “word/utterance recognition” and comprehensibility as the degree to which one can grasp “word/utterance meaning” (Smith and Nelson 1985: 334). Other definitions also exist, for example Derwing and Munro (1997) do not differentiate between word/utterance recognition and meaning. They define intelligibility as "the extent to which a speaker's message is understood by the listener" (Munro 2006: 332), and comprehensibility as "the listener's perceived degree of difficulty in understanding an utterance" (Munro, 2006, p. 334). Thus both terms touch upon the grasp of meaning, but they differentiate along the line of the listener's perceived (comprehensibility) or actual (intelligibility) degree of understanding. As defined by Kuo (2012:14), accent is a purely phonological phenomenon. In order to isolate and measure its effect, Smith and Nelson's (1985) definitions of intelligibility and comprehensibility allow for this distinction. That is to say intelligibility is the ability to identify the utterances on a phonological basis, while comprehensibility is to understand the intended message based on the speech context, or the larger context of world or cultural knowledge.. 2.1.2 Listening comprehension. As the main construct of this study, the term comprehension should be further elaborated. It should first be emphasized that while Smith and Nelson (1985) refer to comprehensibility as the degree to which one can grasp “word/utterance meaning,” this study focuses purely on oral utterances, excluding the comprehensibility of. 6.

(14) written texts. Hence the mode of comprehension referred to here is “listening comprehension.1” Optiz and Zbaracki (2004) study the activity of “listening”, and draw upon the commonalities from several references (Lundsteen, 1979; Samuels, 1984; Wolvin and Coakley, 1996, Goss, 1982) to conclude that listening is “an active process that includes the attention of meaning to the spoken message.” This is very different from “hearing,” which is merely the ability to discriminate among the spoken sounds. For example, one can hear sounds or someone speak, but not recall or understand what was said because of distraction or other factors. In other words, “listening has not occurred unless comprehension has occurred” (Optiz and Zbaracki, 2004:2). Therefore, listening and listening comprehension can be seen as interchangeable terms, which infer that “when one comprehends while listening, one should be able to convert spoken language to meaning in the mind.” (Lundsteen, 1979: 1) The construction of meaning from the spoken language is influenced by several “inthe-head” and “out-of-the-head” factors. In the context of this study, the most relevant “in-the-head” factors include: motivation for listening, good habits of listening (including listening for ideas, concentrating, summarizing the speakers ideas, questioning what has been said, etc.), language facility (including the speed and accuracy with which the listener hears the words and the degree to which the listener understands the speaker’s vocabulary, etc), and background knowledge. “Out-of-thehead” factors include the context and the clarity of the message. Based on these factors, one should be able to interpret2 the meanings comprehended, which results in. 1. Throughout the rest of the paper the term “comprehension” will refer to this construct of “listening comprehension.” 2 “Interpret,” or “interpreting” mentioned in this paragraph mean “to describe the meaning of something; examine in order to explain”, it is different from the “interpreting,” in this study,. 7.

(15) the capacity to explain the spoken text or answer questions about its content. (Optiz and Zbaracki, 2004:3-4). 2.1.2.1 Comprehension in Simultaneous Interpreting. Comprehension becomes more challenging when it comes to simultaneous interpreting. This is due to the requirements and conditions unique to the process of simultaneous interpreting: the simultaneity of comprehension and production under time pressure (Christoffels and de Groot, 2005 in Padilla, Bajo and Macizo, 2005: 207), the necessity to suppress irrelevant information, and the rapid integration of different long-term or short-term memory into the discourse (Bajo, Padilla, Munoz, 2001: 28). It is estimated that at least 80% of the effort or cognitive resources is devoted to listening and understanding the discourse, while only 20% is distributed to speech production (Padilla 1995 in Bajo, Padilla, Padilla, 2000: 128). Due to these circumstances, studies suggest that specific cognitive skills, or a more efficient use of these skills may be necessary for the comprehension process in simultaneous interpreting.. 2.1.2.2 Working Memory Capacity. Among these, “working memory” is most frequently mentioned (Gile 1995; Daro & Fabro 1994; Padilla, Bajo, Canas & Padilla 1995 in Bajo, Padilla & Padilla, 2000; Liu, Schallert & Carroll, 2004). As Bajo, Padilla & Padilla (2000: 128) point out, in simultaneous interpreting, one must simultaneously concentrate on and understand a unit of meaning or chunk of discourse in a given language (L1), while translating it. which is basically “to change what someone is saying into the words of another language.” (Definition retrieved from Cambridge Dictionaries Online.). 8.

(16) into a second language (L2). In order to implement this, the interpreter must be able to maintain the new unit in his/her working memory, access the meaning of the words and phrases involved, connect the information received to previous information, and translate this unit into a new linguistic code while producing the translated version of a previous unit. In short, working memory “is considered to be a limited-capacity mechanism that is involved in both the processing and storing of currently active information while tasks are being carried out” (Baddeley & Hitch, 1974 in Liu, Schallert & Carroll, 2004: 20). When it comes to simultaneous interpreting, it is generally agreed that better management of working memory capacity contributes to the efficiency in comprehension and thus to one’s overall interpreting skill. (Giles 2009; Daro&Fabro 1994; Padilla, Bajo, Canas &Padilla 1995; Christoffels, de Groot and Kroll, 2005:469; Liu, Schallert & Carroll, 2004) As this study focuses on interpreters’ comprehension of accented speech in a simultaneous interpreting-like setting, performance can in part be attributed to the efficient use of one’s working memory capacity.. 2.2 The relationship between accent, intelligibility and comprehension 2.2.1 The inter-relations of intelligibility and comprehension in the context of accented speech. With regards to the relationship between accent, intelligibility and comprehensibility, Munro and Derwing see these constructs as partially independent. It was concluded that a strong foreign accent does not necessarily reduce intelligibility and comprehensibility, in the sense that an utterance could be completely intelligible yet be perceived as having a heavy accent. (Munro and Derwing, 1995: 90; Derwing and Munro, 2009: 479) Likewise, in a research on the perceived accentedness and 9.

(17) intelligibility of native and non-native conversational and clear speech productions,3 it was found that when the speaker adopts clear speech modifications, non-native listeners’ intelligibility of the accented speech increased from 49 to 62%, while perceptions of the foreign accentedness remain constant. (Smiljanic and Bradlow, 2011: 4028-4029). It should be noted, however, that the definition of constructs are different in these studies: in Smiljanic and Bradlow’s research design, intelligibility, as in this study, solely signifies word recognition, while in Derwing and Munro’s study, it was defined as “how much the listener actually understands,” (Derwing and Munro, 2009: 479) therefore it includes word recognition and actually comprehending what was said. Nevertheless, the two studies resonate one another in the sense that accentedness and intelligibility or accent and comprehensibility are in part independent dimensions of non-native speech.. 2.2.2 Factors influencing intelligibility and comprehension in accented speech. According to past research, there are several factors that may affect intelligibility and comprehension in an accented speech: Gass and Varonis looked into “familiarity factors,” including “familiarity with the topic, with listening to non-native speech, with a particular accent, or with a particular speaker.” They concluded that all these have an influence on listeners’ comprehension, with topic familiarity being the prime factor. (Gass and Varonis, 1984:77) Apart from familiarity factors, studies also looked into supra-segmental features that. 3. Detailed acoustic and articulatory analyses have identified typical conversational-to-clear speech modifications, including a decrease in speaking rate, wider dynamic pitch range, greater sound-pressure levels, more salient stop releases and increased energy in the 1000-3000 Hz range of long term spectra. (Picheny et al., 1986, 1989; Matthies et al, 2001; Perkell et al., 2002; Krause and Braida 2004; Bradlow et al., 2003; Liu et al., 2004; Ferguson and Kewley-Port, 2002, 2007; Smiljanic and Bradlow, 2005 in Smiljanic and Bradlow, 2011: 4020-4021). 10.

(18) could account for high variance in participants’ accent and comprehensibility ratings. (Kang 2010; Anderseon-Hsieh et al., 1992; Munro and Derwing, 1998, Munro and Derwing, 2001 in Derwing and Rossiter, 2002: 156) Supra-segmental factors include speech rate, pause patterns, pitch variation and prosody. Of which prosody is considered to be more important in producing intelligible utterances and have more significant effect on listeners’ pronunciation and comprehensibility judgments.” (Anderson-Hsieh, Johnson and Koehler (1992) and Derwing, Munro and Wiebe (1998) in Kuo, 2012: 17) In addition, speakers’ “grammatical and lexical errors” are also relevant to listeners’ subjective experience of the ease or difficulty of understanding L2 speech. (Trofimovich and Isaacs, 2012: 9) As the purpose of this study is to narrow the scope to only the foreign-accented English input factor, which is defined above as someone's structural and systematic articulations of the segmental and prosodic properties of the English language, the research design should exclude potential influences from familiarity, lexical and grammatical factors. To this end, the speakers should read from the same pre-prepared speech script based on generic topics in a natural manner as if they were speaking to a real audience.. 2.3 Differences between Expert and Novice Performance 2.3.1Expertise acquisition. Studies on expertise acquisition is of interest for it illuminates our understanding of how experts become the way they are, and how that could be an inspiration for aspiring expert-to-be novices. Dreyfus and Dreyfus (1986) developed a five-stage. 11.

(19) model of adult skill acquisition in which one goes through the processes of novice, advanced beginner, competent, proficient, and finally the expert level. In the stage of expertise, three characteristics are identified: knowing how to bring knowledge to practice, being able to act upon intuition, and having large amount of experiences (Tsui 2003 in Ju, 2010: 168). Apart from vast amount of experiences, Ericsson also stress on the substantial time period of preparation and deliberate practice that contribute to expertise (1996:10). The key elements of deliberate practice include: (1) a task with a well-defined goal, (2) motivation to improve, (3) feedback, (4) ample opportunities for repetition. On top of that, effort and concentration are characteristics to achieve effective learning that improves performance. (Ericsson, 2008:991) Hence, in expert performances, it is rather the “vast amounts of knowledge and pattern-based retrieval” acquired from the many years of deliberate practice than “exploring new methods with unknown reliability” that makes the experts stand out. (Chase and Simon, 1973 in Ericsson, 1996:14) This is why it is often mentioned in expert studies that experts seem to acquire superior “chunking” skill, which is processing information in a more organized, structured and efficient way. (Seamster et al. 1993: 280 in Moser-Mercer, 1997: 256) The years of deliberate practice not only allow experts to utilize vast amounts of knowledge, but also to gradually do so in an “automatic and intuitive” manner. (Hoffman 1996 in Moser-Mercer, 1997:260). 2.3.2 Expertise performance on comprehension and output in interpreting. Defining expertise in interpreting also serves as a significant reference for assessment, training, and self-growth. In acquiring expertise in interpreting, Moser-Mercer also notes three stages that interpreters would go through: first a cognitive stage of learning relevant facts about interpreting; then an associative stage, during which 12.

(20) novices learn from his errors, corrects and strengthens; and eventually arriving at the autonomous stage, where procedures become more rapid and require fewer processing resources (Moser-Mercer, 1997:260). As have been pointed out in many studies, expert interpreters’ better performance may not result from better verbal fluency (Moser Mercer, 2000), language ability (Cai, 2001 in Liu, 2008) or superior working memory spam (Liu 2004). What sets them apart from novices is better and more efficient coordination effort of their mental resources and utilizing different strategies and skills to meet the different tasks of each specific interpreting situation (Liu, 2008:174). Some interesting findings from past studies on expert-novice differences reflect these abilities and may serve as a point of reference for this study. When it comes to comprehension in interpreting, Moser-Mercer points out that experts may have better performances because they closely tie their semantic knowledge to the context of a speech or text whereas novices’ semantic interpretations are often entirely unrelated to the context. (Moser-Mercer, 1997: 257) This echoes with Gile’s reference that comprehension of interpreters and translators require the understanding of “the functional and logical infrastructure” of the source text, so that the interpreter can utilize appropriate “equivalents,” or other terms or paraphrases to express the same message in the target language. (Gile, 2009: 88) In addition, it is also suggested that expert interpreters are more able to efficiently utilize their working memory resources to avoid the deleterious effect of articulatory suppression, which also contributes to their more superior ability to comprehend and produce simultaneously. (Padilla, Bajo and Macizo, 2005: 208). 13.

(21) In terms of frequent obstacles that may come across in interpreting, such as high input speed or unfamiliar terminologies, experts often swiftly adjust and apply adequate strategies, such as lexical or syntactical compression so as to not lag so far behind. Moser-Mercer also observes that experts tend to proceed from known to unknown information, searching the surrounding context for semantic clues whereas novices tend to focus on the unknown and get stuck. (Moser-Mercer, 1997, p.257) In summary, based on years of practice and experience, professional interpreters adopt skills and judgments that allow them to switch their attention efficiently in each of the comprehension, translation, and production processes in response to the required situation.. 2.4 Research Question The two main purposes of this study are to: 1. Compare. how. novice. and. expert. interpreters’. intelligibility. and. comprehension are affected by foreign-accented input in a simultaneous interpreting-like setting. 2. Determine whether the two groups differ in their effort management skills or strategies to maintain simultaneous interpreting quality when faced with the challenge of foreign-accented input. In Kuo (2012), experiment results show that there are no statistically significant differences between novice and expert interpreters’ intelligibility when listening to both native and non-native speakers. This means that expert interpreters don’t. 14.

(22) necessarily have superior decoding skills when it comes to foreign-accented English speech. However, one should take into consideration that the material used in the experiment of Kuo (2012) to test intelligibility were spoken speeches, which provide topic and context as additional clues to aid intelligibility. As defined previously, accent is a purely phonological phenomenon, and intelligibility is the ability to identify the utterances on a phonological basis. Thus it seems necessary to re-examine both groups’ intelligibility and limit the influencing factor to only the accented pronunciation. This can also better determine whether accent, intelligibility and comprehension are interrelated, or are of independent dimensions. In addition, both professional and trainee interpreters reported that their comprehension of the foreign-accented speech was not greatly impaired by lower intelligibility in Kuo (2012). This means that although both groups’ intelligibility performances have degraded due to the foreign-accented input, they were still able to retain a fair level of comprehension. This was however merely based on participants’ self-perceptions in the retrospective interviews. Therefore, as Kuo (2012: 111) proposed, it is necessary to incorporate objective comprehension assessments to verify this claim. Finally, as mentioned above, in simultaneous interpreting at least 80% of the effort or cognitive resources is devoted to listening and understanding the discourse. In addition, Gile (2004: 160) suggests that foreign accent can be seen as a “problem trigger” that deteriorates output capacity or memorization. Thus the extra time and effort needed to allocate resources for comprehension should also be taken into consideration when assessing the overall impact of foreign-accented input. Therefore. 15.

(23) the second main purpose of this study is to compare how expert and novice interpreters cope with this extra effort requirement.. 16.

(24) Chapter 3 Research methodology. The research methods in this study adopt a within-subjects experimental design, combining shadowing tasks, comprehension tests and retrospective interviews. In order to answer the aforementioned research questions, and to make viable comparisons with Kuo (2012), this study adopts and utilizes the same experimental facilities, location, materials, and seeks assistance from mostly the same group of participants and speakers. However, alternations were made where necessary to meet the purpose of this study.. 3.1 Participants A total of 5 professional interpreters (Group P) based in Taiwan and 10 trainee interpreters (Group T) attending graduate programs of interpreting in Northern Taiwan participated in this study. All participants received a formal email invitation explaining the purpose and procedures of the experiment prior to participation. The setting took place at National Taiwan Normal University, where most of the participants have been training for over 2 years and are familiar with the environment and simultaneous interpreting equipments. Each participant received an honorarium of NT$824 after the experiment. Group P consists of one male and four female Chinese-A and English-B professional interpreters based in Taiwan. They have all attended interpreting schools at the masters’ level and have worked professionally as conference interpreters for over two years. They have thus accumulated at least one hundred to 150 days of conference. 17.

(25) interpreting experiences, which is one of the qualifications to apply for AIIC membership (AIIC, 2012) Group T consists of two male and eight female Chinese-A and English-B interpreting students who have attended graduate schools of translation and interpretation in northern Taiwan for over two years and have received at least one year of simultaneous interpreting training.. 3.2 Materials. 3.2.1 Speakers. The two male speakers invited to take part in this study are the same as those in Kuo (2012), so as to make more reliable comparisons on the results and minimize the effect of possible variables. Speaker A, who serves as the control variable, represents the general American accent, while Speaker S, a native Spanish ESL speaker, represents the non-native English accent in this study. According to Kuo (2012)’s foreign-accentedness rating guideline, Speaker A was rated as having a native English accent and does not have any foreign accent, while Speaker S was rated to have a strong foreign accent. The two speakers should adequately serve the purpose for this study.. 18.

(26) 3.2.2 Speech text Preparation4. In everyday communication, many factors help us understand utterances, even under unfavorable circumstances such as noisy environments. Above all, two main processes are involved: one is “the initial processing of acoustic information through the auditory system, and the other is the utilization of linguistic information that is stored in memory” (Kalilow, Stevens, and Elliott, 1977:1338). Such linguistic information includes the lexical, syntactic, semantic and prosodic properties of a language, which provide a context for the particular utterance and reduce the necessity for listeners to rely on acoustic signals to comprehend the utterance. In other words, such linguistic information serves as cues for listeners to predict and anticipate the utterances when phonetic features deteriorate. Hence, as explained in Chapter 2, in order to clearly distinguish intelligibility and comprehension abilities in a foreignaccented setting, we must exclude such aiding features. Therefore, the first two parts of the shadowing tasks were designed solely to focus on testing listener’s acoustic-phonetic intelligibility, with foreign-accentedness being the independent variable.. 3.2.2.1 Single Words. Task 1 is the shadowing of a set of single, unrelated words. The researcher selected a group of 100 English words from the British National Corpus of frequently spoken words5 which are controlled to be neither rarely used nor very frequently used words in the English language and are balanced in terms of number of syllables and syntactical functions. 4. 5. Please find full texts of the 3 tasks in Appendix 1. http://ucrel.lancs.ac.uk/bncfreq/lists/2_2_spokenvwritten.txt, retrieved 23, Mar, 2014.. 19.

(27) The 100 words are divided in three parts: a practice text of 20 words and two texts of 40 words each. The 40 words texts are evenly distributed by the categories of nouns, verbs, adjectives, and adverbs, while in each lexical category, word syllables evenly range from a number of 1 to 4. The words were also selected to avoid lexical competition, such as “flower” and “flour,” in order to ensure that participants clearly identify the words instead of merely mimicking the sound.. 3.2.2.2 Strings of Words. Apart from shadowing individual words, Task 2 further requires participants to shadow strings of words. In this task, the material was retrieved from Kalilow, Stevens, and Elliott (1977)’s sentence design to test the listening performance in everyday communicative situations. The sentences are balanced in terms of intelligibility, key-word familiarity and predictability, phonetic content, and length (all sentences were constrained to contain five to eight words and six to eight syllables), as well as resembling normal daily spoken sentences. As mentioned previously, listeners usually profit from the semantic, syntactic and prosodic information in a sentence context and are thus able to retrieve lexical information necessary for comprehension when intelligibility degrades. The researcher therefore rearranged the sentences to form semantically anomalous strings of words so that the participants will have to rely on the phonetic properties to complete the shadowing task.. 20.

(28) 3.2.2.3 Speech Transcripts. Task 3 allows participants to perform shadowing tasks in a normal simultaneous interpreting-like setting. The two speech transcripts of approximately 550 words were adapted from different parts of the same speeches used in Kuo (2012). The two speeches, which are each from authentic one-hour long speeches, are titled: 1) What should government do to make people happier? 2) The link between creativity and mental illnesses. According to the Flesch Reading Ease score, Speech one scores 59.5, while Speech 2 scores 61.4, with both being categorized to have a standard/average level of reading difficultness. Moreover, regarding the speech context, no specific domain knowledge is required beforehand to comprehend the speeches.. 3.2.3 Recordings of the Stimulus Materials. The recordings were done with Digital MP3 recording devices with microphone. In terms of Task 1 and Task 2, the speakers were asked to record 20 practice words and 5 practice sentences. This is to accommodate participants with the task requirements and serve normalization purposes in the experiment, which is for a listener to analyze the characteristics of the voice of a speaker that he or she is exposed to for the first time to facilitate accurate processing of the speaker’s speech signals (Field, 2004: 102 in Kuo, 2012: 46). In Task 3, both speakers were asked to the read through the speeches thoroughly in advance and understand the context, then deliver the speeches in a natural way as if they were speaking to an audience. The two speeches were each equally divided in two parts, hence each speaker was asked to record the speech materials in four parts.. 21.

(29) Under the researchers’ supervision, the audio files were then processed by a professional sound-editing studio to optimize the sound quality, and to carefully equalize the volume and speech rate of the speakers’ output.. 3.3 Procedures In this research, all participants were asked to shadow both the accented and nonaccented speakers in the three tasks, followed by a comprehension Q&A test and a semi-structured interview. The full experiment lasts approximately 50 minutes. The experiments all took place in a simultaneous interpreting classroom in a graduate institute of translation and interpretation in northern Taiwan.. 3.3.1 Research Methods 3.3.1.1 “Smart” Shadowing. Shadowing has its similarities with simultaneous interpreting in the sense that both require simultaneous comprehension and production of speech, however excluding the additional complexity of having to reformulate the message and only one language is involved (Christoffels and de Groot, 2004: 228). Therefore, In Kuo (2012), shadowing was used to “elicit tangible evidences of participants’ intelligibility performance in a setting that simulates simultaneous interpreting without the confounding factor of language transition” (Kuo, 2012: 38-91). Shadowing is also the main research method in this study, however with some modifications to meet its specific purpose. An important guideline concluded in chapter two was to clearly differentiate between intelligibility and comprehensibility,. 22.

(30) and that a certain degree of linguistic flexibility should be allowed to reflect participants’ comprehension of the incoming speech. Therefore, rather than instructing participants to shadow word for word, as in Kuo (2012), participants were asked to conduct “smart shadowing,” which allows them to utilize techniques of paraphrasing, summarizing or omissions where they find necessary. According to Danks et al. (1997:52 in Pippa and Russo, 2002:251), “simultaneous oral paraphrasing” is a step closer for its analogies with the simultaneous interpreting process: “when both input and output concern speech rather than writing, and the person producing the paraphrase does so on-line (simultaneously), the demands of the paraphrasing task appear very similar to those of simultaneous interpretation.” Marinetto (1998: 85 in Pippa and Russo, 2002: 251) similarly focuses on the ability to express meaning with linguistic flexibility as an indicator of ones degree of understanding: “a person has really understood a text, i.e. its internal links, only when he/she is able to reword it in many different ways.” Hence, as the purpose of this study is to verify participants’ comprehension of foreign-accented speech, paraphrasing is seen as an aid for participants to reflect their understanding of the incoming message despite the possible degradation of intelligibility. It is however not the purpose to add extra burden on participants to test their language flexibility, thus participants are informed to only do so when they find necessary.. 3.3.1.2 Comprehension tests. Despite the benefits of paraphrasing as an evaluating criterion, it should be taken into account that unsuccessful paraphrasing may not always be the result of poor. 23.

(31) understanding, but probably rather of poor performance in language production. Therefore, comprehension tests are necessary to gain a more holistic grasp of participants’ degree of comprehension of a foreign-accented speech. “Discourse comprehension, from the viewpoint of a computational theory, involves constructing a representation of a discourse upon which various computations can be performed, the outcomes of which are commonly taken as evidence for comprehension” (Kintsch, 1988: 163). Thus, after comprehending a text, one should reasonably be expected to answer questions regarding the context, recall, summarize, paraphrase, verify statements about it, and so on. Hence, this study adopts open Q&A comprehension questions as a method to allow participants to explain and elaborate what they have understood.. 3.3.1.3 Retrospective interviews. Apart from quantitative shadowing and comprehension test results, this study also conducts retrospective interviews to serve as a qualitative compliment to the statistical results. The interviews serve two purposes: to identify where and understand the reasons why participants face difficulties during the shadowing process. Secondly, to gather and compare different points of views by standardized questions directed at every participant.. 24.

(32) 3.3.2 Instructions 3.3.2.1 Shadowing. In each task, participants would hear one speaker delivering the first half of the text/speech, with the other speaker delivering the latter part. In order to avoid ordering effects and progressive errors, (Gravetter and Forzano, 2009 in Kuo, 2012: 47) different treatment protocols were designed for the participants: Table 3.1 Treatment Protocols for the shadowing tasks. Protocol 1. Protocol 2. Protocol 3. Protocol 4. Task 1-. Task 2-. Task 3-. Single Words. Strings of Words. Speeches 1 and 2. Text 1 (A). Text 1 (S). Speech 1-1(A). Speech 2-1 (S). Text 2 (S). Text 2 (A). +Speech 1-2 (S). +Speech 2-2 (A). Text 1 (S). Text 1 (A). Speech 1-1 (S). Speech 2-1 (A). Text 2 (A). Text 2 (S). +Speech 1-2 (A). +Speech 2-2 (S). Text 1 (A). Text 1 (S). Speech 2-1 (A). Speech 1-1 (S). Text 2 (S). Text 2 (A). +Speech 2-2 (S). +Speech 1-2 (A). Text 1 (S). Text 1 (A). Speech 2-1 (S). Speech 1-1(A). Text 2 (A). Text 2 (S). +Speech 2-2 (A). +Speech 1-2 (S). In terms of task 3, in order to retrieve a more reliable mean score of the interaction of intelligibility and comprehensibility, the order of speakers was reversed for the second speech each participant shadowed. The assignment of the Protocols was evenly distributed according to the order participants signed up to take part in the experiment. All the instructions to the participants were written down beforehand, and their clarity. 25.

(33) and procedure were tested in two rounds of pilot tests beforehand. The instructions were closely followed in each experiment and the location was kept constant.. 3.3.2.1.1 Single words. Task 1 begins with a 45 seconds practice text to familiarize participants with the speed, combination of unrelated words, and shadowing. The participants were then informed to shadow a total of 80 words stimuli that lasts about 2 minutes and 42 seconds, which includes a change of speaker in the middle. After the task, participants were first asked to write down their thoughts on the task, and then take out the transcript of the 80 words that was prepared in their drawer beforehand. The researcher then went through the words with the participants to ensure that the words they enunciated were based on clear intelligibility without hesitation or guesswork.. 3.3.2.1.2 Strings of Words. Similarly, before task 2, the researcher explained the nature of the words strings and played 5 practice texts to familiarize participants with the speed and shadowing words strings. This session lasted around 4 minutes. After the session, the researcher also checked immediately with the participants on each hesitation or unclearness to ensure correct intelligibility counts.. 26.

(34) 3.3.2.1.3 Speeches. After the previous two tasks, the researcher explained that the third task involves the shadowing of two speeches. Participants were explicitly informed that they were not required to shadow word for word, but could feel free to paraphrase or utilize other interpreting skills where and when they find necessary. Participants were also informed that there would be a change of speakers in the middle of each speech, and that there would be a comprehension test directly following the end of each speech. Before starting the session, the topic of each speech was read out loud two times to the participants to ensure that they are prepared and have no further questions. Since the content of the speeches are fairly general and do not contain any technical terms, it was not considered necessary to give extra time for participants to prepare. After the session, participants were first given time to write down opinions and reflections of the shadowing process before the researcher conducted comprehension tests. After the comprehension tests, which includes four questions for each speech, participants were invited to come out of the booth and take part in a retrospective interview.. 3.3.2.2 Retrospective Interview. The retrospective interviews, which were conducted in mandarin Chinese, contained standardized questions and individual questions based on the researcher’s notes or participants’ own reflection. Participants were first asked to talk about the notes they have written down after the end of each speech. The researcher also further questioned what has happened or went through their mind where participants seemed to encounter difficulties. Taking into. 27.

(35) account the validity of retrospective interviews, the researcher avoided referring to any possible explanations to what she has observed, this to ensure a fair and accurate reflection of participants’ cognitive process (Ericsson and Simon, 1993 in Kuo, 2012: 54). Finally, based on the feedback from the pilot tests and the purpose of this study, specific questions were asked to solicit participants’ comments on the relationship between foreign-accentedness, intelligibility, and comprehension: 1.. Does the foreign-accentedness affect your intelligibility or comprehension? If you feel intelligibility degrading, do you think it affects your comprehension?. 2.. How do you cope with your effort management under the two different speakers?. 3.. Were there any attempts of utilizing systematic paraphrasing or summarizing skills? Do you consider these a helpful strategy to the shadowing tasks?. 3.3.3 Data Entry. The recordings of the 3 shadowing tasks, comprehension tests and retrospective interviews were transcribed. In terms of intelligibility for the single words, strings of words and speeches, participants’ full recognition and articulate rendering of each word serve as criteria for correctly shadowed words. If there were hesitations, repetitions, self-corrections or wrongly placement of accent, the words were double checked with the participant to ensure correct acknowledgement. For the purpose of evaluating intelligibility, only omitted or wrongly enunciated words were counted as errors. The correctness of the omitted, added, or substituted materials, on the other hand, were evaluated all together under meaning unit scores. This is to determine whether participants still hold a certain level of control over their output and. 28.

(36) comprehension despite the possible degradation of intelligibility. The intelligibility score in this segment thus serve as a comparison to the meaning unit score. It may be likely that a subject receive a low intelligibility score (maybe because of many omissions) but scores well in terms of meaning unit outputs. In order to determine the correct rendering of a meaning unit, two native English speakers from New York State of the United States were invited to participate in the evaluating process. They were first informed to read through the speech text to understand the content and verify the segmentation of the meaning units, then they listened to the participants’ recordings from the perspective of a listener, determining whether participants’ output carry out a clear and apprehensible meaning which is equivalent to the original speech. An inter-rater reliability analysis using Kappa statistic was performed to determine the consistency between the two evaluators. The inter-rater reliability under Speaker A was found to be Kappa = 0.62, and Kappa = 0.87 under Speaker S. These signify that the evaluators reach “a substantial agreement” of participants’ performances under the native speaker, and an “almost perfect agreement” of participants’ performances under the foreign-accented speaker. (Landis and Koch, 1977: 165) The number of correctly shadowed words and meaning units were counted and composed into an Excel spreadsheet. The data was then fed into the SPSS Statistics 17.0.0 system for statistical analysis.. 29.

(37) Chapter 4 Findings The objective of this study is to observe and compare trainee and professional interpreters’ intelligibility and comprehension of foreign-accented speech in a simultaneous interpreting-like mode, and to identify whether there are different uses of techniques and strategies to enhance these capabilities. Based on the results from the aforementioned methods and procedures, the quantitative and qualitative findings are analyzed and presented in this chapter.. 4.1 Quantitative findings In this section, the findings attempt to answer the questions of whether participants’ intelligibility and comprehension degrade due to foreign-accentedness, and if there is a significant difference of such degradation between trainee and professional interpreters.. 4.1.1 Intelligibility and Meaning Unit scores. Table 4.1 and 4.2 show the descriptive analysis of participants’ performance of the three shadowing tasks under a native and non-native speaker (Speaker A and Speaker S). In terms of the number of correctly shadowed words for the three segments, the performance of Group T and Group P under the two speakers are similar. (For example, under Speaker S, Group T’s words intelligibility M = 29.80, Group P, M = 30.00; speech intelligibility Group T, M = .81, Group P, M = .85) However, in terms of correctly shadowed meaning units, whereas both groups’ performance are similar under Speaker A (Group T, M = .94, Group P, M = .94), it differs under Speaker S 30.

(38) (Group T, M = .60, Group P, M = .85), with Group P reaching a higher percentage than Group T. Table 4.1 Descriptive Statistics of participants performance under Speaker A. N. Single Words. Strings of. Mean. Std. Deviation. Group T. 10. 38.10. .994. Group P. 5. 38.60. 1.140. Group T. 10. 83.00. 3.055. Group P. 5. 84.00. 2.5507. Group T. 10. .969844. .0353171. Group P. 5. .939321. .1079761. Group T. 10. .941183. .0588033. Group P. 5. .947957. .0842067. Words. Speech (Intelligibility). Speech (Meaning Units). 31.

(39) Table 4.2 Descriptive Statistics of participants performance under Speaker S. N. Single Words. Strings of. Mean. Std. Deviation. Group T. 10. 29.80. 2.898. Group P. 5. 30.00. 3.937. Group T. 10. 70.30. 1.767. Group P. 5. 71.40. 2.074. Group T. 10. .806958. .1071333. Group P. 5. .845221. .1311574. Group T. 10. .600753. .1449253. Group P. 5. .850753. .1051633. Words. Speech (Intelligibility). Speech (Meaning Units). In order to answer the question of “whether participants’ intelligibility and comprehension degrade significantly due to foreign-accentedness”, a paired T-test was performed. As can be seen from Table 4.3 and Table 4.4, Group T shows significant difference in all four parts under different speakers (p < .05), with significant better performance under Speaker A.. 32.

(40) Table. 4.3 Group T Paired T-test. Paired Differences Sig. Std. Mean. Pair 1. TSWA - TSWS. Pair 2. TSoWA -. Deviation. Std. Error Mean. t. Df. (2-tailed). 8.300***. 2.791. .883. 9.405. 9. .000. 12.700***. 4.244. 1.342. 9.463. 9. .000. TSoWS. Pair 3. TAI - TSI. .1628858***. .0812171. .0256831. 6.342. 9. .000. Pair 4. TAM - TSM. .3404301***. .0974689. .0308224. 11.045. 9. .000. ***p<.001 For Group P, in terms of intelligibility of single words, strings of words and speeches, participants show significant difference under the two speakers (p < .05), with better performance under Speaker A. However, in terms of correctly shadowed meaning units, although Group P shows a weaker performance under Speaker S, there is not a significant difference of performance under the two speakers. (p = .247). 33.

(41) Table. 4.4 Group P Paired T-test. Paired Differences Sig. Std. Mean. Pair 1. PSWA - PSWS. Pair 2. PSoWA -. Deviation. Std. Error Mean. t. Df. (2-tailed). 8.600. 4.879. 2.182. 3.942. 4. .017. 12.600***. 2.074. .927. 13.587. 4. .000. PSoWS. Pair 3. PAI - PSI. .094. .034. .015. 6.128. 4. .004. Pair 4. PAM - PSM. .097. .160. .072. 1.355. 4. .247. ***p<.001 Finally, as can be seen from above, both Group P and Group T are subjected to intelligibility and meaning units loss under a foreign-accented speaker, however, it is in the interest of this research to find whether there is a significant difference of that degradation between trainee interpreters and professional interpreters in the speech segments. To do so, an independent T-test was conducted. The results shown in Table 4.5 and Table 4.6 indicate that in terms of speech intelligibility, there are no significant differences in performance under both speakers between the two groups. (Speaker A: p = .568 > .05, Speaker S: p = .554 > .05). 34.

(42) Table 4.5 Speech Intelligibility Independent T-test. Levene’s Test for Equality of Variances. F. Speaker A Equal variances assumed. t-test for Equality of Means. Sig.. 5.349. t. .038. Equal variances not. df. Sig. (2-tailed). .835. 13. .419. .616. 4.434. .568. -.607. 13. .554. -.565. 6.779. .590. assumed. Speaker S Equal variances assumed. .041. .842. Equal variances not assumed. In terms of correctly rendered meaning units, there is no significant difference between the two groups under Speaker A (p = .858 > .05). However, there is a significant difference under Speaker S (p = .005 < .05), indicating that professional interpreters show significant better performance than trainee interpreters in rendering correct meaning units under a foreign-accented speaker.. 35.

(43) Table 4.6 Speech Meaning Units Independent T-test. Levene’s Test for Equality of Variances. F. Speaker A Equal variances assumed. t-test for Equality of Means. Sig.. .576. t. .461. Equal variances not. df. Sig. (2-tailed). -.183. 13. .858. -.161. 6.029. .877. -3.407. 13. .005. -3.807. 10.854. .003. assumed. Speaker S Equal variances assumed. .752. .402. Equal variances not assumed. 4.1.2 Comprehension test. A comprehension test was immediately performed after each speech. Four openended questions were asked regarding the content of each speech. The four questions are equally divided under the two segments read by the two different speakers. Hence from the speeches, there are a total of four comprehension questions to assess participants’ understanding of each speaker. The questions were designed to be broad and general and not focus on specific details that would require forceful memorization. For example, for the speech “What should the government do to make people happier,” the following questions were asked: Segment 1:. 36.

(44) 1. What is the first big issue that the government needs to address? 2. What can the government do to address this issue? Segment 2: 3. What can the government do to help create a happy family life? 4. Please describe what the survey concerning parents and children was about.. In terms of data entry, only answers that render full comprehension of the source text were counted as 1 point, answers that were incorrect or missing relevant items were not counted. For example, the answer to question one should contain key words such as “lessen competition” or “focus more on cooperation”: T10: “I didn’t hear the thesis statement. I didn’t understand what lessen means in the context. I think that made me lose the picture of what the issue is.” (no points were given) T5: “The problem comes from education. There are too much competition with puts pressure on people and make them unhappy. Cooperation should be encouraged.” (1 point) For subjects that answer all questions correctly under the segments of a speaker, they would receive a maximum score 4 out of 46.. 6. Descriptive results of comprehension test as Appendix IV.. 37.

(45) Table. 4.7 Comprehension Test Paired T-test of Group T and Group P. Paired Differences Sig. Std. Mean. Deviation. Std. Error Mean. t. Df. (2-tailed). Pair 1. TCA - TCS. 1.40000. 1.17379. .37118. 3.772. 9. .004. Pair 2. PCA - PCS. .60000. .54772. .24495. 2.449. 4. .070. The result of the comprehension test is evaluated by paired and independent T-tests. From the paired T-test as presented by Table 4.7, Group T shows significant difference in their comprehensibility of the two different speakers (p = .004 < .05), with better comprehension of Speaker A, while Group P shows no significant difference of their comprehensibility of the two speakers (p = .070 > .05).. Table 4.8 Comprehension Test Independent T-test. Levene’s Test for Equality of Variances. F. Speaker A Equal variances assumed. t-test for Equality of Means. Sig.. 4.721. t. .049. Equal variances not assumed. 38. df. Sig. (2-tailed). -1.347. 13. .201. -1.744. 12.788. .105.

(46) Speaker S Equal variances assumed. 10.428. .007. Equal variances not. -2.402. 13. .032. -3.206. 12.112. .007. assumed. In terms of the comprehension performance of the two groups under Speaker A and Speaker S, results from the independent T-test as presented in Table 4.8 show that under Speaker A, comprehension performance between the two groups shows no significant difference (p = .201 > .05). However, there is a significant difference (p = .032 < .05) of comprehension performance under Speaker S, with Group P showing significant better comprehension of the foreign-accented speaker than Group T. The results tell us that the two groups are similar in their ability to comprehend the native speaker, however when it comes to the foreign-accented speaker, Group T degraded significantly in their ability to grasp the meaning, while Group P still remained a high level of comprehension which doesn’t differ significantly from that under the native speaker.. 4.2 Retrospective interviews The retrospective interviews were transcribed by the researcher based on the recordings and are cited below if it serves as a representative example to the topic or phenomenon discussed.. 39.

(47) 4.2.1 Findings of Interview Results 4.2.1.1 Intelligibility, comprehension and processing output. In Kuo (2012) research, participants reported that although they were susceptible to intelligibility degradation while shadowing the foreign-accented speaker, their overall comprehension was nevertheless not as much affected. Participants also attributed their omissions to not being able to paraphrase even though they have understood the meaning. Due to the lack of comprehension assessment and the instruction to shadow word-by-word, participants’ comprehension capacity could not be fully verified in that research. Hence, it is the objective of this research to evaluate and observe how the foreign accent affects participants’ comprehension. Two questions were asked in this regard: 1. Has comprehension degraded along with intelligibility loss? Different from Kuo (2012)’s results, in which both professionals and trainees reported that their comprehension was not compromised by intelligibility loss, there is a noticeable difference between professional and trainee interpreters’ perceptions toward comprehension of a foreign-accented speaker in this study. Professional interpreters in general reported that although they would miss a few words from time to time, their overall comprehension was not as much affected due to the aid of the general context or specific keywords. 7 out of 10 trainee interpreters, however, reported that their comprehension was affected by the loss of intelligibility. They claimed that if they missed some words or topic sentences, it would be hard for them to understand the message; hence a lot of omissions and guesswork occurred during the process.. 40.

(48) T2: “I missed the topic sentence, so I couldn’t understand anything that followed, I had to guess. The accent caused comprehension difficulties for me, I have to guess what he’s saying from my background knowledge and the context.” T4: “When there is the foreign accent, I would miss out on a lot of words, and when I don’t get the words, it’s left blank, so I have to wait till he finishes the whole sentence and try to guess by sporadic words. I think I can still figure out what he’s saying, but it’s not as precise and clear as when listening to the native English speaker.” P2: “When there is the foreign accent, I would miss out on words more frequently, but this does not affect my comprehension, because I can rely on the general context.” 2. How does intelligibility loss affect shadowing output? Were there any attempts of systematic paraphrasing? In terms of intelligibility degradation affecting shadowing output, most trainee interpreters reported that they were less able to monitor their output, they stated that listening to the accented speaker was exhausting enough, thus there was limited energy for comprehension, let alone paraphrasing or summarizing. It was also claimed that when they do paraphrase, it was because they haven’t understood the word so they had to quickly guess a word to replace it. T4: “Shadowing affects my comprehension because I have to listen and speak at the same time. I cannot lose focus by summarizing, and when I omit. 41.

(49) something, I find it hard to fill it in with something else. I have to follow closely on a word or sentence basis.” T10: “When I didn't get the keyword or thesis statement, I would feel lost, because I don't know what to say when I didn’t understand the speech. I would only paraphrase when I lagged too far behind, but mainly I would just guess some words and quickly move on.” Professional interpreters, in comparison, had a tendency to utilize more of paraphrasing and summarizing, notably P4 paraphrased extensively under both speakers. Most professional interpreters claimed that they do not paraphrase for the sake of paraphrasing, but that their primary focus was on comprehension, hence they would focus on listening and understanding, and then deliver what they have understood in their own words if necessary. P4: “I wouldn't paraphrase on purpose, I would listen and reproduce the message into what I have understood.”. 4.2.1.2 Phonological variations that interfered with intelligibility. For the sections of single words and word strings, both trainee and professional interpreters found it very difficult to shadow the foreign-accented speaker, in some cases also the native speaker. The reason to this almost all attributed to the lack of context for anticipation. An interesting phenomenon during the word strings section was when the text doesn’t make sense and become unintelligible under the foreign accent, since the strings of words were rearranged from real sentences, many participants would shadow the text into what they thought would be more grammatically correct or would make more sense. For example, “Tom thinking about 42.