以語料庫為本之近似詞教學成效之研究:以台灣大學生為例 - 政大學術集成

全文

(1)國立政治大學英國語文學系碩士班碩士論文. 指導教授: 張郇慧博士 Advisor: Hsun-huei Chang, Ph.D. 以語料庫為本之近似詞教學成效之研究:以台灣大學生為例政治. 大. 立. The Effect of Teaching Near-synonyms to Taiwan EFL. ‧ 國. 學. University Students: A Corpus-based Approach. ‧. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. 研究生: 陳聖其撰 Name : Sheng-chi Chen 中華民國 103 年 1 月 January, 2014. v.

(2) The Effect of Teaching Near-synonyms to Taiwan EFL University Students: A Corpus-Based Approach. A Master Thesis Presented to Department of English,. 政治大. National Chengchi University. 立. y. ‧. ‧ 國. 學. Nat. io. sit. In Partial Fulfillment. n. al. er. of the Requirements for the Degree of. C Master of Arts hengchi. i n U. by Sheng-chi Chen January, 2014. v.

(3) 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v.

(4) ACKNOWLEDGEMENTS I am very grateful to many people that have contributed in indirect or direct ways to this thesis. First, I would like to express my gratitude to my dear advisor, Dr. Hsun-huei Chang for her patience, encouragement, and guidance throughout the research process. During the thesis writing, she provided me with clear directions and. 政治大 Gratitude is also extended to my committee members, Dr. One-Soon Her and Dr. 立. revised my thesis. I really appreciate what she has done for me as an inspiring teacher.. Shu-hui Chen, for providing me with valuable suggestions to help me improve the. ‧ 國. 學. quality of the thesis. In addition, my special thanks go to my senior Claire who helps. sit. y. Nat. help.. ‧. find the participants in my research. This study would not be completed without their. io. er. Fourth, I would like to express my indebtedness to Professor Siaw-fong Chung. I learned a lot of corpus knowledge and techniques from her courses. Her work in. al. n. v i n C h language learning corpus linguistics and second vocabulary inspired me for this thesis. engchi U I owe most of my skills as a researcher to her.. Next, I would like to thank my classmates in NTUE and NCCU. With their generous assistance and encouragement, my graduate school life was enriched. Moreover, I also want to express my gratitude to my friends, Winona and Zack, who help me check my thesis grammars and sentences. Finally, my deepest gratitude goes to those whose influence may not be so evident but who nevertheless provided the educational and emotional foundation without which this thesis would have never been possible: my parents. Their love and faith gave me power and strength to go through every difficult time. I love you all. iii.

(5) TABLE OF CONTENTS Acknowledgements…………………………………………………………………...iii Chinese Abstract…………………………………………………………….................x English Abstract……………………………………………………………………...xii Chapter One: Introduction....……..…………………………………………………1 1.1 Background and Motivation………………………………………………….1 1.2 Significance of the Study..……………………………………………………4. 治政 Chapter Two: Literature Review…………………………........................................7 大立 ‧ 國. 學. 2.1 Data-driven learning (DDL) in a classroom………………………………….7 2.2 Concordance lines and Data-driven learning.................................................11. ‧. 2.3 Related studies on data-driven learning…………………………………….14. sit. y. Nat. 2.3.1 Vocabulary…………………………………………….......................14. n. al. er. io. 2.3.2 Collocation…………………………………………………………..15. i n U. v. 2.3.3 Grammar……………………………..………………………………15. Ch. engchi. 2.4 Synonym and near-synonym in language learning and teaching…...............16 2.5 Chapter Summary…………………………………………………………...18 2.6 Research Purposes and Research Questions…………………......................19 Chapter Three: Methodology…………….……………..........................................21 3.1 Research Design…………………………………………………………….23 3.2 Research Procedure and Duration…………………………………………..26 3.3 Setting and Participants……………………………………………………..31 3.4 Experimental Instruction Design……………………………………………33 iv.

(6) 3.4.1 Teaching Content…………………………………………………….33 3.4.2 Teaching Activities…………………………………………………..35 3.4.3 Teaching Procedure………………………………………………….37 3.5 Control Instruction Design………………………………………………….39 3.6 Instruments………………………………………………………………….41 3.6.1 Background Questionnaire…………………………………………..41 3.6.2 Self-Designed Vocabulary Test….......................................................42 3.6.3 Student Feedback Questionnaire……………………………………42. 政治大 3.7 Chapter Summary ………………………………………………………......44 立 3.6.4 Semi-Structured Student Interviews....................................................43. 3.7.1 Quantitative Data…………………………………………………….44. ‧ 國. 學. 3.7.2 Qualitative Data……….......................................................................44. ‧. Chapter Four: Results and Discussions………………..…………………………..47. sit. y. Nat. io. er. 4.1 Pretest……………………………………………………………………….48 4.2 Posttest…………………………...................................................................49. al. n. v i n C hAchievement Score……………………………..49 4.2.1 Learners’ Learning engchi U 4.2.1.1 Paired samples t-test………………………………………….49 4.2.1.2 Independent sample t-test…………………………………….51 4.2.1.3 Summary of result……………………………………………53 4.2.2 Questionnaire results: Learners’ Perceptions………………..............55 4.2.2.1 Learners’ attitude: DDL Learning Value……………..............55 4.2.2.2 Learners’ attitude: Hands-on Activities………………………61 4.2.2.3 Learners’ attitude: Teacher’s teaching value…………………66 4.2.3 Qualitative Findings: Interview…………...........................................67 4.2.3.1 High Achievers: prefer doing tasks with Chinese translation v.

(7) Chinese……………………………………………………….68 4.2.3.2 Low Achievers: prefer challenging tasks….............................70 4.2.3.3 Summary of Interview………………………………………..73 Chapter Five: Conclusions and Implications…………………………………….. 75 5.1 Summary of the study………………………………………………………75 5.2 The Major Findings…………………………………………………………76 5.2.1 Data-driven learning approach improved participants’ vocabulary. 政治大 5.2.2 Data-driven shows different impact on participants’ proficiency…...76 立. knowledge…………………………………………………………..76. 5.2.3 Students’ Positive Reactions...............................................................77. ‧ 國. 學. 5.3 Pedagogical implications………....................................................................78. ‧. 5.3.1 Adaptive Data-driven Learning Approach in English Near-synonyms. sit. y. Nat. learning……………………………………………………………...79. io. er. 5.3.2 Instructional Activities and Design of Lesson.....................................79 5.4 Limitations and Delimitations………………………………………………82. al. n. v i n 5.5 Suggestions for future C study……..................................................................84 hengchi U. REFERENCES……………………………………………………………………..86 APPENDICES…………………………………………………………………….95 Appendix A: Pilot study-lesson plan for corpus-based data-driven learning......95 Appendix B: Concordance lines handout: idioms and phrases…………………99 Appendix C: Teaching materials of the control group………………………...103 Appendix D: Vocabulary Acquisition Questionnaire………………………….106 Appendix E: Pre and post-test…………………………………………………108 Appendix F: Chinese Version of learners’ perception in corpus-based data-driven vi.

(8) learning questionnaire……………………………………………113 Appendix G: English Version of learners’ perception in corpus-based data-driven learning questionnaire……………………………………………116 Appendix H: Sample of group interview Questions…...……………………...119. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. vii. i n U. v.

(9) LIST OF TABLES. Table 3.1 Idioms and Phrases in Three Near-synonyms………………………….….22 Table 3.2 Timetable of the Study…………………………………………………….30 Table 3.3 Number of students in Each Subgroup…………………………………….32 Table 3.4 Format of Data-driven learning activity…………………………………...32 Table 3.5 Near-synonyms List of the Study………………………………………….34 Table 3.6 Teaching Activities and Timetable of the Study…………………………..36 Table 3.7 The Selective Procedure of ‘people’………………………………………40. 政治大 Table 3.8 Five Senses of ‘people’ in Selected Dictionaries…………………………..41 立 Table 3.9 Research Measurement and Statistic Methods Corresponding to Research. ‧ 國. 學. Questions…………………………………………………………………..45. ‧. Table 4.1 Independent sample T-test of pretests……………………………………..48. sit. y. Nat. Table 4.2 Paired sample T-test in the two groups……………………………………49. io. er. Table 4.3 Independent sample T-test of posttest of high achievers in the two groups……………………………………………………………………...51. al. n. v i n C h of posttest of lowUachievers in the two groups..51 Table 4.4 Independent sample T-test engchi. Table 4.5 Results of Learners’ attitude on Data-driven learning Value……………...60. Table 4.6 Results of Learners’ Feedback on Language Learning Value…………….62 Table 4.7 Results of Learners’ Feedback on Teacher’s teaching Value……………..67. viii.

(10) LIST OF FIGURES Figure 2.1 Pedagogic continuum from product to process grammar learning through DDL………………………………………………………………………..8 Figure 2.2 Example ‘people’ for manual concordance lines (extracted from BNC web Corpus)…………………………………………………………………...13 Figure 3.1 Research Structure………………………………………………………..24 Figure 3.2 Research Procedure……………………………………………………….29 Figure 4.1 Pre- and Post-test Scores………………………………………………….53. 政治大 Figure 5.1 Apple Daily English 立 Column……………………………………………..80 Figure 4.2 Pre- and Post-test Results of Progress Gains……………………………..54. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. ix. i n U. v.

(11) 國立政治大學英國語文學系碩士班碩士論文提要. 論文名稱：以語料庫為本之近似詞教學成效之研究: 以台灣大學生為例指導教授：張郇慧博士. 立. 研究生：陳聖其. 政治大. ‧ 國. 學. 論文提要內容：. y. Nat. 教學，致使學生無法於新的情境靈活使用字彙。. ‧. 台灣英語教育多以考試取向為主，許多教師進行英語字彙指導時採用填鴨式. er. io. sit. 本研究旨在於探究以語料庫為本之教學對於台灣大學生在英語近似詞學習成效的影響，以台北市某一所大學 86 位英語學習背景及能力相似之大一生為研. al. n. v i n 究對象。研究人數均分成兩班進行教學實驗，一班為實驗組，以資料觀察法進行 Ch engchi U. 教學，另一班為對照組，以傳統形式教學為主，每週一次五十分鐘，共進行十週。資料蒐集包含近似詞學習成就測驗前、後測，並且依據研究對象於實驗教學結束後接受語料觀察教學法回饋問卷，蒐集研究對象對於語料觀察法之反應與感知，進行量化分析。最後，透過訪談高分組和低分組學生，蒐集其質性資料進行研究探討哪些因素會影響不同英語能力學生對於資料觀察法的意願與需求。本研究發現如下:. 一、近似詞教學有助於提升台灣大學生的英語字彙能力。兩組教學均在後測有. x.

(12) 進步。但就後測成績來說，實驗組顯著優於控制組。資料觀察法之近似詞教學均較傳統教學法更能有效提升學生的英語字彙能力。. 二、在不同程度的學生學習成效上，高、低分組學生均在後測成績有進步。對於高分組而言，實驗組後測成績顯著優於控制組後測。但對於控制組而言，實驗組的與控制組的後測成績未呈顯著差異。. 立. 政治大. 三、大部分的學生對於運用資料觀察法學習單字均給予正面回饋，也肯定資料觀. ‧ 國. 學. 察學習法活動的效益。另外，根據高、低分組學生訪談結果發現，英語程度的高低的確會影響學生對於資料觀察法的喜愛和需求。高分組的學生希望先以資. ‧. 料觀察學習法為開端，再以傳統講解式方式做總結。但對低分組的學生而言，. y. Nat. sit. 喜歡參與小組討論。由於單字量的不足，低分組學生希望在語料庫為主的教材. n. al. er. io. 旁能附上中文解釋，降低學習焦慮。. Ch. engchi. i n U. v. 根據上述研究結果，本研究建議大學英語教師在教學現場能夠融入語料觀察學習法並依照不同程度的學生進行教材設計，以助提升學生學習英語單字。關鍵字:資料觀察學習法、近似詞、語料庫為本. xi.

(13) ABSTRACT. Corpus Linguistics has progressively become the center in different domains of language research. With such development of large corpora, the potential applications and possibilities of corpora in second language teaching and learning are extended. A. 政治大. discovery-based authentic learning environment is provided as well as created by such. 立. corpus-based language learning. Synonym or near-synonym learning is a difficult. ‧ 國. 學. aspect of vocabulary learning, but a linguistic phenomenon with ubiquity. Hence, this. ‧. research aims to investigate the effectiveness of the application of data-driven. Nat. io. sit. y. learning (DDL) approach in near-synonyms instruction and compare the teaching. er. effect on the high and low achievers through the near-synonyms instruction.. al. n. v i n Participants of this study C were throughout the eight-week h given e n ginstruction chi U. corpus-based teaching with materials compiled by the teacher. This is a quasi-experimental study consisting of comparison between two experimental conditions, with a pre-post test and control-experimental group design, followed by qualitative method of semi-structure interviews and questionnaire provided to the experimental group of EFL university students in Taiwan. Two intact classes of 86 college students participated in this study. The quantitative analysis of the pre- and xii.

(14) posttest scores and questionnaire were conducted through descriptive statistics and frequency analysis in order to explain the learning effects and learners’ perceptions. The results of the study revealed that: (1) participants in the experimental group made significant improvement in the posttest; (2) EFL high proficiency learners with. 政治大. DDL approach performed better than high achievers who were taught by the. 立. traditional method. However, low achievers may not be able to benefit from DDL. ‧ 國. 學. approach in the form of concordance teaching materials; (3) the majority of the. ‧. participants had positive feedback on DDL activities. Also, types of preferred DDL. Nat. io. sit. y. activities were strongly influenced by students’ proficiency level. Low achievers. er. preferred activities that should involve Chinese translation as the supplementary note. al. n. v i n while as for the high achievers, C they for the teacher’s explanation of hwere e nlooking gchi U words’ usages and functions in the end. This study demonstrates the importance in illuminating the dynamic. relationship between DDL approach and second language near-synonyms learning, as well as provides English EFL teachers with a better concept to incorporate corpus or concordance lines into vocabulary instruction.. Key words: data-driven Learning, near-synonym, corpus-based approach xiii.

(15) CHAPTER 1 INTRODUCTION. 1.1 Background and Motivation. Taiwan is a test-oriented society, where many EFL learners are taught to learn English vocabulary through drilling and memorization because this is thought to be the. 政治大. best way to improve learners’ vocabulary. Although this method seems to be an effective. 立. way to cope with tests in school, students who undergo the school entrance exam system. ‧ 國. 學. in Taiwan might become confused and have problems in using precise vocabulary when dealing with new contexts.. ‧. Nevertheless, when students encounter a new vocabulary, teachers, before they. y. Nat. sit. think of the best way to explain or explore vocabulary, tend to use the traditional ‘drills. n. al. er. io. and repetition’, methods to make students memorize vocabulary. Due to the fact that the. Ch. i n U. v. majority of junior or senior high school teachers are under the pressure of limited time. engchi. and needing to maintain students’ academic performance, few teachers try to find a new way in helping students understand the usage. This is the situation in Taiwan, a test-oriented, EFL context, in which rote-learning is the most common method of learning vocabulary. When the students enter colleges, the teaching context will be changed with a more flexible surrounding. Students have no test pressure as previous learning stages. The teacher in college might have flexibility to arrange his or her own curriculum and to teach students in different ways. This study focuses on college students because they have no test pressure and more time to explore the usage of each word. For college teachers, 1.

(16) although the class still has limitation to some extent, teachers can still manage to do a lot of creative activities in class. In addition to the need of vocabulary development, the strategy of how to deal with bimodal distribution phenomenon of students’ English proficiency level in Taiwan EFL context has become an important issue for teachers.The performance gap in English proficiency has troubled a lot of English teachers to meet the needs of all students. The effectiveness of vocabulary acquisition lies much on learners’ ability to produce or recognize a word in different contexts, which requires the retrieval of the. 治政大understand the learning teachers want to design teaching materials, they need to first 立 vocabulary knowledge in long-term memory (Craik & Lockart, 1972). Hence, when. groups’ needs towards learning vocabulary.. 學. ‧ 國. distinctions between high and low achievers. Teachers will be able to respond to both. ‧. With the rapid development of large corpora and the availability of powerful. sit. y. Nat. personal computers, many researchers and teachers in ELT have begun to explore the. io. er. potentials of corpus linguistics in language teaching and language learning. One of the. al. ways to learn vocabulary is using a concordancer to learn vocabulary. Concordancer. n. v i n is not only a sophisticated computerC retrieval with a large amount of h e nprogram gchi U information in the form of computer language corpora facilitating data-based. inductive learning (Chan & Liou, 2005), but also a bundle of examples of a specific word or phrase that can be displayed efficiently. Thus, the amalgamation of corpora and concordances provide a new framework in the language learning and teaching by allowing learners to discover patterns through extensive naturally occurring examples in real texts (Hill, 2000). Several scholars have conducted the use of concordancing for second or foreign language acquisition, such as collocation learning (Lin, 2002; Tseng, 2002; Chan & Liou, 2005), lexical acquisition (Cobb, 1999; Sun & Wang, 2.

(17) 2003; Supatronant, 2005; Yeh, Liou & Li, 2007; Boulton, 2009), and grammar acquisition (Hsieh, 2008; Ewa, 2011; Wang, 2012). Data-driven learning (DDL) exploits the computer corpus for teaching and is developed by Tim Johns (1991), a pioneering teacher who combined corpus with language teaching (Leech, 1994; Hunston, 2002). The essence of DDL, a research-then-theory method, is that students acquire grammar rules and regular patterns inductively through exploring corpus materials. In the DDL approach, based on the theory that students can act as ‘language detectives (Johns, 1997:101),’ learners. 治政 regularity of the language. Teachers present learners大 with authentic examples and ask 立 are not regarded simply as recipients of knowledge, but as researchers studying the. them to interpret clues and draw conclusions from context and this encourages. ‧ 國. 學. learners to ‘search for information without knowing in advance what patterns they. ‧. will discover (Huston, 2002:184)’.. sit. y. Nat. With the large interest in data-driven learning (DDL), a growing number of. io. er. studies have suggested that data-driven learning may indeed facilitate processes beneficially to second language learning, especially to develop learners’ writing and. al. n. v i n CCobb, vocabulary ability (Gaskell& 2004; Scott& U Tribble, 2006; Boulton, 2009) to hen gchi name but a few. More Specifically, both low and high proficiency level students. increased their writing ability (Yoon& Hirvela, 2004; Tian, 2005; Boulton, 2009) and correct usage of grammar increased (Ciezielska-Ciupek, 2001; Tian, 2005). Students also view data-driven learning as a conducive method, which can result in their progressively accurate usage of vocabulary and grammar (Nesselhauf, 2003; Chen, 2004; Koosha & Jafarpour, 2006). Wu (2010) has discussed the effect of integrating corpus data into grammar instruction by using three near-synonyms ‘advise’, ‘recommend’, and ‘suggest’ as examples. Near-synonyms, which can be defined as 3.

(18) lexical pairs that have very similar cognitive and denotational meanings, but which may differ in collocational or prosodic behavior, such as ‘strong’ and ‘power ’(Halliday, 1976). However, the issue of teaching near-synonyms has been totally ignored in previous literature. In order to help fill this gap, this study establishes a niche to investigate the possible existence of using DDL approach to teach learners’ near-synonym and also investigate their collocational competence. The goal of the study is to find out whether the learner knows the correct usage of near-synonyms and their phrases expressions through the DDL approach.. 1.2 Significance of the Study. 立. 政治大. ‧ 國. 學. Data-driven learning (DDL) is really a “brand-new” teaching methodology in. ‧. Taiwan with regard to the EFL teaching context. In Taiwan, the majority of. sit. y. Nat. data-driven related studies were conducted in four domains: vocabulary learning and. io. er. teaching, collocational acquisition, second language grammar learning, and second. al. language writing. For EFL students, vocabulary building is really important for their. n. v i n C h and studies related academic success. However, most literature e n g c h i U to the topic of. vocabulary have emphasized the effects of learners’ collocational competence on vocabulary learning regardless of students’ proficiency level, and little research has been explored on the vocabulary issues of ‘near-synonyms’. The current study aims to explore the effectiveness of the application of data-driven approach in near-synonyms instruction. A secondary purpose of this study is to compare the teaching effect on the high and low achievers through the teaching of near-synonyms. The expected benefits of this research would show that data-driven learning of this research could be helpful not only for university students in a mix-competency 4.

(19) classes but also for college teachers on the front line who are considering incorporating corpus or concordance lines into their vocabulary instruction.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 5. i n U. v.

(20) 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 6. i n U. v.

(21) CHAPTER 2 LITERATURE REVIEW. This chapter is divided into three sections. The first section elaborates how data-driven learning (DDL) is applied in classrooms, using consciousness-raising activities and concordance lines as the teacher-made teaching materials. Secondly, some. 政治大. studies related to the impact of data-driven learning and teaching also reviewed in terms. 立. of three teaching perspectives: vocabulary, collocation, and grammar learning. In the third. ‧ 國. 學. section, some local and international studies regarding synonym and near-synonym applied in English learning and teaching are explored.. ‧. io. sit. y. Nat. 2.1 Data-driven learning (DDL) in a classroom. n. al. er. Applied Linguists working on technology-related issues have for some time noted. Ch. i n U. v. the relevance of technological changes in the digital global economy for TESOL (e.g.,. engchi. Cummins, 2000; Warschaucer, 2000; Chapelle 2001). In the area of language teaching and learning, the potential applications of electronic corpora, the electronic authentic language databases available on the Internet or stored in personal computers (Serkan, 2011), in language teaching and learning have receiveconsiderable attention in recent years. It is also possible for teachers and learners to access corpora by themselves (Gaskell& Cobb, 2004; Scott& Tribble, 2006; Boulton, 2010) to name but a few. The English language teaching (ELT) profession is currently undergoing what has been labeled by some as a major paradigm shift (Woodward, 1996). Starting from the mid 1980’s (Swain, 1985), some began to question many aspects of Communicative Language 7.

(22) Teaching (CLT). Since then, many more have gotten together to express the idea concerned with the direction that the position of CLT will be taken. Later on, the idea of DDL was proposed by Johns (1991) as an innovative approach to the implementation of concordancing materials, which lists all the occurrences of a word in a text with enough co-text in each line in the field of second language acquisition (SLA). The readers can tell how the word is used and where it was used. According to Johns (1994), the language learner using this approach is essentially a research worker whose learning is driven by access to authentic linguistic data.. 治政大since the specific aspects of product to process. It has the advantage of product approach 立 Batstone (1995) claims that DDL is a pedagogic continuum (see figure 2.1) from. language are presented to the learners by multiple exposures within contexts.. ‧ 國. 學. Simultaneously, it offers a completed process approach because DDL promotes creativity. ‧. and self-discovery learning among learners. While DDL is a very new methodology, DDL. sit. y. Nat. appears to utilize the strengths of both product and process approaches to teaching. io. n. al. er. grammar and collocation successfully (Hadley, 2004).. Manipulation. Ch. Regulation. engchi. i n U. v. Abdication. Pedagogic Continuum. PROCESS. Data-Driven Learning. Language is. Pre-selected. Authentic Language discovered. generated by. for the learner. by learner for improved communication. PRODUCT. Language. Figure 2.1. the learner. A pedagogic continuum from product to process grammar learning through DDL. Adapted from Batstone (1995).. DDL approach suggests that grammar and collocation learning should consist 8.

(23) largely of consciousness-raising activities rather than teaching rules (Willis & Willis, 1996). Consciousness-raising (CR) is defined by Rutherford and Smith (1988:108) as “the deliberate attempt to draw the learner’s attention specifically to the formal properties of the target language.” Kennedy (2003) also holds the similar view, and explains that the second vocabulary acquisition process as depending on the consciousness raising and learners should be exposed to authentic language materials as much as possible. Moreover, Schmidt (1990) argues that it is necessary to ‘notice’ features of the target language at the beginning in learning process. In Ellis’ (2003:163) study, he also claims that ‘consciousness-raising constitutes an approach to grammar teaching which is compatible. 治政大 If the teacher can place with current thinking about how learners acquire L2 grammar.’ 立 students in the position of researchers, this will increase learners’ self-esteem and. ‧ 國. 學. confidence (Johns, 1994; Willis & Willis, 1996; Johns, 2002).. ‧. Data-driven learning, a “research-then-theory” method which can be used to study. sit. y. Nat. grammar and collocation, is just like the consciousness-raising activity in that. io. er. language learners start with a question, and then come to their conclusions after analyzing. al. the corpora with a concordance program. Hence, the teaching material has become an. n. v i n C hwants to cultivateUlearners’ linguistic awareness. important element when the teacher engchi. Recently, scholars who support DDL have written an increasing amount of material to show how data from corpora and concordances can be used in the classroom. The results all indicate that learners have a big improvement after corpus-based teaching instructions (Tribble & Jones, 1990; John, 1994; Kettemann, 1995; Tribble, 1996). In these teaching materials, concordance plays an important role in spite of the different teaching. Teaching and learning purpose is to learn grammar patterns and collocations through the concordance line. More research will be necessary to determine if learning actually takes place from a DDL approach. If it could be shown the DDL approach is not only engaging 9.

(24) but also effective in language learning, it would be a major step toward establishing it in Taiwan EFL classroom. This study uses data-driven learning to investigate the learning effectiveness of learners. There is a controversial issue in the relationship between data-driven learning and learners’ language proficiency. Some scholars think DDL is helpful only for high achievers in learning English vocabulary, while some claim DDL helps low achievers too. For high achievers, the proponents of DDL take it as a task-based learning. which. requires good language awareness to find patterns and rules of teaching materials. 治政大 for its application. Johns the learners at advanced levels and a lot of training is essential 立. presented to them (Touraj & Zahra, 2012). Thus, it is claimed that DDL is only useful for. (1986) pointed out that DDL is appropriate for the learners who are adults and have high. ‧ 國. 學. motivation and intelligence. Similarly, in 2008, Boulton surveyed fifty empirical DDL. ‧. studies and noted that only four were conducted with beginning and low level studies. A. sit. y. Nat. lot of research on DDL reveals that this type of learning has been used for the learners at. io. er. advanced level. Boulton (2008:39) also noted that, “current research encourages the belief. al. that DDL is only useful for advanced learners in a computer laboratory, and with the. n. v i n C hto developing corpora researchers devoting considerable time e n g c h i U and training learners in small groups”. The reason why most previous DDL studies focus on high proficiency learners might be that the participants were chosen according to the researchers’ preference. Although there is a tendency towards thinking that DDL is only useful for the learners at advanced proficiency level, some investigations present evidence that DDL is helpful for the low proficiency level students as well. Sealy and Thompson (2007) proved that even primary school students can take advantage of corpora in their native language without a higher order thinking skills. However, Braun (2007) pointed out that when 10.

(25) corpus is used for learning in ESL or EFL context, the situation is totally different from using DDL to learn first language. It might become less efficient. This is the reason why in most articles about DDL, the participants are chosen from high proficiency level. Since very little DDL research ventures on low achievers in EFL learning context, this present study tries to investigate the effect of using corpus-based DDL on low achievers’ synonymous vocabulary learning, compared to traditional teaching.. 2.2 Concordance lines and Data-driven learning. 治政大 by Tim Johns (Sinclair, the 1980s as part of the “data-driven learning” (DDL) proposed 立. Concordance lines have been used as the teaching material in the classroom since. 1991). It is widely recognized as providing a fast and powerful means for learners to. ‧ 國. 學. become more aware of the language and the patterns of its various grammatical features,. ‧. words, and lexical patterns (Tribble & Jones, 1990; Sinclair, 1991; Mindt, 1997; Fox,. sit. y. Nat. 1998; J. Willis, 1998; Honston, 2002; O’Keefe et al., 2007). An important effect of. io. er. concordance lines is that they enable learners to take a more objective look at the language. They are like ‘tiny snapshots’ (Willis, 1998) of a linguistic landscape. Hunston. al. n. v i n C hinvolves setting upUsituations in which students can (2002: 170) also noted that, “DDL engchi answer questions about language by studying corpus data in the form of concordance lines or sentences”. Johns (1997) points out that a concordance can bring to the class abundant. authentic language examples that can be studied and exploited in many ways by learners, so that the learners can see how the target word collocates with other words, which grammar patterns will be followed, which preposition the target word goes with, and so on (Willis, 1990). Furthermore, using concordance line as the teaching material may be a more 11.

(26) learner-centered methodology. Johns (1994:101) refers to learners who are studying by concordance line as so-called “ language detectives ” whose task is to discover the rules or patterns of the language they are focusing on by finding, identifying and inferring these linguistic implication from the context (Serkan, 2011). Learners can be motivated to discover new meanings and to examine collocations because language learning is more likely to occur when adequate examples are noticed and processed by learners (Cobb, 2003). As for learner motivation, Tribble and Jones (1990) claim that concordance line can. 政治大 exercises is drawn from real life rather than concocted by teachers to increase motivation, 立 activate learner motivation. They believe that “ the fact that the source material for. ‧ 國. 學. as it gives learners immediate contact with the target language in use” (p. 38). However,. Tomlinson (1990) warns instructors against assuming that all of their students are “ready. ‧. and willing” to learn any particular teaching points chosen for presentation. Though it. sit. y. Nat. proves to be an effective teaching methodology, many people do not believe concordance. io. er. lines are suitable for lower or beginning level students, especially for weak and remedial. al. v i n C h Manual concordance lines before actual teaching in the classroom. lines have become an engchi U n. beginning students (Willis, 1998). The teacher has to compile and tailor the concordance. important teaching material when the teacher wants to teach through corpus-based approach. Figure 2.2 shows an example of the manual concordance lines of the target word ‘people’. Learners can read the context with each target word in each concordance line and then generate the usage and collocation by themselves.. 12.

(27) 1. a three-month contract with ACET, is looking forward to caring. people with for HIV/. AIDS. Letters to the Editor Letters. 2. To reduce the number of new HIV infections by giving young. people the facts about. AIDS. AIDS Care Education. 3. , under the auspices of the United Nations. At least nine. people were killed. between November 1990 and January 1991 by death squads linked. 4. may be POCs. Among the possible prisoners of conscience are 37. people sentenced to. up to 20 years' imprisonment for their involvement in justice. PRISONER LETTER WRITING CAMPAIGN. Each of. people whose story. the follows is a prisoner of conscience. Each has been. 立. 政治大. Figure 2.2 Example ‘people’ for manual concordance lines (extracted from BNC web. 學. Corpus). ‧ 國. ‧. With manual concordances, learners will have the chance to not only ‘notice’ the language in each text they encounter, but also ‘recognize’ the word they have seen it. y. Nat. io. sit. before (Allen, 2012). The rationale of manual concordance lines is like the ‘spiral structure’ in the textbook development. The intertexual nature of course books will also. n. al. er. 5. Ch. i n U. v. allow learners who choose to develop it to break through a new language awareness and. engchi. increase the sharpness of their learning curve. Since, a good general English course book will recycle important language from one lesson to another; manual concordance lines can serve as the same purpose. As Willis (1998:63) observes that manual concordance lines, as a learner-centered methodology, is ideal for mixed level class, since it allows students to work at their own level, in their own time and in their own way. Manual concordance will not only help learners process new language more effectively, but also produce the language repertoire more easily.. 13.

(28) 2.3 Related studies on data-driven learning. With the emergence of data-driven learning, the central task of the language learner changes. Rather than dull memorizing grammatical rules and word lists which lack context, learners are more focusing on learning the lexical chunks (e.g. collocation, discourse) and pragmatic convention that shapes the use of the lexis (Tribble & Jones, 1990; Lewis, 1993; Fox, 1998; Willis, 1998; O’Keefe et al., 2007). With rich ‘contexts’, generally speaking, data-driven learning can be conducted in three realms: vocabulary, collocation and grammar learning and teaching.. 2.3.1 Vocabulary. 立. 政治大. ‧ 國. 學. Cobb’s (1999) study found that the experiment group who learned language through concordance has a higher proportion in their retention than the control group, a. ‧. dictionary-based approach. Boulton also (2009) concluded that the experiment group who. y. Nat. sit. learned adverbs from corpus has better retention than the control group who was taught. n. al. er. io. by traditional teaching methodology. Yeh, Liou and Li (2007) conducted a synonym. i n U. v. research by means of using data-driven approach in Taiwan colleges. They tried to find. Ch. engchi. whether the learner can use synonyms effectively in their writing. For example, the students may substitute great, large or huge for big. The result showed that the scores of post-test are higher than pre-tests, but, the delayed post-tests score is lower slightly than post-tests. All in all, the students writing ability improved significantly after conducting the teaching experiment. Similar results were reported in Supatronant (2005) study. The study elaborated on the purpose of how concordance output software is in direct relation to definitive and productive knowledge of learners at the engineering department. The findings reported that students achieved greater progress in operating concordance output software. 14.

(29) 2.3.2 Collocation. The importance of collocations has long been stressed by scholars involved in teaching second or foreign languages (Xiao & McEnery, 2006). In second language learning, the knowledge of collocation effectively helps learners to generate sentences, to attain proficiency in writing, and to become fluent, native-like L2 speakers (Zang, 1993). On Taiwanese learners, Lin (2002) examined high-achievers’ and low-achievers’ performance in receptive and productive verb-noun collocation tasks after six weeks collocation instruction. Thirty-two students were divided into two groups. One class was. 政治大. conducted with the deductive teaching approach and the other with inductive approach.. 立. The result pointed out all students made more progress in receptive tests than productive. ‧ 國. 學. ones, and high-achievers generally performed better than low-achievers. The result suggested that learner’s competence might have impact on the effect of data-driven. ‧. learning. Chen and Liou (2005) investigated the influence of using data-driven learning. y. Nat. sit. with a bilingual concordancer on English verb-noun collocation learning. The study. n. al. er. io. revealed that students had more progress than their pre-test scores. In that, it is beneficial. i n U. v. to use data-driven learning approach to teach collocation (Wu, 1996; Liou, 1999 a, 1999. Ch. engchi. b; Chen, 2002; Lin, 2002; Liu, 2002; Chen & Liou, 2005; Li, 2005; Shin & Wang, 2006).. 2.3.3 Grammar In EFL context, an important feature for developing students’ grammatical competence is to provide them with a sufficient amount of language samples. Researchers such as Dekeyser (1995) and Doughty and Varela (1998) have all suggested that grammatical correctness in communication could be achieved by the analysis of forms and of functions with a bundle of examples. This has been one of the reasons that support the teaching rationale to use a corpus-based material as an example of the natural 15.

(30) resources. In Wang’s (2012) study, she investigated whether data-driven learning, as compared to focus on forms of grammar rules, is helpful for 5th grade English language learners in Taiwan in understanding and acquiring the English grammar points. The result indicates that the data-driven learning group performed better than the focus on forms learning group on the ‘third-person singular’. Ewa (2011) showed how competence in the understanding and creation of discourse can be achieved by the use of a corpus concordance and by the students performing various language-learning tasks. The concordance and the related task were assumed to be useful tools in developing students’. 治政大findings showed that a competence in formulating appropriateness in discourse. The 立. corpus-based approach not only raised students’ grammar awareness, but also guided the. ‧ 國. 學. students’ attempts at ‘rule discovery’. Hsieh (2008) provided a syllabus which was based. ‧. on corpus-based grammar syllabus design with three near-synonyms ‘see, watch, look at’.. sit. y. Nat. She suggested that grammar should be learnt with the context and a small number of. io. n. al. er. concordance line is not sufficient.. i n U. v. 2.4 Synonym and near-synonym in language learning and teaching. Ch. engchi. Vocabulary is the most common subject taught directly by teacher through corpora. Among vocabulary, synonym or near-synonyms are the most difficult vocabulary to learn (Higa, 1963; Lanfer, 1990; Tsui, 2005; Liu, 2010). Tsui’s (2005) research points out that for ESL teachers, one challenge in vocabulary instruction concerns the synonyms and near-synonyms. Synonym or near-synonym is a common yet complex linguistic phenomenon with its ubiquity in language. Liu (2010) assumed that while synonyms express basically the same concept, they often do so in different fashions, for different contexts, and/or from different perspectives. In other words, synonyms are often not identical in meaning and hence not completely interchangeable. Therefore, synonyms are 16.

(31) a challenging and simultaneously important lexical category because they are essential for expressing shades of meaning to help us convey our ideas and feelings accurately for effective communication (Edmonds & Hirst, 2002; Taylor, 2003; Divjak, 2006; Divjak & Gries, 2006). In linguistic term, they “are neither in free variation, nor in complementary distribution” (Divjak, 2006). Cruse (1986) remarked that the synonym relationship seems to be unstable, in that one of the synonyms, being dysfunctional, will be liable to fall into disuse; alternatively some semantic or stylistic nuance will in time come to differentiate words. Bolinger asserted (1977) that “if two ways of saying something differ in their. 治政大woods is a complex mentioned in his paper, the distinction between forest and 立. words or their arrangement they will also differ in meaning. Similarly, as Room (1981). combination of size, proximity to civilization, and wildness as determined by the type of. ‧ 國. 學. animals and plants therein. Research near-synonyms includes ‘little/small, high/tall,. ‧. start/begin, stop/finish, motor/engine’ (Taylor, 2003), cut/break (Taylor, 2007), fangbian. sit. y. Nat. ‘convenient’ and bianli ‘convenient’ (Chief, Huang, Chen, Tsai, & Chang, 2000),. io. er. gan3 趕‘rush’ and qiang3 搶‘rush’ (Chung, 2008) and many more.. al. v i n C means of data-driven vocabulary learning, especially by h e n g c h i U learning, a corpus-based n. However, only limited research has investigated the effects of synonymy on. approach. Higa (1963) found that learning two synonyms at one time is more difficult than learning two unrelated words. Laufer (1990) also found that synonymy is one of seven interlexical factors that can reduce the chances of vocabulary acquisition. She gave two reasons that synonyms may be more difficult to learn than other words. First, learners often make mistakes using synonyms because some of them may be substituted effectively in some contexts but not in others. For example, strong and powerful have similar meanings, but usually tea is only strong, and engines may be powerful but are rarely strong. This is an important point and demonstrates that some synonyms might be 17.

(32) more difficult to learn than others because synonyms with similar meanings do not always have the same collocates. Certainly, some words such as good and nice are synonymous in many contexts, while others such as powerful and strong are synonymous to a lesser degree. The second reason she gave is that less advanced learners are unlikely to try to learn words with similar meanings when they have a greater need to learn unknown L2 meanings. Since near-synonyms are hard to learn, teachers might put more emphasis on them when teaching vocabulary.. 治政大 learning. For most EFL To summarize, vocabulary is important for any language 立. 2.5 Chapter Summary. learners, learning vocabulary requires time and efforts. When it comes to classroom. ‧ 國. 學. instruction, language teachers should include activities that facilitate vocabulary learning. ‧. process. In Taiwan EFL context, teachers often use the traditional deductive methodology,. sit. y. Nat. they might face the common problem of ‘how to give students correct and practical. io. learning and teaching.. er. language usages (Conrad, 2000)’. This is still a controversial issue in second language. al. n. v i n C h ‘DDL’ proposed Then the corpora teaching methodology e n g c h i U by Tim Johns, is the key. methodology to teaching learners through corpora. Later, more and more researchers dedicated themselves to the research of corporal-based teaching method (Willis and Willis, 1996; Hunston, 1998; Cobb, 2003; Sun and Wang, 2003; Yeh, Liou & Li, 2007; Hsieh, 2008). They point out that through data-driven learning (DDL), students can not only learn the correct target language functions and usages from authentic materials but also acquire the vocabulary knowledge and grammar knowledge from concordance lines. Students might feel learning vocabulary is not a dull and abstract curriculum. Taiwanese EFL teachers still have doubts about the effectiveness of DDL. According to the literature, 18.

(33) DDL can be conducted in three realms: vocabulary, collocation and grammar learning and teaching. Identifying and classifying near-synonyms are a big challenge for second language learners because teachers seldom emphasize this part. Hence, it is necessary to teach near-synonyms in class. There is very little research on the effects of synonymy on vocabulary learning, especially by means of data-driven learning. Hence, the purpose of the present study is to explore the effect of using corpus-based DDL on learners’ synonymous vocabulary learning, compared to traditional teaching (with dictionaries). It also aims at focusing on the learners’ reactions toward the DDL approach.. 立. 政治大. 2.6 Research Purposes and Research Questions. ‧ 國. 學. The study adopted two groups, experimental group with data-driven learning. ‧. approach and control group with traditional teaching method. Participants received 10. sit. y. Nat. weeks of data-driven learning instruction and traditional teaching methodology separately. io. al. er. and the study would use both qualitative and quantitative methodologies. The present. n. study is designed to address the following research questions that have not been answered in previous studies: 1.. Ch. engchi. i n U. v. Does the data-driven learning (DDL) approach with manual concordances improve better in students’ near-synonyms knowledge and performance in comparison to control group taught in traditional method?. 2.. Are there any significant differences between high and low achievers in the experiment and the control group?. 3. What are learners’ reactions and preferences regarding using the corpus to learn vocabulary?. 19.

(34) 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 20. i n U. v.

(35) CHAPTER 3 METHODOLOGY. The researcher had a pilot study to see whether the DDL method and questionnaire are flexible or not. The pilot study is described first and then the quasi-experiment is introduced. The pilot study aims to investigate the influence of data-driven learning on. 政治大. three near-synonyms ‘people, person and human’ through corpus-based activity. There. 立. were three research questions in the pilot study similar to the quasi-experiment: (1) Does. ‧ 國. 學. the data-driven learning (DDL) approach with manual concordances improve students’ knowledge and performance of near-synonyms? If so, is it only effective after a certain. ‧. period of time? (2) What and which aspect of the three near-synonyms do the students. y. Nat. io. sit. have to notice? (3) What are the learners’ reactions and preferences regarding using the. n. al. er. corpus to learn vocabulary?. Ch. i n U. v. First, corpus technique was put to use in one first-year EFL reading and writing. engchi. class I for 13 English major students at National Chengchi University in the spring semester of 2012 academic year. Second, pre-test, post-test and delayed-posttest were conducted in order to compare the effectiveness of learners’ vocabulary knowledge. The main purpose of implementing this pilot study was to know whether students’word knowledge improved or not on three near-synonyms idiomatic expressions and the word usages. All the students’ native language was Chinese. Most had been learning English for 12 years at school. In the Activity, the researcher put more stress on learners’ vocabulary learning strategies which based on the Schmitt’s (2012:207-208) taxonomy. The lesson plan in 21.

(36) pilot study is displayed in Appendix A. The activity in quasi-experiment was followed by the same procedure in pilot study. For teaching materials, teacher gave every student a reading package. The reading package consisted of a lyric (where is the love: The Black Eyed Peas) and seven pages concordance handouts. The lyric was used as a warm-up activity to activate students’ motivation. Concordance lines were extracted and selected with careful consideration from the BNC web corpus and then compiled into four page handouts. The four-page concordance handout involved seven target ‘phrases’ and one ‘idiomatic expression’ which were based on three near-synonym (see Appendix B). Seven. 治政大 Table 3.1 Idioms and Phrases in Three Near-synonyms 立 Phrase Idiom. phrases and one idiomatic expression are shown in Table 1.. ‧ 國. 學. people boat people. lay person, people person,. snake oil person. ‧. person living person, in the person of, be on one’s person. y. Nat. n. er. io. al. sit. human complete human, human person. i n U. v. The results implied that participants made great progress in vocabulary knowledge. Ch. engchi. in immediate posttest after data-driven learning instruction. In addition, the participants’ delayed posttest values are still higher than pretest. This suggests that DDL (data-driven learning) can improve students’ near-synonymous vocabulary knowledge. In addition, the result also suggests that data-driven learning might not improve participants’ word retention ability. Such a finding echoes the previous studies (Sun &Wang, 2003; Chan & Liou , 2005; Koosha & Jafarpour, 2006; Yeh, Liou & Li, 2007; Alex, 2009; Ewa, 2011; Serkan, 2011) and proves that data-driven learning enhances participants’ second language vocabulary acquisition. For the learners’ perception, learners’ perceptions towards computer learning were 22.

(37) very positive. Learners thought that doing vocabulary exercises through paper-printed concordance is more interesting than doing identical exercises of traditional teaching from textbooks. Every learner would like to do other similar corpus activities in later classes and was satisfied with the experience of learning English vocabulary in a corpus-based learning environment. More students had a sense of achievement in the corpus-based data-driven learning class and more were satisfied with this method. Since the pilot study showed that the learners have significantly improved on near-synonyms, the main study will try to enlarge the population and increase teaching. 治政 The present study intends to investigate whether 大 learners’ near-synonymous 立. hours to investigate whether the learners improve their word knowledge or not.. achievement is enhanced through corpus-based data-driven learning (DDL) instruction. In. ‧ 國. 學. this chapter, detailed information about methodology is presented in the following order.. ‧. It will begin with the introduction of the research design, followed by the participants and. sit. y. Nat. setting selected in this research. Later, specific procedure and materials will be outlined.. al. n. 3.1 Research Design. io. be applied.. er. Finally, the instruments will be described, both quantitative and qualitative methods will. Ch. engchi. i n U. v. This is a quasi-experimental study with a pre-, and posttest control group design, accompanied with qualitative method of semi-structure interviews and questionnaire with the experimental group of EFL university students in Taiwan. The independent variable is the different teaching delivery modes (corpus-based data-driven learning vs. dictionary-based traditional vocabulary instruction) and the dependent variable is learners’ vocabulary knowledge achievement. Control variables include learner’s age, instruction hours, designed activities and previous learners’ English proficiency. There are two 23.

(38) groups in this study: one experimental group and one control group. The experimental group received corpus-based data-driven learning while the control group received dictionary-based instruction in a traditional classroom. Posttests of near-synonyms were conducted in the 10th week. The detailed of the research structure is shown in Figure 3.1.. Control variables: Independent variable:. Previous English Proficiency:. 1. Corpus-based data-driven. high achievers and low achievers. learning instruction. (Initial English proficiency). 2. Traditional instruction. English learning duration (Instructor and Instructor time). 立. 政治大. Nat. sit. Learners’ near-synonymous vocabulary knowledge. y. ‧. ‧ 國. 學. Dependent variables:. er. io. Learners’ perceptions. al. n. Preferred corpus-based data-driven learning activities. i n C Figure 3.1 Research h e n gStructure chi U. v. The variables involved in the study are explained as follows:. Control variables In order to control the reliability of the study, several significant factors were strictly controlled in the experimental teaching procedures. Since the classes were selected randomly by the researcher, students’ English proficiency might be different from one another. The experimental group and control group had similar English proficiency test which was based on their General English 24.

(39) Proficiency Test (GEPT) intermediate level in the last semester of academic year 2013. The GEPT intermediate level test was used as a standard to group the participants in this study. Later, the researcher chose the highest and the lowest thirty percent of learners separately as the high and low achievers in the experimental and control group.. English learning duration Duration means the time span the instructor gives lecture to learners. The instruction duration in both the experimental and the control group were the same. The instruction was given in each class for 50 minutes once a week, and it was given on 10. 政治大. consecutive weeks. And the total instructional time is five hundred minutes.. 立. ‧ 國. 學. Instructor. The instructor, who taught according to school schedule on Monday afternoon, was. ‧. the same as researcher in this study. He remained the same attitude in order to avoid the. Nat. sit. y. so-called ‘teacher’s factor’ interference, having no bias toward any particular students in. n. al. er. io. the ongoing research. In the study, the researcher plays the role as a curriculum designer,. i n U. v. activity facilitator, as well as an English instructor of near-synonymous vocabulary teaching.. Ch. engchi. Teaching materials All students were taught near-synonyms in each class. The students in the experimental group received data-driven learning instruction based on the teacher-made concordance lines selected from the BNC web corpus. There were forty pairs of near-synonyms in the teaching materials. The control group was taught with the dictionary-based words’ definitions which were extracted from the five on-line dictionaries (Merriam Webster, Longman dictionary of contemporary English, Oxford 25.

(40) dictionary, Cambridge dictionary, Collins English dictionary). Since both of on-line dictionaries were based on the word frequency, the teacher used these two dictionaries as the material to teach the control group.. Independent variable This study adapted only one independent variable which was the teaching methodology related to data-driven learning approach. Students followed the same activity procedures in ten weeks.. Dependent variables. 立. 政治大. The dependent variables were referred to the students’ performances which were. ‧ 國. 學. affected by the instruction and their reactions toward the data-driven learning approach.. ‧. 3.2 Research Procedure and Duration. sit. y. Nat. io. er. The procedure of the study is shown in Figure 3.2 which was conducted through. al. three phases, the preparation phase, the experimental instruction phase, and complete. n. v i n C h learning (DDL) phase. In the preparation phase, data-driven e n g c h i U was determined to be implemented in the college freshmen English class to facilitate near-synonymous. vocabulary learning. Simultaneously, the research questions were generated to help explore the effectiveness of DDL approach. Later, the researcher started to design activity and questionnaires as well as the format of pretest and posttest. After designing the whole research tool, a pilot study was conducted in National Chengchi University English major freshmen (Reading and Writing Class I ) to determine if the research design was viable or not. After gathering the data from the pilot study, the research analyzed the results and revised the research tools to make sure the quasi-experiment can be conducted in a more 26.

(41) accurate way. Since participants in the two-year junior college had no displacement test, not like the majority of universities that divided students into different English proficiency level, it was necessary to identify the students level in two classes and then selected the high and low achievers. Hence, the researcher separated the students into two groups based on their GEPT test scores (intermediate level) before conducting the qusai-experiment. The students took the GEPT test in the last semester of 2012 academic year. The GEPT score was the reference for their initial English proficiency. The GEPT test includes forty-four multiple choice questions which consist of four parts—. 治政大and the lowest thirty percent of percent of the participants was defined as high achievers, 立. vocabulary and structure, fill-in-the blank and reading comprehension. The highest thirty. into experimental and control group randomly.. 學. ‧ 國. the participants belonged to low achievers. Then the researcher divided the two classes. ‧. During the experimental instruction phase, at first, a background questionnaire was. sit. y. Nat. delivered to survey students’ learning experience and background information. In the first. io. er. week, an introduction and an orientation were given in the first class. From the second. al. class, data-driven learning approach was implemented in the experimental group while. n. v i n the traditional dictionary-based C approach the main instructing method in the control h e nwas gchi U group in the vocabulary session in each English class. The experimental instruction. continued for ten weeks, fifty minutes each time and once a week. After ten weeks instruction, students had a posttest to make sure if their near-synonymous vocabulary knowledge was improved. Later, a learner’s perception questionnaire was given only in the experimental group to measure learners’ affective skills toward data-driven approach. After filling the questionnaire, oral interviews were conducted to find out what effect data-driven learning instruction had on participants. Four of experimental groups, made up of high and low proficiency students, were served for the interview. The participants 27.

(42) were interviewed one by one. Finally, it was the data collection and analysis phase. After ten weeks, the data was analyzed through SPSS (17.0) independent t-test and pair-sample t-test. Later, according to the data results, the researcher wrote the thesis. The whole time table is presented in Table 3.1.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 28. i n U. v.

(43) Preparation Phase. Decide research topic and questions Literature review and design research framework. Design Activity and research tools (questionnaires, teaching materials, pre-and posstest) posttest). Select participants & and setting. Pilot study ( English major freshmen in National Chengchi University) Revise research tools and analyze the results of pilot study. 治政 Analyze learners’ proficiency test (GEPT) and 大decide high and low achievers in both groups. Divide research group into two groups. 立 ‧ 國. 學. Conduct the background questionnaire and pretest in two groups. ‧. Control group: traditional dictionary-based vocabulary teaching (8 weeks). n. sit. er. io. Phase. al. y. Nat. Experimental. Experimental group: Receive corpus-based data driven learning instruction (8 weeks). Ch. i n U. v. Conduct the posttest. engchi. Conduct questionnaire: Learners’ perception toward data-driven learning approach. Conduct interview. Complete Phase. Analyze data collected. Figure 3.2 Research Procedure 29.

(44) Table 3.2 Timetable of the study Year. 2012. Month. 10. 11. 2013 12. 1. 2. 3. 4. 5. Decide the research and questions Design research framework and design Activity I in pseudo-class. Pilot Study Phase. Revise Activity I and design Activity II in pseudo class Conduct research tools Implement the pilot experimental instruction (include pre-and posttest). 立. Conduct delayed posttest. 政治大. ‧ 國. 學. Analyzed collected data. n. al. er. io. sit. y. Nat. Quasi-experiment phase. Deliver pretests and background questionnaire Conduct quasi-experiment instruction. ‧. Measure learners’ initial language proficiency and group participants (high and low achievers). Ch. engchi. Conduct posttest Conduct perception questionnaire Conduct interview Data analysis and thesis writing Review Literature. 30. i n U. v. 6. 7. 8. 9. 10.

(45) 3.3 Setting and Participants. Setting This study was conducted in a College in Taipei City. This school is located in the center of a popular educational district, which is situated on the southernmost part of the Taipei City. There are five departments in this college. Students in this college have to study two years and then have a special test. There are totally 84 classes in the two-year college, 47 classes for the first-year, and 37 for the second-year. There are 44 students in one class. The enrollment is approximately two thousand students. Most students want to. 政治大. engage in government jobs in the future. Students can get familiar with English because. 立. they have a special examination in the second year of their college life. If they pass the. ‧ 國. 學. examination, they will become a government official worker in Taiwan. The classrooms in this college are equipped with a computer, a projector and a. ‧. speaker. The teacher is able to access to the internet via overhead projector to display. y. Nat. n. al. er. io Participants. sit. information from the websites or PowerPoint.. Ch. engchi. i n U. v. There are 47 classes in the first year in this school. Two English classes were selected randomly from all junior classes. Each class had 43 students (Female: 5, Male: 38). So, 86 students were participating in this experiment. Since the college did not have the English placement test, each English class was a mixed English proficiency class. The two classes were randomly assigned to the experimental group with concordance lines approach by using paper-based DDL material and the other was the control group with traditional grammar teaching respectively. According to previous studies mentioned in literature review, the results showed that DDL might have influenced on low and high achievers. Hence, in order to explore the influence of English proficiency on the 31.

(46) preference of data-driven learning, the 86 participants were further divided into high and low achievers according to their GEPT (intermediate level) scores in the previous semester. The researcher only discussed the highest and lowest thirty percent of learners in experiment and control group. Table 3.2 shows the distribution of participants in each subgroup.. Table 3.3 Number of students in each subgroup Groups. Proficiency. Total. High Experimental group. 立. 15. 13 政 Low治大 High. 15. Low. 17. 學 ‧. ‧ 國. Control group. In each group, students were divided into eight subgroups, each group had five to. sit. y. Nat. six people. In each lesson, the teacher taught four pairs of near-synonyms, which meant. al. er. io. the teacher taught at least eight words in one class. The words were selected from the. v. n. ‘vocabulary column’ of learners’ textbook and then the teacher added one or two words to. Ch. engchi. i n U. form near-synonyms pairs. The format of data-driven learning activity is shown in Table 3.3. Table 3.4 Format of Data-driven learning activity. 3/4. 3/11. Group1. Group2. Group3. Group4. Group5. Group6. Group7. Group8. human. human. see. see. spend. spend. tour. tour. person. person. watch. watch. take. take. travel. travel. people. people. look at. look at. cost. cost. trip. trip. eventful. eventful. creative. creative. advantage. advantage. alter. alter. Significant. Significant. imaginative. imaginative. benefit. benefit. change. change. important. important. 32.