雙語工具對台灣高中生句子翻譯影響之研究

全文

(1)國立臺灣師範大學英語學系碩士論文 Master’s Thesis Department of English National Taiwan Normal University. 雙語工具對台灣高中生句子翻譯影響之研究 The Effects of Two Bilingual Referencing Tools on Senior High Students’ Sentence Translation. 指導教授：陳浩然博士 Advisor: Hao-Jan Chen, Ph.D. 研究生：韓雙憶 Student: Shuang-yi, Han. 中華民國 108 年 8 月 August, 2019.

(2) 中文摘要本文旨在比較不同類型的雙語檢索工具輔助台灣高中生翻譯表現的成效，研究對象為普通高中二年級的三個班級，共計 113 人參與本研究。一組使用線上雙語檢索系統 Concordancer Alpha，一組使用線上雙語字典 Yahoo Dictionary，一組為控制組，沒有使用任何檢索工具。三組以一份相同題目的中翻英測驗題目進行前後測與延遲後測，藉以觀察使用雙語檢索工具前後的翻譯成效，以及該成效是否能維持。前測結果顯示三組在該翻譯題目上的知識具有同質性。前測後，三組各接受一節課的句子翻譯教學，接著兩組實驗組分別接受所分配到的雙語工具的使用說明與訓練，在訓練中受試者皆需要使用各自的雙語工具完成練習題，訓練結束後的一個禮拜進行後測，在這一個禮拜期間，受試者被鼓勵繼續使用並探索該雙語工具。前測的一個禮拜後實施後測。在後測中，控制組被要求不能使用任何檢索工具來完成句子翻譯，而兩組實驗組的受試者則利用雙語工具進行句子翻譯。完成句子翻譯後，兩組實驗組學生皆須填寫問卷，以了解學生對於使用雙語檢索工具完成翻譯測驗的想法，接著於兩個禮拜後，三組受試者皆接受延遲後測，在延遲後測中，三組皆無使用任何檢索工具。研究結果顯示雙語工具 Concordancer Alpha 以及 Yahoo Dictionary 對於受測高中生的翻譯成效有所幫助，兩組實驗組學生在後測的表現，皆顯著優於前測，但控制組並無明顯進步，意即兩組實驗組於後測表現皆顯著優於控制組；另外，使用 Concordancer Alpha 的受試者在後測的表現顯著優於使用 Yahoo Dictionary 的受試者。然而，兩組學生在延遲後測表現有顯著退步，但仍然優於前測表現；且 Concordancer Alpha 的受試者在延遲後測的表現仍然顯著優於 i.

(3) 使用 Yahoo Dictionary 的受試者。除此以外，問卷結果顯示，學生對於利用雙語工具進行翻譯學習與練習持正向態度，且未來願意繼續使用雙語工具學習不會的字彙與進行翻譯輔助。學生使用雙語工具遇到的困難及對雙語工具的建議皆在研究中討論。. 關鍵字：翻譯、雙語檢索系統、雙語工具、雙語字典. ii.

(4) ABSTRACT The current study investigates the effects of two bilingual referencing tools, Concordancer Alpha and Yahoo Dictionary, on EFL senior high school students’ translation performance. The subjects are three intact classes of 11th graders in EFL senior high school. Among them, two experimental groups were instructed to use bilingual referencing tools, Concordancer Alpha and Yahoo Dictionary respectively, while the other group served as control group, having no referencing tools to turn to. One pretest, one posttest and one delayed posttest utilizing the same sentence-based translation task were employed in order to compare the translation performance among three groups. The results of pretest ensured the homogeneity of three groups in terms of their knowledge of the translation task. After the pretest, the three groups received instruction on how to do sentence translation. Next, the two experimental groups received instruction on how to use their assigned tools, and they were given an exercise that required them to finish with the tools. After the training, participants in the two experimental groups were encouraged and advised to continue to explore the tool. One week after the pretest, a posttest was administered. In the posttest, the control group was asked to finish the translation task without the help of referencing tool, while the two experimental groups were required to finish the sentence translation task with the help of their assigned bilingual referencing tools. After the posttest, participants in the experimental groups were asked to complete a questionnaire inquiring their perception and experience of doing sentence translation with referencing tools. Two weeks after posttest, the delayed posttest was administered. The results indicated that using bilingual tools was helpful in improving the quality of sentence translation, and learners in both experimental groups made significant iii.

(5) progress in the posttest. In addition, Concordancer Alpha Group outperformed significantly than the other two groups. The results of delayed posttest demonstrated that the although the scores of the two experimental groups regressed significantly after two weeks, their scores in delayed posttest were still higher than those in their pretest, indicating that the residual effect of both referencing tools could be maintained for a period of time. Moreover, in delayed posttest, Concordancer Alpha Group still outperformed the other two groups. As for learners’ perception reported in the questionnaires, most of the participants held positive attitudes toward consulting bilingual tools when doing translation task, and they expressed willingness to continue using the tools for future consultation. In light of the findings of the current research, some pedagogical implications are drawn. Limitations of the present study and suggestions for future research are also presented.. Keywords: bilingual concordancers, bilingual tools, bilingual dictionary, translation learning. iv.

(6) ACKNOWLEDGEMENT This thesis owes its existence to many people. Without their professional guidance, great encouragement and strong support, this thesis would have never been completed. First and foremost, I owe my deepest gratitude to my advisor, Dr. Hao-jan Chen, for his introducing me to the corpus field and useful tools like concordancers. These benefit me not only in research areas but also in pedagogical fields. In addition, his insightful advice, great patience, continuous support and inspiring encouragement during the process of writing my thesis also contributed to the completion of the thesis. Furthermore, I am also grateful to my committee members, Dr. Hsiao-Hui- Yang and Dr. and Chih-Cheng Lin, who spent time reading my thesis and kindly provided me with expert and thorough guidance on how to revise my thesis. All their constructive advice has made my thesis better. Also, this study wouldn’t have been possible without the help of my students in National Hu-Wei Senior High School. They needed to completed the same translation tests three times, and they did not complain once. And many of them tried really hard to finish the questions. I really appreciate their efforts. Without them, my experiment wouldn’t have been completed. Finally, I want to thank my family for supporting me in many ways during this difficult and long journey. I really appreciate they were really encouraging and understanding instead of being pushy the whole time. Without my family and my dearest friend, Michelle Wu, I couldn’t have persevered until the end. Thank you for being there for me.. v.

(7) TABLE OF CONTENTS 中文摘要 ................................................................................................................... i ABSTRACT ............................................................................................................. iii ACKNOWLEDGEMENT ......................................................................................... v CHAPTER ONE INTRODUCTION ......................................................................... 1 1.1 Background................................................................................................... 1 1.2 Motivation .................................................................................................... 2 1.3 Purpose of the Study and Research Questions ............................................... 6 1.4 Significance of the Study .............................................................................. 7 CHAPTER TWO LITERATURE REVIEW .............................................................. 8 2.1 Translation and Language Learning............................................................... 8 2.1.1 The role of translation in Language Teaching and Learning ............... 8 2.1.2 The Application of Translation on Language Learning ...................... 11 2.1.3 L2 Learners’ Difficulties in Translation ............................................14 2.2 Language Learning and Reference Tools ..................................................... 16 2.2.1 The Effects of Bilingual Dictionary on Translation and Language Learning....................................................................................................20 2.2.2 The Effect of Parallel Concordancer on Language Learning .............23 2.3.2 The Application of Parallel Concordancers on Translation Tasks ......34 CHAPTER THREE METHOD.................................................................................39 3.1 Participants ................................................................................................. 39 3.2 Instruments ................................................................................................. 40 3.2.1 Online Bilingual Tools ......................................................................40 3.2.2 Tests .................................................................................................44 3.2.3 Translation Exercise Task .................................................................45 3.2.4 Online questionnaire .........................................................................46 3.3 Procedures .................................................................................................. 47 3.4 Data Analysis .............................................................................................. 51 3.4.1 Inter-rater reliability .........................................................................51 3.4.2 Marking and scoring .........................................................................51 3.4.3 Statistical Procedures........................................................................52 CHAPTER FOUR RESULTS AND DISCUSSION ..................................................54 4.1 Results ........................................................................................................ 54 4.1.1 Pretest Results ..................................................................................54 4.1.2 The results of the Three Tests ...........................................................55 4.1.3 Learners’ Translation Improvement ..................................................62 116.

(8) 4.2 Learners’ Perceptions toward Translation Learning via Bilingual Tools ....... 74 4.2.1 Participants’ Overall Perception of Bilingual Referencing Tools .......75 4.2.2 Strengths and Weaknesses of the Tools Perceived by Learners ..........80 4.2.3 Willingness of using the tools for future words/translations consultation...............................................................................................85 4.2.4 Learners’ suggestions for the bilingual tools......................................87 4.3 Discussion .................................................................................................. 88 4.3.1 Discussion on Learners’ Translation Performance .............................88 4.3.2 Discussion on Learners’ Perception toward the Tools........................91 CHAPTER FIVE CONCLUSION ............................................................................95 5.1 Overview of the Study ................................................................................ 95 5.2 Pedagogical Implications ............................................................................ 95 5.3 Limitation of the Present Study and Suggestions for Future Research ......... 96 REFERENCES.........................................................................................................98 APPENDICES ....................................................................................................... 105 APPENDIX A. Year and Source of Target translation questions and their suggested answer provided by CEEC .............................................................. 105 APPENDIX B. Pretest, Posttest and Delayed Posttest Questions ..................... 106 APPENDIX C. Translation Exercise ............................................................... 110 APPENDIX D. Questionnaire after posttest (Concordancer Group) ................ 112 APPENDIX E. Questionnaire after posttest (Dictionary Group) ...................... 114 APPENDIX F. Target Translation Questions and the Grading Guideline ......... 116. ii.

(9) LIST OF TABLES Table 3.1 Data Collection Procedure ................................................................................................. 49 Table 4.1. Results of ANOVA test on the pretest scores ...................................................................... 55 Table 4.2 The results of the three tests in Three Groups ..................................................................... 56 Table 4.3 Repeated Measures ANOVA Summary for Group, Time, and Type ..................................... 56 Table 4.4 Post-hoc test analysis using Bonferroni .............................................................................. 58 Table 4.5 Summary of Simple Main Effect of Group and Time .......................................................... 59 Table 4.6 Pairwise Comparisons of Time*Treatment ......................................................................... 61 Table 4.7 Pairwise Comparisons of Treatment *Time ........................................................................ 62 Table 4.8 The Comparison of Participants’ Translations in Question 1 ............................................... 63 Table 4.9 The Comparison of Participants’ Translations in Question 2 ............................................... 64 Table 4.10 The Comparison of Participants’ Translations in Question 3.............................................. 67 Table 4.11 The Comparison of Participants’ Translations in Question 4 .............................................. 69 Table 4.12 The Comparison of Participants’ Translations in Question 5.............................................. 71 Table 4.13 The Comparison of Participants’ Translations in Question 6.............................................. 73 Table 4.14 Results of Participants’ Perceptions toward the Use of the Two Bilingual Tools ................. 77 Table 4.15 Participants’ Responses to the Questionnaires ................................................................... 78 Table 4.16 Participants’ Perceptions toward the Strength of the Two Bilingual Tools .......................... 81 Table 4.17 Participants’ Perceptions toward the Weaknesses of the Two Bilingual Tools..................... 84. iii.

(10) LIST OF FIGURES Figure 3.1 A search of “獲取” in Concordancer alpha. ....................................................................... 42 Figure 3.2 A search of “獲取” in Yahoo! Dictionary. ......................................................................... 43 Figure 3.3 A search of “obtain” in Yahoo! Dictionary. ........................................................................ 43 Figure 3.4 Information shown below “Related words” ....................................................................... 44 Figure 3.5 Data Collection Procedure ................................................................................................ 50 Figure 4.1 Mean scores of the Three Tests ......................................................................................... 57. iv.

(11) CHAPTER ONE INTRODUCTION 1.1 Background The advance of technology has made web-based reference tools widely and easily accessible to language learners. Among those reference tools, when encountered with unknown words, most language learners tend to consult online dictionaries because of the user-friendliness and convenience of online dictionaries as well as learners' familiarity with dictionaries (Dziemianko, 2012; Levy & Steel, 2015). Rasmussen also (2010) discovered that a large percentage of learners tended to consult bilingual dictionaries when faced with writing problems. When having difficulty coming up with an expression in the target language, language learners usually turn to bilingual dictionaries (Lai, 2010). With a quick click, learners are presented with a list of translations they may or may not have seen before, from which they choose a word and place it into the sentence they are making. A bilingual dictionary seems to provide sufficient assistance for learners to achieve communication success. However, misuse and misinterpretation of words have been frequently reported, which might be caused by “the ignorance of connotation or lexical collocation” (Nesi & Meara, 1994). Zughoul (1991) indicated that much of learners' writing problems are related to lexical choice and that many of the errors identified can be attributed to learners' use of bilingual dictionaries which fail to provide adequate example sentences. Without being demonstrated how the word is used, learners can only assume that the word works the same as the equivalents in their native languages. Since they are not aware of the impact of contextual clues on word choice, learners may produce grammatically correct sentences but these sentences still sound weird to native speakers. This is why some researchers suggest that bilingual dictionaries be used with caution and that learners should use other reference tools that provide more authentic example sentences as 116.

(12) reference resources (Boulton, 2012; Frankenberg-Garcia, 2005; Gao, 2011). With the rising popularity and significant expansion of electronic corpora, a growing interest in the use of online corpus and concordancer has emerged in the field of language learning and teaching. The potential applications of electronic corpora in language teaching and learning have received considerable attention. Numerous studies have established the effectiveness of concordancers in facilitating grammar, vocabulary, writing and collocation learning (Benavides, 2015; Chan & Liou, 2005; Flowerdew, 1996; Frankenberg-Garcia, Flowerdew, & Aston, 2011; Jafarpour, Hashemian, & Alipour, 2013; John, 2001; Kaur, 2004; Kaur & Hegelheimer, 2005; Liu, 2013; Yeh, Liou, & Li, 2007). More and more researchers and teachers have attempted to introduce corpus and concordancers into classroom. Compared with bilingual dictionaries, corpus tools offer more example sentences, allowing learners to explore the authentic use of the target word and its uses and meanings in different contexts. Among these corpus tools, bilingual concordancers allow users to enter their queries in both the source and target languages, just like bilingual dictionaries do. This function enables users to retrieve information without language barrier and to understand many ways to express their ideas in target language. In short, with the aforementioned advantages, a bilingual concordancer seems to be a more reliable reference source than a bilingual dictionary in terms of writing. Nevertheless, bilingual concordancer as a reference tool is still far from being a mainstream use in the field of language teaching and learning.. 1.2 Motivation Referencing tools are often needed when language learners are doing production task, including writing and translation tasks. Most beginning and intermediate language learners cannot write a sentence without conceiving ideas in their L1 and doing mental translation (Sasaki & Hirose, 1996). In the process of language production, learners are 2.

(13) often in desperate need for bilingual reference tools because they have no idea how the word is translated or what word follows the other word. Some researchers maintained that translation is a skill that most second L2 learners employ on a daily basis for a variety of purposes (Campbell, 2013; Kavaliauskienė, 2010; Mckay, 2003; Naimushin, 2002). For example, Kavaliauskienė (2010) discovered that translation is a common strategy employed by many L2 learners when doing language learning tasks and as a real-life, naturally-occurring activity that many learners engage in every day either in formal or informal circumstances. Ross (2000) also considered translation essential in social interaction since it promotes communication and understanding between people with different L1. In addition, many researchers consider translation as a fifth skill (Naimushin, 2002; Ross, 2000) or “fifth macro-skill” (Campbell, 2002). While translating, learners are actually practicing a multi-skilled activity that involves both receptive and productive language skills. Nevertheless, in a language classroom, translation training has not yet received sufficient attention, and teachers are prone to neglect the importance of translation skills in language learning (Lin, 2011). In fact, a translation task is mostly assigned with the purpose of helping teachers to determine whether learners have learned a given expression or sentence patterns. Furthermore, it is also indicated that teachers use translation mainly for explaining complex language concepts, checking understanding and classroom management (Inbar-Lourie, 2010; Littlewood & Yu, 2011). In short, language learners rarely receive formal training in translation skills. Nevertheless, language learners still need to acquire this language skill for the propose of preparing themselves for social activity and professional practice where translation skill is needed (Campbell, 2013). Therefore, it is essential for language learners to receive more training in developing translation skills. 3.

(14) To help learners promote translation skills, there is a need to provide proper training and introduce referencing tools for learners to use while translating. However, in reality, teachers have limited time in classroom for teaching translation skills or allowing students to practice translation in class. For lack of training and practice, students also tend to perform poorly in translation tasks. To compensate for the lack translation practice in classroom setting, there is a need for introducing proper referencing tools for assisting learners in doing translation tasks. Also, a proper bilingual referencing tool is needed when learners are self-learning translation and teachers are not always available to provide immediate feedback. Currently, language learners usually rely on bilingual dictionaries when they do translation tasks. However, even if they are provided with bilingual dictionaries that helps them translate and looking for the expressions they need, they might produce lexical errors, with the aforementioned problems inherent in bilingual dictionaries. With the increasing demand for translation services, computer-assisted translation (CAT) tools have been developed, including spell checkers, electronic dictionaries, concordancers, and machine translation (MT) engines (Federico, Cattelan, & Trombetti, 2012). CAT tools not only benefit professional translators in terms of their productivity and accuracy, but they can also provide assistance to language learners performing translation tasks (Zanettin, 1998). Beginning language learners, due to their unfamiliarity with the target language and relatively lower language proficiency, are likely to rely on Machine Translation tools, such as Google Translate, when they are requested to translate a text into the target language. Unlike bilingual dictionaries, machine translators allow users to query a sentence instead of merely a word or a phrase. However, the translation results provided by a machine translator is based on the correspondence of words, and these tools sometimes overlook a word’s different usage, 4.

(15) which results in poor word choice, collocational and syntactic errors. Worst of all, many beginning and intermediate-level learners may not be able to discern the appropriateness of their wording, hence producing semantically and syntactically unacceptable translation outcomes. As a matter of fact, when it comes to the field of translation, many professional translators use bilingual concordancers to produce more fluent and accurate text in the target language (Bourdaillet, Huet, Langlais, & Lapalme, 2010). Also, corpora have been extensively used in the training of translators. Therefore, researchers suggest that electronic corpora and corpus tools can be employed to train language learners to perform better on translation task (Rodríguez-Inés, 2009). Bilingual concordancers not only provides different meanings of the target word but also provides a compilation of how they are used and interact with other words in different contexts. In the process of reading and analyzing concordancing lines, learners can construct their understanding of the word meaning, learn to observe and discover word usage on their own and apply them in their writing or translation task. Although researchers have tried to stress the usefulness and effectiveness of corpus tools like concordancers, they are rarely used in actual classroom settings. It comes as no surprise that corpus tools and concordancers are underused by language learners. Because online corpus tools are relatively new compared to online dictionaries, most language learners are not familiar with the use of corpus tools (Chan & Liou, 2005). Kilgarriff (2009) indicated that most language teachers, especially those teaching in high schools, have never heard of concordancers, let alone using it and training their students to use it. However, most learners have learned how to use a dictionary since they started to learn a new language, and hence are more used to searching in dictionaries rather than corpus tools. Furthermore, most language learners 5.

(16) find it difficult and stressful to use corpus or concordancers as reference tools because learners may be overwhelmed by the abundance of texts and may not be able to successfully locate the information they need or generate the rules by themselves (Gabel, 2001; Granger & Tribble, 1998; Yoon & Hirvela, 2004). Even if they do use concordancers, it is discovered that they use dictionaries and concordancers for different purposes (Lai & Chen, 2015). Learners usually consult a dictionary when they need information on the meaning of a word, while they use corpus tool when they need to know more about how the word is used. For most learners, their needs can be satisfied with the use of bilingual dictionaries. This may decrease their motivation in using concordancers. Although the usefulness and effectiveness of CAT tools have been recognized, the debate about which tool is the most beneficial and proper one for language learners remains inconclusive. Furthermore, a limited number of studies have compared the effectiveness of different bilingual referencing tools and which tools can benefit lowerintermediate EFL learners the most when doing sentence-based translation task. The aim of the current research was to assess the effectiveness of two different types of bilingual referencing tools and examine whether and to what extent they can help EFL learners to perform on translation tasks.. 1.3 Purpose of the Study and Research Questions As mentioned, bilingual dictionary and bilingual concordancer have been used to assist translation. Nevertheless, few studies have compared and assessed these two bilingual referencing tools and their effect on translation tasks. In order to examine the effect of these two referencing tools on EFL learners’ translation skills, the current study focuses on comparing the effectiveness of the two different referencing tools, bilingual dictionary and bilingual concordancer, and how they assist language learners in 6.

(17) improving their translation. The research questions are as follows. 1. Can the use of bilingual referencing tools improve EFL learners’ translation performance? 2. Can the use of bilingual referencing tools help learners to retain the learned expressions for a period of time? 3. Between the two bilingual referencing tools utilized in the study, which one is more effective in helping students to translate more correctly? 4. What are EFL students’ perceptions toward the bilingual referencing tools?. 1.4 Significance of the Study The present study is conducted to investigate the effects of two different types of web-based bilingual referencing tools on improving learners’ translation performance. The current research attempts to provide more empirical evidence for the existing research literature on the use of bilingual dictionary and concordancer in language learning. It is hoped that the results from the present study allow both language teachers and learners to understand the potentials of applying bilingual concordancers to language learning. Furthermore, with the exploration and comparison of the two referencing tools, the present study aims to offer an alternative tool to compensate for some perceived deficiencies in bilingual dictionaries.. 7.

(18) CHAPTER TWO LITERATURE REVIEW This section provides a review of the literature concerning two major topics relevant to this present study. The first part will discuss how translation and language learning are related. The second part provides a review of how data-driven learning with the help of reference tools have been used to improve translation and promote language learning.. 2.1 Translation and Language Learning To further understand the relationship between translation and language learning, a clear definition must be made. The translation discussed in this research refers to “pedagogical translation” (Klaudy, 2003) and “school translation” (Gile, 1995). Unlike real and professional translation, pedagogical translation denotes a tool for language teachers to improve or to gain information about a learner’s language proficiency (Klaudy, 2003). Gile (1995) added that school translation focuses on the language itself and the correct use of vocabulary and grammatical rules will be strictly examined while professional translation focuses on the content of the translation text instead of linguistic problems. In the following section, the role of translation in language learning and teaching will be firstly discussed. Furthermore, how translation can be utilized to facilitate language learning and what difficulties learners may have in translation tasks will be discussed later. 2.1.1 The role of translation in Language Teaching and Learning In SLA research, studies concerning translation mostly focus on the use of learner’s use of their native languages in teaching and learning. The role of translation in language learning and teaching has been vigorously debated. In the early years of the nineteenth century, the Grammar Translation Method (GTM) had been the dominant teaching method in language learning. GTM focuses mainly on learners’ vocabulary 8.

(19) and grammar and required learners to do considerable rote learning and translation practices. However, it encountered rejection because it failed to enhance learners’ communicative ability (Richards & Rodgers, 2001; Brown, 2007) and hindered “the development of the ability to think directly in the new language” (Rivers & Temperly, 1978, p. 326). With the diminishment of the grammar-translation method and the growing popularity of direct method and communicative language teaching approaches, translation was rejected and degraded in the field of language teaching (Carreres, 2006). Not only translation as a teaching device has met its ups and down, learners’ use of their mother tongue in language learning was also once faced with discouragement. Opponents against involving L1 in L2 acquisition claim that the overuse of L1 may cause interference and negative transfer which will induce more errors in language learning (Kaplan, 1972). However, in learning process, translation behavior cannot be denied or ignored. When reading for comprehension, for example, L2 words are automatically and unconsciously translated into L1 (Thierry & Wu, 2007). From the cognitive linguistic perspective, L2 is conceptually mediated by L1 (Kroll & Stewart, 1994). At the early stage of second language acquisition, when language learners need to produce output in the target language, they resort to their L1 to form meaning and then mentally translate it into target language before writing it down or speak out (Kobayashi & Rinnert, 1992). Only when their language proficiency increases can learners decrease their reliance on mental translation (Sasaki & Hirose, 1996). Moreover, it is inevitable for learners to use translation to help them understand word meaning and memorize vocabulary (Chamot & Kupper, 1989; Liao, 2006, Prince, 1996). Research on the bilingual lexicon has shown that when beginning learners receive L2 input, they will associate it with L1 forms during their processing (Talamas, Kroll, & Dufour, 1999). Although with their language proficiency developing, learners are more 9.

(20) capable of directly mapping meanings to forms in L2, it is inevitable for L2 input to induce L1 associations in advanced learners or even proficient bilinguals (Spivey & Marian, 1999). Based on the findings, Horst, White, and Bell (2010) recognized that L1 certainly resides in L2 learners’ mind no matter how hard language teachers try to avoid using L1 in class. Now that it is impossible for language learners to avoid using translation, some researchers have begun to investigate how and to what extent translation can be applied in language teaching to facilitate language learning. Also, to avoid the negative transfer and interference, some researchers have advocated exposing learners to the language problems they may encounter by saliently showing them the differences between the two languages (Lucas & Yiakoumetti, 2017; Laufer & Girsai, 2008; Nesselhauf, 2003; Phoocharoensil, 2012; Xiao & McEnery, 2006), and that translation practice is one way to do it Kavaliauskienė (2010) claimed that discussing differences and similarities between languages with learners can help them understand their difficulties caused by their native language and raised their consciousness of nonparallel nature of languages. If students are aware of the differences, language interference and intervention from their native language may be reduced. Furthermore, translation activities involve integration of all four language skills, including reading, writing, speaking and listening. Therefore, when translation skill is practiced, these skills are actively involved and developed, especially accuracy and flexibility on communicative aspect. In fact, Kavaliauskienė (2010) regarded translation as a common strategy employed by many learners when doing language learning tasks and as a real-life, naturally-occurring activity that many learners engage in every day either in formal or informal circumstances. Some researchers even recognized translation as the fifth language skill and the most important social skill since it promotes communication and understanding between people (Ross, 2000). Therefore, 10.

(21) Kavaliauskienė (2010) commented that it is natural and logical to develop learners’ skills in translation for the purpose of enhancing language proficiency. 2.1.2 The Application of Translation on Language Learning In recent years, researchers have advocated that translation should not be neglected in language teaching (Carreres, 2006; Cook, 2007). One of the reasons is that translation activities are believed to be effective in enhancing accuracy and clarifying the notion of equivalence (Cook, 1998; Pellowe, 1998). Proponents of contrastive instruction even state that learners’ exposure to cross-linguistic information will lead to learners’ awareness raising and noticing in the difference of two languages, which assist them in difficult grammatical acquisition (Lucas & Yiakoumetti, 2017) and vocabulary learning (Laufer & Girsai, 2008). From their point of view, translation can serve as a way to “make a foreign language feature noticeable or salient in the input” (Laufer and Girsai, 2008, p.697). Laufer & Girsai (2008) investigated whether form-focused instruction on vocabulary items when adopting explicit contrastive analysis and translation as teaching approach can facilitate vocabulary retention when compared with mere meaning-focused instruction. After finding that teaching vocabulary with comparison with L1 and translation task helps learners to perform better in retention test, they propose that translation activities prevent learners from avoid words or structures they have problems with and result in “pushed output” (Swain, 1985). In Laufer and Girsai’s research, the target items also include collocation, which is widely recognized as one of the obstacles that hinder learners from achieving a native-like command of the foreign language. Researchers have also found that most of learners’ collocational errors are derived from the L1 transfer and this phenomenon is observed in beginners and advanced learners alike (Boonyasaquan, 2006; Lin, 2010; Nation, 2001; Nesselhauf, 2003; Phoocharoensil, 2012; Xiao & McEnery, 2006). However, researchers argue that 11.

(22) this does not imply collocation should be taught without L1. On the contrary, Xiao & McEnery (2006) proposed that teachers should make distinction between the collocational behavior in L1 and L2 to enhance learners’ awareness of L1–L2 differences and therefore reduce the number of collocational errors caused by L1 interference. Many researchers have supported this view and proved this approach effective (Nesselhauf, 2003; Phoocharoensil, 2012; Xiao & McEnery, 2006). The research conducted by Laufer & Girsai (2008) also showed that explicitly explaining L1-L2 differences along with specific translation exercise can lead to better collocation learning. The example of collocation learning also helps to illustrate the second reason why translation activities are claimed to be helpful for language learning. Some researchers propose that translation activities enable learners to be aware that they cannot always find equivalences in the two languages. Learners usually attribute their failure in translation and writing to their small vocabulary size. However, Liao (2011) indicated that poor word choice doesn’t necessarily result from small vocabulary size or insufficient grammatical knowledge. Instead, sometimes it is caused by learners’ lack of collocational knowledge. Due to insufficient collocational knowledge and awareness, L2 learners rely on word-for-word translation or literal translation when producing output. However, a L2 word does not necessarily have a clear-cut equivalence in L1 or L2 words are not always semantically overlapped with its L1 correspondence (Jiang, 2000), which results in learners’ poor word choice and lexical errors. This is why many researchers oppose word-for-word translation and suggest that an idea-to-idea translation should be applied (Weschler, 1997). Liao (2011) believe that translation practice can be effective in collocation learning because learners can learn to translate by a phrase rather than by a word. He also suggested that teacher can design translation 12.

(23) practices based on learners’ common collocation errors so that learners can perceive that the perfect equivalence does not exist as they correct and revise their translation. Moreover, Liao (2011) also proposed that corpus tools can be applied when learners do translation practices because corpus tools provide a large number of actual collocations used in authentic context, which can assist learners in reducing their weird-sounding word combinations. Other researchers also confirmed that translation practice plays an essential role in language teaching and learning (Malmkjar, 1998; Zanettin, 1998). In addition to lexical translation, translation practice enables learners to attend to and familiarize the structural similarity and difference in different languages. Researchers also maintained that translation training can promote the understanding of the source language, and that translation as testing device can reflect how much and whether learners can use the language (Newmark, 1988). Unlike free writing tasks, translation tasks specify the structures and the vocabulary items in a given context, so learners cannot avoid making errors if they are not familiar with the target sentence patterns and vocabulary items (Källkvist, 1998). Based on the theory of Error Analysis, learners’ errors allow teachers to know the areas learners have problems with so that they can focus on the area to help learners improve. Nevertheless, translation as a testing device has also raised some concerns. To lower the difficulty in marking and avoid potential dispute, many translation exercises or tests employ blank-filling or lexical translation items instead of sentence-based translation. Learners’ performance on this type of translation tests cannot fully represent learners’ knowledge of word-choice and syntactic structure. Furthermore, in order to test a target linguistic item, translation questions are usually composed of unnatural and specially-arranged language. Learners can notice what specific linguistic items are 13.

(24) being tested and therefore write down the correct forms in these gap-filling blanks. However, in real communication situation, they still fail to use these words or forms correctly (Matthews-Břeský, 1972). In short, researchers believe that translation practice and questions are merely a part of writing activity or one of the approaches to testing one’s writing ability and that they can never replace other writing tasks which help to develop learners’ logic and creative thinking. In addition, since the results of translation tests cannot thoroughly reflect learners’ writing ability, researchers suggest that other production tests should also be included together with translation tests so that examiners can assess learner’s different aspects of language proficiency. 2.1.3 L2 Learners’ Difficulties in Translation To gain more insight in learners’ difficulties in translation for pedagogical purposes, researchers have observed learners’ translation patterns and analyzed the errors made by language learners in translation tasks. Wu (2014) investigated Taiwanese senior high school students’ errors and difficulties in translation questions, and observed that the errors senior high school students generally made in Chinese-toEnglish translation questions included passive voice, article, quantifier, conjunction, relative clause and conversion of parts of speech. Aside from grammatical errors, many of learners’ errors are attributed to the influence of question keyword in L1 and their lack of collocational knowledge. She found that both high and low-achieving students heavily relied on Chinese keyword and failed to apply their existing vocabulary to their translation, and she claimed that this resulted from learners’ deficiency in a word’s semantic knowledge and its usage. For most language learners trying to know a word, the easiest way is to look up the target word in dictionaries, mostly bilingual ones, and read the first one or two translations listed, and memorize the word form and its translations together. However, one word can have many different meanings in different 14.

(25) contexts. When the keyword in translation questions does not correspond with the L1 translations in their mental storage, learners will be unable to retrieve the information and use the target word correctly (Wu, 2014) On the other hand, Wu (2014) also found that students’ lack of collocational knowledge contributed to many of students’ errors in translation. When required to produce word combination, learners tend to resort to their L1 knowledge and adopt literal translation strategy due to their lack of collocational knowledge (Fan, 2009), thus producing weird combinations. In fact, previous findings have shown that learners’ collocation production is greatly influenced by their L1 transfer (Fan, 2009; Nesselhauf, 2003; Phoocharoensil, 2013). Nesselhauf (2003) explored the use of verb–noun collocations by advanced German learners of English in free writing and investigated the causes of learners’ improbable collocation production. She stressed learners generally had hard time acquiring L2 collocations that were incongruent in their native languages and these collocations should be the focus of teaching. In addition, Liao (2011) investigated the influence of collocation on college-level language learners’ Chinese-to-English translation and confirmed that there existed positive correlation between learners’ knowledge of and their translation performance. He also identified the most frequent miscollocation types in translation were verb-noun and noun-verb collocations, which corroborated previous research findings (Chen, 2002; Nesselhauf, 2003; Tang, 2004) whose analyses were based on college learners’ free writing performance. According to Nesselhauf (2003), language learners had difficulty using correct collocation involving verbs because of the restricted sense in the verb. Verbs with more restricted sense in collocations preclude substitutability of its nearsynonyms, while some verbs in collocations had less restricted sense and if learners cannot distinguish the subtle difference in the synonymous verbs, they will produce 15.

(26) collocational errors. To improve learners’ translation competence and reduce learners’ collocation errors, Liao (2011) proposed that learners should be trained to explore authentic translations or parallel texts in corpus and concordancers where learners can compare and contrast the different word combinations and collocations in the two languages, thus generalizing the rules on their own. Reynolds (2016) also encouraged teacher to adopt translation as a technique for teaching collocation to raise students’ collocation awareness in their L1 and L2 and to increase the use of L2 collocations in students’ writings.. 2.2 Language Learning and Reference Tools When learners need to comprehend a text, to write in target language or to translate a text, they usually need to consult bilingual reference tools, including dictionaries, machine translators and concordancers. The potential of these reference tools is attracting widespread interest in fields such as language learning and translation. Especially in past decades, the use of corpora in language teaching and translation has drawn much attention, which has also given rise to data-driven learning (DDL) (Johns, 1991). DDL refers to a learning approach where learners are considered as researchers who can explore natural linguistic data in corpora or concordancers. In the process, learners are able to scrutinize the data that are relevant and useful to them to find linguistic patterns and induce rules on their own, which is considered to facilitate learner autonomy because they are no longer passively given the rules by teachers (Boulton, 2009; Boulton, 2010). Thanks to the advancement in technology and the ubiquity of computers, researchers can build and develop computer-based corpora and make these tools accessible to both teachers and learners. To gain more understanding of the potential of DDL, researchers have applied 16.

(27) DDL to language teaching and have proved its effectiveness in grammar and vocabulary learning (Benavides, 2015; Gabel, 2001; Guan, 2013; Oghigfian & Chujo, 2010; Yılmaz & Soruç, 2015). Furthermore, with the rising interest in collocation learning which is considered a great challenge for language learners, data-driven learning has been proposed to tackle learners’ collocation difficulties because corpora can provide a large number of authentic texts showing how the target word is used in context and what words it is used with, especially in the form of concordancers (Boulton, 2010). Therefore, the use of corpus tools in helping learners to learn accurate and appropriate English collocations has been extensively investigated (Chan & Liou,2005; Kheirzadeh& Marandi, 2014; Rezaee, Marefat, & Saeedakhtar, 2015). On the other hand, since the use of parallel or multilingual corpora and concordancers have been widely accepted by translators for decades, researchers have also begun to probe into whether data-driven learning can benefit language learners’ translation competence. At the beginning stage of investigation on data-driven learning, monolingual concordancers have been extensively adopted in the research that attempted to examine the implementation of data-driven learning in the classroom and the effectiveness of data-driven learning have been established. Boulton (2009) conducted an empirical study to compare the use of materials from corpus and conventional reference as a learning source and reference. The results showed that corpus data seemed to be more effective in helping learners to deal with language problems than traditional ones from bilingual dictionary entries or usage manual. However, drawbacks of the corpus tools have also been perceived and these drawbacks and the unfamiliarity to the tools have deterred many teachers from utilizing these tools and adopting the approach in the classroom. First of all, asking learners to read and analyze the data can be timeconsuming and a large number of sentences may intimidate and frustrate learners, 17.

(28) especially for beginning-level learners (Gabel, 2001; Granger & Tribble, 1998; Yoon & Hirvela, 2004). It has been suggested that DDL in class should not last too long, and different tasks must be adopted to prevent the data analysis work from being too mechanical and tedious (Allan, 2006; Whistle, 1999). Second, sometimes the chosen corpus did not tailor to learners’ needs and proficiency level. The languages contained in the corpus may be too challenging for beginners or too general for advanced learners who need information in specialized fields (Frankenberg-Garcia, 2005). Oghigian and Chujo (2010) advised that teachers should take into consideration the size, relevance and accessibility of the corpus when choosing a corpus appropriate for learners. For example, an immense corpus may have a risk of intimidating learners with huge amount of information, while a corpus with too small data may not present sufficient example sentences. The written language in newspaper may be less technical than academic journals, but it still pertains to formal language and is less colloquial than language in magazines. Introducing an inappropriate corpus to learners may hinder learners from retrieving the information they need and result in learners’ resistance to adopting datadriven learning approach. In addition, not all learners can be accustomed to this self-research, learnercentered learning process and develop learner autonomy. In Reynolds’ study (2016), 25 Taiwanese medical students were encouraged to exploit a web-based English/Chinese parallel collocational concordancer, TANGO, to self-edit their essays for verb-noun errors. Although the results demonstrated that the adoption of a web-based collocational concordancer largely decreased learners’ collocation errors, especially for descriptive writing, students expressed negativity toward being asked to self-edit their essays with a software and their difficulty in determining which queries to input in order to obtain desired results due to their lack of collocational knowledge. 18.

(29) Other researchers also expressed concerns regarding data-driven approach and corpus tools. In Wang’s research (2001), he claimed that concordancers help learners to develop in-depth lexical knowledge and compared data-driven learning to inductive teaching. However, he noted that these two approaches were distinctive in the way that the materials in data-driven learning were “primary, and the teacher does not know in advance exactly what rules or patterns the learner will discover. (p.182)” This implied that learners might be misled by the concordancing results if they were not guided properly. Despite all the concerns above, data-driven learning still attracted much research interest. The emergence of parallel concordancers encouraged more practitioners to test the tools and implement them in translation studies, the development of bilingual lexicography, machine translation and language learning. Researchers indicated that bilingual texts in parallel concordancers alleviate learners’ anxiety in reading through abundance of sentences and boost learning efficiency. In addition, bilingual concordancers allow users to query in both source and target languages based on their need. This means users can retrieve information they need even if they do not know the expression in L2, which is the most distinctive feature that sets bilingual and monolingual concordancers apart (Gao, 2011). Moreover, parallel concordancers enable users to compare the two languages and notice the differences between the two languages (Chujo et al., 2005). When users query a term in their native languages, parallel concordancers may show different ways to express the same idea in target languages as well as the context in which these equivalent terms are used. In this way, learners are able to tell the subtle differences between these L2 expressions and to decide the appropriateness of these terms in different contexts (Frankenberg-Garcia, 2005). With the assistance of parallel concordancers, data-driven learning has been 19.

(30) proved to be accessible and effective for beginning level learners (Chujo, Utiyama, & Miura, 2006; Chujo & Oghigian, 2006). Nonetheless, it was noted that parallel concordancers should not be considered as a panacea for every language problems and learning contexts. As indicated in Frankenberg-Garcia (2005), parallel concordancers consist of corpora that are smaller than the ones constitute monolingual concordancers and can only represent only a part of language because parallel corpora contain only the texts that have been translated. It means that it is highly possible that users cannot find words or expressions they need in these comparatively small corpora. Furthermore, if the class is composed of learners of different native languages, the introduction to a bilingual concordancer is likely to serve no useful purpose (Frankenberg-Garcia, 2005). As a matter of fact, FrankenbergGarcia (2005) pointed out that it was unnecessary to retire monolingual or bilingual concordancers because they play “non-conflicting, complementary roles (p.191)” in language learning. Under certain circumstances, a monolingual concordancers is more useful than a bilingual one, and vice versa. With proper guidance and in appropriate contexts, data-driven learning with the assistance of different corpus tools can surely help learners to achieve better learning outcome. To date, the majority of the research concerning data-driven learning has focused on monolingual concordancers because parallel concordancers are relatively new tools. Only a comparatively small body of research has tackled with the usefulness of parallel concordancers in different contexts. On the other hand, when it comes to most accessible bilingual reference tools, bilingual dictionaries and machine translators account for a large proportion of the of total user accesses. The following sections will examine the effectiveness of different reference tools and how they assist language learners. 2.2.1 The Effects of Bilingual Dictionary on Translation and Language Learning 20.

(31) Since translation is suggested to be a part of language teaching and learning, researchers have discussed how and with what materials or tools translation should be taught. With rapid development of technology, various types of technology-based referencing tools have been used to help translators and language learners to accomplish translation tasks, including online dictionary, machine translators and translation memory. And researchers also suggest that these referencing tools can be utilized in the language teaching to promote language learning (Chen & Yu, 2013; Yepes, 2011; Zanettin, 1998). Among these referencing tools, bilingual dictionary is one of the most popular and easily-accessed tools (Levy & Steel, 2015). For many language learners, when encountered unknown words, they tend to consult dictionary of any kinds, especially electronic dictionary and online dictionary by virtue of their convenience. Although bilingual dictionaries were often criticized for limited example sentences owing to space limitations, especially in hardcopy dictionaries, many electronic dictionaries nowadays eliminate the constraints (Kilgarriff et al., 2014), such as Cambridge Dictionary and Bing Dictionary. Many web-based dictionaries have started to offer many new features. Aside from additional example sentences, they also provide word derivation, audio pronunciation, visual aid and even the hyperlink features that allow users to click any word as an entry to search for the word information. Furthermore, the positive effects of online bilingual dictionaries as a writing aid for improving EFL learners’ writing quality have been established, especially in the aspects of vocabulary use and accuracy (Lew, 2016). Laufer and Hadar (1997) compared three types of dictionary and their effect on learners' comprehension and production. In their study, they found that although bilingualized dictionary, which contains translation and definition in both L1 and L2, 21.

(32) did help learners of different levels of language proficiency to perform better, bilingual dictionaries would be more appropriate for novice dictionary users. They pointed out that the monolingual parts in bilingualized dictionary were often neglected by unskilled dictionary users and that average users would consult monolingual definition to help them comprehend the word meaning but it did not help them to produce L2 output better. Thompson (1987) also indicated that monolingual dictionary might be misleading and confusing for language learners, especially those at beginning level, and that bilingual dictionary might provide more suitable assistance for language learners. With bilingual dictionary, learners can search for the meaning of the target word and its translated equivalence in their L1. Meanwhile, they can also query in their L1 and search for the possible translation in L2. However, the translations provided by the dictionary does not end learners’ quest. It is highly possible that bilingual dictionary usually provides more than one L2 translations, which might lead learners to use the inappropriate one because learners may not know the distinction among these L2 translations. Another problem was that “a simple one-word translation can be misleading, especially when there are semantic incongruencies between the two languages” (Laufer and Hadar, 1997, p.189). Because learners usually assume that a word is equal to its translation in every way, they will combine words the way they do in their L1 and produce unacceptable word combination. Nevertheless, many online dictionaries nowadays have revolutionized their interfaces and contents. For example, most online dictionaries provide audio pronunciation for users to learn how to pronounce the word. Furthermore, since they no longer have limited space on the webpage, many online dictionaries begin to list more example sentences, and even other information. Yahoo Dictionary, for example, provide collocations and frequently-used phrases for users to understand some of the 22.

(33) word usages. In brief, the potentials of these new functions provided by modern online dictionaries are still yet to be explored. 2.2.2 The Effect of Parallel Concordancer on Language Learning Copious research has illustrated several benefits of the use of concordancers in different aspects of language teaching and learning. However, most of the previous research examined the effectiveness of monolingual concordancers, and only a limited number of research studies explored the potentials of parallel concordancers due to the difficulty in compiling parallel concordancers. Bilingual concordancers can come in handy when learners have no idea how to express an idea in L2, while monolingual concordancers are more useful when learners want to confirm the correctness and appropriateness of a certain expression. More details on the differences between monolingual and bilingual concordancers have been illustrated in Section 2.2. Since one of the reference tools targeted in the current research is parallel concordancer, this section will focus on the impacts of parallel concordancers on language learning. To further understand the use of parallel concordancers in language learning, an increasing number of studies have probed into the use of parallel concordancers in different contexts and a variety of aspects of language learning, including vocabulary, collocation, grammar, writing and translation. Parallel Concordancers and Vocabulary and Collocation Learning Mastering a sufficient amount of vocabulary has always been a cognitive challenge for language learners (Zimmerman & Schmitt, 2005). It has been suggested that varied exposures to new words occurring in meaningful contexts may facilitate the process of vocabulary learning (Myers & Chang, 2009). Therefore, researchers have proposed to involve corpora in vocabulary learning since one of the most significant characteristic of a corpus is that it can provide a large number of example sentences, showing how 23.

(34) the word is used in different context. Moreover, concordancers usually present the keyword in a format that allow users to do rapid scanning and comparison, which is referred to as keyword-in-context (KWIC) (Johns, 1991). Sentences may be truncated and target word may be highlighted in a certain way in KWIC format. One way of keyword enhancement in parallel concordancer is alignment. Alignments allow users to observe the existence of the equivalent text segment. Learners can easily spot the target equivalents of the source word, also allowing them to know the meaning. However, since different languages have different syntactic characteristics, alignment provided may not fully represent the “precise” equivalents of the source language. In addition, some argued that target word presented in “unfinished sentences” may be confusing, especially for beginners (Yoon & Hirvela, 2004). However, Boulton (2009) actually discovered that in his experiment, lower-intermediate learners provided with reference in KWIC format performed better than in full sentence contexts. In addition, concordancers can provide information related to the keyword, such as collocation, derivation and word frequency (Flowerdew, 1996). With these features, concordancers are believed to be beneficial in facilitating vocabulary learning. Wang (2001) developed a system called English-Chinese Parallel Concordancer which was aimed for intermediate learners and investigated whether data-driven learning with the support of his tool can facilitate lexical learning. Eight college students learning Chinese were recruited in the research, and they were asked to analyze the meaning and use of a Chinese adverb in the parallel concordancing output. In the study, participants were reported to perceive the target word could be used in certain structures and contexts and to induce the rules of the word usage on their own. Wang (2001) thus discovered that learners could acquire in-depth knowledge of lexical meaning and use by closely examining the texts. In the process of comparing the 24.

(35) languages in the data, learners realized no exact corresponding translations existed between the two languages. Based on his findings, Wang (2001) suggested that parallel concordancers were promising tools to fulfill the aims of data-driven learning. One of the limitation of corpus application is that it is difficult for beginning level learners to uncover the linguistic regularity hidden behind the enormous number of concordance lines (Gabel, 2001; Granger & Tribble, 1998; Yoon & Hirvela, 2004). In addition to being overwhelmed by the long list of retrieved example sentences, beginning level learners also have problem understanding the texts containing advanced vocabulary, let alone analyzing it. However, in John’s pioneering pilot study (2001) in which a beginning German learner was asked to answer lexical and grammatical questions via a parallel corpus and concordancer, he found that beginning level learners can benefit from parallel corpus tools even in an unsupervised environment. The researcher observed that the translation in learner’s native language was proved to be helpful in his comprehension and analysis process. Also, the translation not only allowed the learner to realize that one-to-one equivalent did not always exist, but it also helped him learn new vocabulary. In addition, the researcher found that the learner adopted a strategy to only select the sentence that contained no complex structures. Although the researcher noted the learner’s deduction might not be reliable without considering and analyzing all data, this still meant that the learner was able to develop a learning strategy and be more selective and aware of the data he needed to helped him reach a certain conclusion on his own. Finally, considering a parallel concordancer as a supplement for teaching, the researcher supported the view that corpus and concordancer can facilitate learner motivation because the learner gained more confidence by finding answers for themselves. Although the participant in John’s study (2001) was at beginner’s level learning 25.

(36) German, he was a college student majoring in linguistics. Also, there was only one subject in his study, which might not be a valid evidence for proving the usefulness and applicability of parallel concordancer for beginners. To examine beginning level learners’ use of parallel concordancer on a bigger scale, Chujo et al. (2006) carried out a study involving 72 non-English major beginning level EFL students. The researchers utilized a self-designed Japanese-English parallel corpus program integrating parallel news-based corpora into CALL-based activities, and students’ evaluation and response to the tool were collected. The researcher found that although corpus-based activities were generally difficult for beginning level language learners, parallel corpus eliminated the language barrier, enabling learners to compare the vocabulary and language patterns between two languages. In the process, students not only notice the collocation pattern but they also learn new words. The learners were also discovered the word can be translated into different meanings in another language, identifying the one-to-many relationships between two languages. In addition, it should be noted that in the researchers’ attempt to explore the optimum interface features, they found that several user-friendly features provided by concordancers were helpful in comparing and analyzing the texts, such as sorting the concordance lines in a certain order, highlighting the accurate translations and displaying concordance lines in KWIC format. Although the sample size in the study by Chujo et al. (2006) was sufficiently large, these subjects were actually at college or graduate level engineering students. This means these students had higher level of cognitive development and computer skills than students from senior or junior high schools. The researchers found that they became used to the concordancer very soon, and this was probably because these learners were familiar with the computers. Furthermore, the study failed to provide an empirical evidence for the effect of the parallel concordancer. There remains a need to 26.

(37) examine the influence of parallel concordancers on younger students’ language performance. Yu, Liou, Chang, and Vongpumivitch (2011) designed a Chinese-English specifically for primary school students and carried out a study to examine the effectiveness of web-based concordancer on children’s vocabulary learning. The researchers constructed a corpus containing 10211 sentences modeled from sentence patterns in nine elementary English textbooks. After the pretest that confirmed the participants displayed no knowledge of the six target words, a total of seven fifthgraders were asked to take part in the twelve 30-minute sessions of concordancing activities over six weeks. In each session, students searched in the concordancer for one unknown word’s meaning, usage, and the example sentences. Afterwards, they were asked to complete one or two translation sentences that were adopted from the concordancing output. One week after the treatment, they received two posttests and an interview. Four weeks after the posttest, they were given a delayed posttest to observe their retention of the word knowledge. The results showed that all participants made significant progress in the posttests. Although their scores of all the aspects in the delayed posttest regressed, there existed no significant difference between the scores during the treatment and the delayed posttest. The findings suggested that even elementary school beginning level EFL learners could benefit from the parallel concordancer as they induced the usage of the target words based on the concordancing output and used them to make sentences. Also, the results echoed with Chan and Liou’s study (2005): the residual effects of concordancing could be maintained over a period of time. However, although the participants commented that the parallel concordancer was a meaningful reference for language input, they were encountered several difficulties when using the tool, including a limited number of sentences and unfamiliar vocabulary in concordance output. With regards to these problems, the researchers 27.

(38) suggested that the teachers’ intervention should be involved to improve the effectiveness of concordance-enhanced language learning. Although the research conducted by Yu et al. (2011) demonstrated that integrating concordance-based vocabulary instruction was beneficial to elementary school beginning learners because it could empower learners to discover and induce vocabulary knowledge, the sample size in the study was too small to perform contrastive analyses between control and experiment groups. On the other hand, since the significance of collocational knowledge has been established and collocation has been considered fundamental in vocabulary learning (Nation, 2001; Nesselhauf, 2003; Lewis, 2000; Hyland, 2008), several researchers have proposed the use of corpora and concordancers in collocation teaching and learning to help learners have a native-like command of collocations (Woolard, 2000; Conzett, 2000). Therefore, studies have been done to investigate how to use online concordancers to raise learners’ collocational awareness. Many research studies have emphasized that it is helpful to immerse learners in the abundant authentic language input provided by corpora (Chan & Liou, 2005; Chang, 2004; Chao, 2010; Chen, 2011; Sun & Wang, 2003; Wu et al., 2010; Wang, 2013). However, since low to intermediate level learners might be overwhelmed by the number of concordance lines, some researchers have explored the potential of parallel concordancers in collocation learning to reduce the language burden of difficult L2 words. For example, Chan and Liou (2005) compared the effects of utilizing web-based practice units on English verb–noun collocations with and without the help of TotallRecall, a web-based Chinese–English bilingual concordancer, on EFL learners’ collocation learning. TotalRecall was based on the corpus of Sinorama Magazines (written in bilingual languages) and records of the Hong Kong Legislative Council, containing over two hundred thousand bilingual 28.

(39) sentences. Only three of the five units focusing on V-N lexical collocations involved the use of the concordancer, while the other two units were taught with pattern explanation and a few example provided. Along with the units, participants were also asked to complete the interactive exercises, including multiple choice, Chinese–English sentence translation, and gap-filling sentences. Thirty-two non-English major college students’ collocational knowledge was measured with a pretest, an immediate posttest and a delayed posttest three months later. Before the treatment, most participants were found to have little awareness in collocation. The progress in the immediate posttest results showed students’ overall V–N lexical collocation knowledge has been significantly enhanced, and even though learners’ degree of retention of collocations decreased as shown in delayed posttest, the final performance was still better than the students’ pretest performance. Also, the results demonstrated that the learners improved more on the collocations taught with the help of the online concordancer. The researchers suggested that reading abundant examples of de-lexicalised verbs in concordances and mutual translations in the bilingual concordancer have helped students acquire de-lexicalised verbs and collocations that lack direct translations in another language (Chan & Liou, 2005) In order to explore the influence of bilingual concordancer on beginning level learners, Wang (2013) compared the use of the two web-based Chinese-English bilingual referencing tools, JuKuu, a parallel concordancer and Bing Dictionary, a bilingual dictionary, among junior high students in collocation learning. 63 ninth graders in the EFL junior high school were was randomly divided into “JuKuu group” and “Bing Dictionary group”. Two groups of participants utilized the two online bilingual tools respectively to complete the collocation retrieval task. A week later, all the participants took a posttest. The improvement shown in the task scores indicated 29.

(40) that both the online bilingual tools helped junior high students find proper collocates and facilitated their collocation learning even without teachers’ guidance, and it also indicated that students made significantly more progress with the help of the bilingual concordancer. However, their collocational knowledge regressed later in the posttest and there existed no differences between two groups. The findings were consistent with the research by Chan and Liou (2005) where learners’ degree of retention of collocations decreased after the treatment. The researcher suggested that this might result from the short exposure to the collocates and that teacher could design some practices so that learners could store collocation input in the long-term memory (Wang, 2013). Although previous research has not studied in depth whether providing learners with longer or repeated exposure to target words via concordancers will help learners attain better retention of knowledge due to limited experiment duration, these studies have proven that teaching collocations to low-level learners via explicit approaches with a parallel concordancer can enhance learners’ vocabulary production. Parallel Concordancers and Grammar Learning To date, the debate of whether data-driven learning facilitates grammar learning remains unsettled because a limited number of studies have addressed the issue, some of which even yielded unsatisfactory findings. For instance, Vannestål and Lindquist (2007) undertook a semester-long project that engaged college learners in corpus consultation for grammatical rules. A monolingual corpus was adopted as a complement to grammar textbooks and a reference in problem-solving assignments. Learners were asked to answer the questions about grammatical rules based on the concordancer output after typing in the queries that had been formulated for them. Based on the findings, the researchers indicated that the improvement at the grammatical and sentence level was limited and that some problems arose during the process. For 30.