教師回饋對台灣高中EFL低成就學生段落寫作之效用：「直接訂正法」與「語意重述法」 - 政大學術集成

全文

(1)國立政治大學英國語文學系碩士在職專班碩士論文. 指導教授：葉潔宇博士 Advisor: Dr. Chieh-yue Yeh. 立. 政治大. ‧ 國. 學. 教師回饋對台灣高中 EFL 低成就學生段落寫作之效用：「直接訂正法」與「語意重述法」. ‧. The Effects of Corrective Feedback on Taiwan High School EFL Low-achievers' Paragraph Writing: “ Direct Correction”vs. “ Reformulation”. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. 研究生：許凱絨撰 Name: Kai-jung Hsu 中華民國 99 年 6 月 June, 2010. v.

(2) The Effects of Corrective Feedback on Taiwan High School EFL Low-achievers' Paragraph Writing: Direct Correction vs. Reformulation. A Master Thesis Presented to Department of English,. 政治大. 立 National Chengchi University ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. In Partial Fulfillment of the Requirements for the Degree of Master of Arts. by Kai-jung Hsu June, 2010.

(3) To My Parents 獻給我的父母. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. iii. i n U. v.

(4) Acknowledgments. My warmest thanks go to Dr. Chieh-yue Yeh for her inspiring guidance and encouragement throughout my research for this work. For their reading of the manuscript and for helpful suggestions and other support, I want to thank Dr. Chen-kuan Chen and Dr. Yi-ping Huang. I greatly appreciate Dr. Hsueh-ying Yu and. 政治大 my upper classmate, Shu-ping Huang, for their suggestions. Also, I would like to give 立. ‧ 國. 學. my thanks to my classmates, Hui-ling Chan and Chiao-yu Chen for their. ‧. encouragement. My gratitude is also extended to my colleagues in Tunxu High School. sit. y. Nat. of Commerce and Industry for their help and kindness. I also greatly appreciate the. n. al. er. io. financial aid from my husband, I-ting Chou, whose thoughtfulness supported me. Ch. i n U. v. throughout the final stage of my research. I am thankful for the cooperation of my. engchi. students in Class 241 and Class 242 in Tunxu High School of Commerce and Industry. Finally, I would like to extend my heart-felt thanks to my family. Their support has made this work possible.. iv.

(5) Table of Contents. Ac knowl e dg e me nt s ……………………………………………………...…………iv Chinese Abstract……………………………………………………………………viii English Abstract…………………………………………………………………….ix Chapter One: Introduction……………………………………………….……….1. 政治大 Purpose and Value立 of the Study………………………………………………..3 Background and Motivation……………………………………….…………..1. ‧ 國. 學. Definition of Terms…………………………………………………………….4 Chapter Two: Literature Review…………………………………….……………5. ‧. Noticing in Second Language Acquisition (SLA)……………………………..5. sit. y. Nat. Language Processing……………………………………………………...7. n. al. er. io. Input-intake-output Hypothesis…………………………………………...9. i n U. v. Corrective Feedback in Writing……………………………………………….10. Ch. engchi. Reformulation…………………………………………………………….12 Research Questions…………………………………………………………….15 Chapter Three: Methodology…………………………………….………………..17 Participants……………………………………………………………………..17 Instruments……………………………………………………………………..18 EBCT in 2009……………………………………………………………..18 GEPT Elementary Level Writing Test…………………………………….19 GEPT Holistic Scoring Guide……………………………………………..19 Reformulation……………………………………………………………...20 v.

(6) Error Record Form………………………………………………………...21 Error Classification System……………………………………………….21 Interview…………………………………………………………………..22 Procedures……………………………………………………………………...22 Step 1: Sampling and Orientation…………………………………………23 Step 2: Pre-test…………………………………………………………….23 Step 3: Rating and Grouping……………………………………………....23. 政治大 Inter-rater Reliability……………………………………………........25 立. Intra-rater Reliability…………………………………………………24. Step 4: Treatment………………………………………………………….26. ‧ 國. 學. The Experimental Group……………………………………………..26. ‧. The Control Group…………………………………………………...27. sit. y. Nat. Step 5: Post-test and Rating……………………………………………….28. io. er. Step 6: Interview…………………………………………………………..28 Data Analysis…………………………………………………………………...29. al. n. v i n Ch Chapter Four: Results…………………………………….………………………..32 engchi U. RQ1……………………………………………………………………………..32 RQ 2………………………………………………………………………….....34 RQ 3…………………………………………………………………………….37 The Experimental Group………………………………………………….38 The Control Group………………………………………………………...40 The Experimental Group vs. the Control Group…………………………..41. RQ 4…………………………………………………………………………….44 Students’ Views on Reformulation as a Way to Improve Writing…………46 Easy Parts for Students…………………………………………………….46 vi.

(7) Difficult Parts for Students………………………………………………...47 Students’Suggestions……………………………………………………...49 Students’Change in Writing Behavior…………………………………….49 Chapter Five: Discussion and Conclusion…………….…………………….……..51 Summary and Discussion……………………………………………………….51 Helpfulness in Low-achievers’ Writing: Reformulation vs. Direct Error Correction…………………………………………………………………….…51. 政治大 writing……..……………………………………………………………51 立. 1. Direct error correction may not be so helpful in learners’. 2. Modified reformulation technique appears effective…………………...52. ‧ 國. 學. 3. “ Memory”and “ time”can make a difference…………………………..53. ‧. Effect of Modified Reformulation on Low-achievers at Different Proficiency. sit. y. Nat. Levels…………………………………………………………………………...54. io. er. Change in Errors after Treatment……………………………………………….58 Students’ Views on Reformulation as a Way to Improve Writing………………61. al. n. v i n CReformulation Students’ Views on h e n g c hActivities………………………….…..62 i U Students’Suggestions on Reformulation and Change in Writing. Behavior…………………………………………………………………...64 Implications of the Study……………………………………………………….66 Limitations……………………………………………………………………...68 Suggestions for Further Studies………………………………………………...69 Conclusion……………………………………………………………………...70 References……………………………………………………………………………72 Appendixes…………………………………………………………………………...79. vii.

(8) 國立政治大學英國語文學系碩士在職專班碩士論文提要. 論文名稱：教師回饋對台灣高中 EFL 低成就學生段落寫作之效用：「直接訂正法」與「語意重述法」指導教授：葉潔宇博士研究生：許凱絨論文提要內容：. 政治大本研究針對台灣 EFL 低成就學生之段落寫作，比較「直接訂正法」與「語立. 意重述法」兩種寫作回饋之成效。本研究對象為台北市某高職學生，共 56 名學. ‧ 國. 學. 生全程參與這項從 2009 年 9 月到 2010 年 1 月的研究。進行修改寫作時，教師對. ‧. 實驗組使用「語意重述法」，學生比較原稿與老師保留學生原意但改寫成符合英. y. Nat. 文語法的段落，並將發現的文法錯誤記錄並自行訂正；對照組則運用「直接訂正. er. io. sit. 法」，學生審視老師直接在上面訂正的原稿。經過看圖英文段落寫作的前測與後測、實驗組與對照組後測結果比較、以及針對實驗組的訪談，研究結果如下：(1). al. n. v i n 整體性評量上，「語意重述法」對學生改進寫作較為有效；(2)兩組中程度較差之 Ch engchi U 低成就學生進步程度均優於程度較好之低成就學生，尤其實驗組之程度較差者進步程度猶勝於對照組的；(3) 「直接訂正法」對減少學生文法錯誤之功效優於「語；(4)絕大多數參與者認為「語意重述法」有助增進寫作能力。論文最意重述法」後討論此研究在教學上的意涵與提出對之後研究的建議。. viii.

(9) Abstract This study aimed to compare the efficacy of “ direct correction”with that of “ reformulation”on Taiwan EFL low-achievers’paragraph writing. Fifty-six students in a vocational high school in Taipei City participated in this study from Sep. 2009 through Jan. 2010. When conducting revision activities, the teacher implemented the “ reformulation”technique in the experimental group. The students compared the originals with the reformulated versions given by the teacher, and detected, recorded, and corrected all the grammatical errors mainly on their own. The control group. 政治大. received the “ direct correction”treatment, examining their originals with the teacher’ s. 立. corrections on them. With the pre-test and the post-test on a paragraph-length English. ‧ 國. 學. picture description, the comparison of the post-test results between the experimental and control groups, and interviews with the experimental group, the results are as. ‧. follows: First, in holistic rating, “ reformulation”was more helpful than “ direct. y. Nat. sit. correction”in improving the participants’writing performance. Second, the. n. al. er. io. low-achievers with lower proficiency benefited more from “ reformulation”than those. i n U. v. with better proficiency. Third, “ direct correction”was more effective than. Ch. engchi. “ reformulation”in reducing the participants’grammatical errors. Fourth, the majority in the experimental group were positive of “ reformulation”as a way to improve writing. Finally, some implications for pedagogy and suggestions for future studies were made.. ix.

(10) 1. CHAPTER 1 INTRODUCTION Motivation and Background Writing assignments and follow-up feedback characterize any language writing course (Kroll, 2001). Students have to write in order to improve writing ability; at the same time, teachers’feedback can be referred to as directions to their improvement. Furthermore, in terms of learning English as a second or foreign language (ESL or. 政治大 language, especially at the aspect of form. Based on the written texts, corrective 立. EFL), students’written production can reveal whether they have acquired the target. feedback (CF) can be provided for learners to correct their misinterpretations about. ‧ 國. 學. some aspect of the target language (Larsen-Freeman, 2001).. ‧. CF studies on English writing are found dominantly in the context where students. sit. y. Nat. have English proficiency at the intermediate level or above, and the written output. io. er. investigated are compositions, essays, or papers (Ashwell, 2000; Fathman, & Walley, 1990; Huang, 2006; Naeini, 2008; Wu, 2003; to name but a few). Different aspects. al. n. v i n C h In terms of student regarding CF have been investigated. processing, Cohen (1987) engchi U. surveyed 217 second language learners in a university and found that 80 % of them preferred to make a mental note when receiving written feedback. Li (2004) classified the errors and discussed possible causes in the 94 ESL college students’free writings on self introduction. In addition, more attention is given to the relationship between CF types and their effects (Ashwell, 2000; Bitchener et al., 2005; Fathman, & Walley, 1990; Ferris, & Roberts, 2001; Huang, 2006; Naeini, 2008; Wu, 2003). Many of the researchers mentioned above are positive of CF effects on English writing, while some researchers oppose corrective feedback in language learning. Truscott (2007) indicates that corrective feedback should be abandoned because it.

(11) hinders students’freedom of producing the target language. In addition, Guénette (2007) doubts the CF effects, as he finds that many students do not read the corrected papers. When these students ignore what to notice in their EFL/ESL use, the errors tend to repeat themselves (Miao, Badger, & Zhen, 2006). Even so, Swain (1985) stresses the importance of CF and advocates it as the medium between comprehensible input and output in the target language. Actually, the EFL low-achievers, like many in Taiwan, have to depend greatly on their teachers’. 政治大 Larsen Freeman (2001), CF is what students need to reject or modify their hypotheses 立 corrective feedback to find out what to improve in their English writing. According to. about the target language. In fact, whether CF makes effect on writing ability relies. ‧ 國. 學. heavily on learners themselves. Responsibility for the correction of any given piece of. ‧. writing should fall mainly on the student, not on the teacher (Rosen, 1987). Learners. sit. y. Nat. should be trained to search for, find, and correct their own mistakes. This. io. er. awareness-raising purpose may be served through reformulation (Cohen, 1982, 1983), a CF procedure through which learners make cognitive comparison between their. al. n. v i n C h and the target language, second/foreign language (L2/FL) and become more engchi U conscious of their L2/FL use.. Though the studies on reformulation have been conducted with L2 learners at the higher intermediate level and above (Cohen, 1982, 1983; Qi, & Lapkin,2001;Sachs, 2003), few empirical studies, if any, have targeted at the EFL learners in Taiwan, who learn English as an exam subject and have little access to communicative use of English. That is, English learning in most EFL classrooms in Taiwan is synonymous with developing accurate English grammar, vocabulary, and translation skills since the goal is to pass high school and university entrance exams (Fotos, 2001). Therefore, further research needs to be conducted in such an English learning context to 2.

(12) determine the effect of reformulation on Taiwan EFL learners’writing. In addition, it is interesting to uncover the relationship between s t ude nt s ’English proficiency levels and the extent of improvement when reformulation is employed in the writing process. Purpose and Value of the Study The present study aims to uncover whether Taiwan EFL low-achievers progress in paragraph writing through reformulation, and to explore how reformulation may. 政治大 each participating student may pay attention to his or her own errors and decrease the 立. influence EFL learners with different English proficiency levels. It is expected that. frequency of error making through comparison and contrast of their erroneous. ‧ 國. 學. sentences with the reformulated version. In the meantime, the participants are. ‧. expected to perceive the form of the target language through awareness-raising so that. sit. y. Nat. they may advance in writing performance.. io. er. To put it differently, through reformulation, or awareness-raising in the target language, the participants are expected to turn to their knowledge of grammar, and to. al. n. v i n Ch see what can fit in with the relationship they map out between form and meaning. engchi U. Therefore, in the present study, reformulation is supposed to facilitate the interaction of input and output, which leads up to language acquisition. More importantly, the reformulation procedure is adopted to force students to become more involved in and responsible for their learning since they have to notice and compare the similarities and differences between two language systems (the target and native languages). It is hoped that reformulation would encourage more student involvement in learning how to write more correctly in English.. 3.

(13) Definition of Terms Reformulation: Reformulation is an alternative corrective feedback in writing. It means that a native speaker rewrites a second language l e a r ne r ’ sc ompos i t i onbut the latter’ s original ideas are retained. Then the learner is asked to make cognitive comparison between his or her original sentences and the native-speaker version. He or she is expected to find how his or her ideas can be conveyed in the target language and thus to adjust his or her interlanguage to become more similar to the very. 政治大 This procedure has been adopted in several L2 writing studies (Cohen, 1982, 立. language.. 1983; Qi, & Lapkin,2001;Sachs, 2003). The researchers found, compared to. ‧ 國. 學. traditional reconstruction or correction in writing, that reformulation appeared to be. ‧. more stimulating and meaningful feedback for the L2/FL learners to attend to their. sit. y. Nat. misuses of the target language, and hence to pursue more native-like language use. In. io. er. other words, reformulation is a procedure where students’output turns out to be more digestible input, or intake, which may better raise the target students’attention to their. n. al. i n C hup language acquisition. L2/FL use in writing and speed engchi U. v. In the present study, the part of the reformulator, or the rewriter is altered to. better fit the real English classroom situation in Taiwan. This is due to the fact that an English native speaker is seldom available in English writing classes in Taiwan’ s senior high schools. Therefore, the rewriter in the present study is the teacher-researcher, a senior high school English teacher with 7-year teaching experience in English writing.. 4.

(14) 5. CHAPTER 2 LITERATURE REVIEW This chapter will present two aspects of literature relevant to this research. First, Noticing Hypothesis is mentioned to help the readers understand the role of noticing in second language acquisition, which concerns language processing and input-intake-output hypothesis. The second aspect is about corrective feedback (CF) in writing. It is considered essential to lead student-writers to become more conscious. 政治大 is reviewed about its application. 立. of their use of the target language. More significantly, reformulation, a CF technique,. Noticing in Second Language Acquisition (SLA). ‧ 國. 學. The relationship between the knowledge of grammatical forms and rules and the. ‧. ability to use them accurately in communication remains unclear. However,. sit. y. Nat. “ noticing”(Schmidt, 1990) has been suggested as an interface by the growing body of. io. er. empirical evidence in favor of positive effects of formal instruction in SLA. That is, although universal grammar (Chomsky, 1957) appears to guide the first language. al. n. v i n acquisition in small children,Cafter the critical period, h e n g c h i U or puberty, learners may need explicit feedback to direct their attention to relevant input toward structures so that they can notice and integrate new forms in the target language. Schmidt (1990), based on the theory of consciousness (Baars, 1988), stresses the importance of conscious experiences in interlanguage development. His Noticing Hypothesis states that it is possible for implicit learning to happen incidentally during meaningful interaction but learners may require tasks that force them to notice certain aspects of the input. Actually, many ESL/EFL learners, especially those at lower levels of proficiency, cannot process the target language input for both meaning and form at the same time (Skehan, 1998; Tomasello, 1998). This is true especially when.

(15) they are focusing attention to the message being conveyed in the target language. Therefore, even while input and interaction are emphasized in communicative language learning, it is crucial to realize that learners may need certain kind of direct intervention (Ellis, 1995). It is also suggested that particular kinds of noticing may be necessary for SLA.. That is, not only must learners pay attention to linguistic. features of input in order for it to become intake, but they must notice the gap between their interlanguage output and the target language input.. 政治大 -r a i s i ng , ”us e dbyRut he r f or da ndSha r wood Smith (1985), refers to increased learner 立 Not i c i ngi nSLAha sbe e na ddr e s s e di ndi f f e r e ntt e r ms .Fi r s t ,“ c ons c i ous ne s s. awareness of particular linguistic features. Then Klein (1986) claims that language. ‧ 國. 學. l e a r ne r ss houl dbea bl et ous e“ ma t c hi ng ”t oc he c kt he i rout puta ga i ns ta ne xt e r na l. ‧. me a s ur e ,whe r e a sO’ Ma l l e ya ndCha mot( 1990) propose two strategies of noticing:. sit. y. Nat. “ s e l e c t i vea t t e nt i on”a nd“ s e l f -e va l ua t i on. ”The former means paying attention to. io. er. particular linguistic items in input, and the latter refers to making sure that output answers to internal accuracy measures. In addition, related studies on corrective. al. n. v i n Cdt f e e dba c k( CF)ha vee mpl oy e m“ i npute nha nc e me nt ”t ode s c r i beCFa sone hheeter ngchi U specific form of conscious-raising (Lightbown and Spade, 1990; Sharwood Smith,. 1991; White et al., 1991). Finally, Ellis (1995) terms noticing as“ g r a mma r / c og ni t i ve c ompa r i s on, ”whi c hs pot l i g ht st hef a c tt ha tt hel e a r ne rmus tnot i c ebot hs i mi l a r i t i e s and differences between the interlanguage and the target language. That is, one can confirm and disconfirm hypotheses that exist in his or her implicit knowledge by comparing what one has noticed in input to what one is able to produce in output. Among the noticing-related strategies above, the awareness of grammatical form is viewed as crucial in helping learners to restructure their interlanguage systems.. 6.

(16) Language Processing Concerning the role ascribed to noticing as a trigger in language processing, Table 2.1, cited from Skehan (2002, p. 88-89), briefs the interaction between input and output in language acquisition and highlights the comparison and interaction between the interlanguage and the target language. In the process, only some portion of new information of the target language may be perceived by the learner. At the same time, explicit knowledge (learned linguistic structures) usually does not turn directly into. 政治大 constraints. That is, when learners’interlanguage is not sufficiently advanced, they 立 implicit knowledge (acquired linguistic competence) because of learnability. may not be able to integrate certain kinds of new information. However, according to. ‧ 國. 學. Ellis (1995), explicit knowledge might help learners to notice forms, to think about. ‧. what they mean, and to compare those form-function mappings with their. Table 2.1. er. io. sit. y. Nat. interlanguage systems.. SLA processing (cited from Skehan, 2002, p. 88-89). n. al. SLA processing stage Noticing. Ch. i n U. v. Nature of stage. i eThe n glearner c h directs attention to some aspects of the language system, or is led to direct attention in this way.. Pattern identification. On the basis of the focal attention, the leaner makes a hypothesis or generalization, implicitly or explicitly, about the target language, based on a perceived pattern or regularity.. Extending. The learner extends the domain of the hypothesis, without changing it fundamentally in kind.. Complexifying. The learner apprehends the limitations of the identified pattern and restructures it, as new aspects of the target language are noticed.. 7.

(17) Integrating. The learner makes an attempt at the output by integrating the new subarea of interlangauge into a learning structure.. Becoming accurate, avoiding error. The learner becomes able to use the interlangauge area without making errors, although this use may be slow and effortful.. Creating a repertoire, achieving salience. Not only can errors be avoided, but the interlanguage form can be accessed at appropriate places--it becomes part of salient (not latent) language repertoire.. Automatizing rule-based language, achieving fluency. The domain is now used correctly with reasonable speed, and the role has become, to some degree, proceduralized (implicit).. 立. 政治大. ‧. ‧ 國. 學. The learner, at this stage, is also able to produce the interlanguage form in question as a lexicalized element. Namely, language which may be analyzable (and has been produced on the basis of analysis) is now also available as a lexical element. In addition, the learner is able to choose freely, and appropriately, between lexicalized representation of an interlanguage form and its rule-based creative version.. Lexicalizing. n. er. io. sit. y. Nat. al. Ch. i n U. v. One thing to notice here is that noticing does not guarantee that input will. engchi. become intake, and its usefulness may depend on a learner’ s developmental readiness. However, if noticing truly is a prerequisite to acquisition as Shmidt maintains, instruction that promotes noticing should make language acquisition more likely. Moreover, several researchers (Ellis, 2001; Gass, 1988; Robinson, 1995; Schmidt, 1990) have discussed a variety of factors which may influence whether learners notice forms and achieve intake. Considering the external factors, one’ s quality of noticing and acquisition can be affected by the materials for instruction, the task demands, the learning environment, as well as the frequency, perceptual salience (defined as availability of data), and complexity of the forms. The internal factors may include a 8.

(18) learner’ s developmental stage, learnability constraints, prior knowledge, skill level, memory, and attentional capacities, social distance, status, motivation, and attitude. All the above points will be useful when the results of this empirical study are discussed later. Input-intake-output Hypothesis According to Krashen (1984), input is either comprehensible or not; comprehensible input will automatically trigger language acquisition. In contrast to. 政治大 Ga s s( 1988)pr opos e s“ c ompr e he nde di nput , ”whi c hi sde vi s e dt os t r e s st ha ti ti st he 立. comprehensible input, which is controlled by the person providing the linguistic data,. l e a r ne rwhoi sdoi ngt he“ c ompr e he ns i on”wor k.I not he rwor ds ,al e a r ne r ,whe n. ‧ 國. 學. exposed to input, should be able to pay selective attention to certain part of the new. ‧. linguistic information.. sit. y. Nat. Comprehended input is usually multi-staged. One may first realize the semantic. io. er. meaning of the input, then perform linguistic analysis, and finally understand its phonological and syntactic patterns. However, not all that is comprehended becomes. al. n. v i n CspherceivedforprocesU i nt a ke ;onl yc e r t a i npor t i oni s i ng ,whi c hi s“ pr e l i mi na r yi nt a ke ” engchi (Chaudron,1985). Intake is not simply a subset of input, but the process of attempted integration of linguistic information, which mediates between target language input. a ndt hel e a r ne r ’ si nternalized set of rules. Moreover, when preliminary intake is further processed, it turns into “ f i na li nt a ke ”(Chaudron, 1985), a part of the larger corpus of comprehensible input responsible for enabling the learner to understand and use stored linguistic data. To achieve successful language acquisition, learners would need the opportunity for meaningful use of the integrated linguistic resources. “ Comprehensible output,” according to Swain (1985), can extend learners’l i ng ui s t i cr e pe r t oi r ea sthey attempt 9.

(19) to create precisely and appropriately the meaning desired. When learners attempt to produce the target language, the difficulties which they experience may push them to become aware of what they need to express themselves effectively. Learners would attend to the form of the very language, deal with their problematic linguistic product, and modify it. Such a process assists them in extending their knowledge of the target language. Ellis (1990), echoing Swain, regards production as a medium between the learner’ s semantic and syntactic processing. Such output can stem from final intake.. 政治大 Figure 2.1, adapted from Seow, & Tay (2004, p. 17). 立. The relationship among input, intake, and output in SLA may be summarized in. Figure 2.1. ‧ 國. 學. input-intake-output (adapted from Seow, & Tay, 2004, p. 17). ‧ y. Comprehension. io. sit. Nat. Input. n. al. er. Comprehended input1 Preliminary intake2. Ch. Comprehensible output3 Final intake4. engchi. i n U. v. Production. 1. Gass (1988) Chaudron (1985) 3 Gass (1988); Schmits (1990); Sharwood Smith (1986); Swain (1985) 4 Chaudron (1985) 2. Corrective Feedback in Writing Theoretically, in SLA there exists a close relationship between noticing and output. Noticing is the prerequisite of language acquisition, while comprehensible output is the evidence of whether language acquisition is successful or not. Pedagogically, errors which appear in production have been treated as opportunities to facilitate SLA. Errors in output, according to Swain (1998), can prompt second 10.

(20) language learners to recognize consciously some of their linguistic problems. They may thus become aware of something they need to find out about their second/foreign language (L2/FL). In Swa i na ndLa pki n’ s study in 1995, by means of think-aloud, the French immersion participants, average aged 13, reported noticing the gaps between their interlangauge and the target language while producing their compositions. The students were also found to engage in certain thought processes, including grammatical analysis for accuracy, when they encountered difficulties in producing. 政治大 sometimes faulty, which led to incorrect hypotheses and inappropriate generalizations. 立 the target language. In addition, the substance of these participants’t houg ht swa s. This implies that relevant feedback can play a central role in improving L2 writing. ‧ 國. 學. development.. ‧. Many L2/FL learners regard corrections as essential and want to be corrected. sit. y. Nat. regularly (Havranek, 2002;Leki, 1991; Schulz, 2001; Zhang, 1995). However,. io. er. regarding feedback, there seems to be a debate over issues such as form/meaning focus (Horowitz, 1986; Hyland, 1998; Paulus, 1999; Zamel, 1988) and effectiveness. al. n. v i n C oft e a c he r s ’c omme nt sore r r orc n( Fe r r i s , 1999;Polio, Fleck, & Leder, 1998; U hoerrenctio i h gc. Semke, 1984; Truscott, 2007; Zamel, 1985). Whatever form feedback may come in, its purpose is to help learners to notice the gaps between their L2/FL output and the native speaker’ s version. In fact, if learners can be conscious of the gaps, they are doing their own error analysis (EA), which can be more effective than mere grammatical correction or comments from peers and teachers (James, 1998). Learners first notice something. problematic in their own production, and then they compare it with a native-speaker version of it. Practically speaking, James’EA strategy seems more workable in real L2/FL classrooms, especially when the students are at lower levels of proficiency. 11.

(21) Such students can only depend on their own recent learning experiences (especially negative ones) to notice the forms and to make cognitive comparisons (James, 1998). This EA sense of cognitive comparison is consistent with the technique of “ reformulation”(Levenston, 1978). Reformulation Reformulation is an alternative procedure of giving corrective feedback, which refers to a native s pe a ke r ’ sr e wr i t i ngofaL2 l e a r ne r ’ sc ompos i t i onwithout changing. 政治大 native-like, the content the learner provides in the original draft is maintained, but its 立 the latter’ s ideas. As proposed by Levenston (1978), to make a composition more. awkwardness, rhetorical inadequacy, ambiguity, logical confusion, style, and so on, as. ‧ 國. 學. well as lexical inadequacy and grammatical errors are tidied up.. ‧. Inspired by Levenston, Cohen incorporated reformulation in his studies in1982,. sit. y. Nat. 1983, and 1989 with higher-intermediate and advanced L2 learners. He contended. io. er. that reformulation could benefit L2 writers in vocabulary, syntax, paragraphing, and cohesion. The student-writers, who received the reformulation procedure in his 1982. al. n. v i n Ctohdiscern problems U and 1983 studies, were found of cohesion, grammatical rules, engchi. precision in the use of vocabulary, as well as differences in levels of formality in the target language. Qi, & Lapkin (2001) highlights that reformulation is superior to other kinds of CF in three aspects: 1. It provides the structures most relevant to the learner’ s personal needs and interests since they are tailor-made for his or her original ideas and intentions. 2. It offers appropriate forms for those who have little experience with the target language and who cannot figure out solutions for themselves. 3. It can be an activity either meaning-driven or form-focused, or both, 12.

(22) depending on the task design and chosen text. In addition, the reformulation procedure seems in accord with Gass’(1988) comprehended input. That is, the reformulated version of L2/FL writing serves as a kind of comprehended input for the learners. Since it is directly related to their previous output, according to Myers (1997), this later input, or intake, can particularly evoke their personal responses. When language learners evaluate it with their original intentions and knowledge of rules, they may increase their awareness about their. 政治大 grammar to express their ideas. 立. common mistakes, and think about how they could have used certain structures and. s 1983 study, when compared with direct correction, the Even so, in Cohen’. ‧ 國. 學. reformulation treatment was found less successful, particularly in the areas of syntax. ‧. and cohesion, where it was supposed to excel. Therefore, he explained that only the. sit. y. Nat. learners at higher-intermediate levels and above appeared to benefit more in the. io. er. process of reformulation. In the meantime, he considered that the reformulation technique needed refining, suggesting that further research could be conducted to. al. n. v i n C h feasible with beginning investigate whether this technique or intermediate students. engchi U. Sachs (2003), following Qi, & Lapkin’ s study in 2001, conducted two successive. studies, the first with 15 high-intermediate ESL university students and the second with 54 ESL university students from a variety of levels (none of whom participated in Study 1). She devised three writing conditions--explicit error correction, reformulation, and reformulation plus think-aloud--to investigate’the students’ evidence of noticing by having them compare their originals and the corrected, or reformulated versions. The research measures for Study 1 are demonstrated in Table 2.2, cited from Sachs, 2003, p. 64.. 13.

(23) Table 2.2 Three-day sequences of the three experimental conditions (cited from Sachs, 2003, p. 64) Condition. Tuesday (30 min). Thursday (15 min). Friday (20 min). Error Correction. Write a 30-minnute picture description.. Look at explicit error corrections of the essay.. Revise a clean copy of the original essay.. Reformulation. Write a 30-minnute picture description.. Compare the essay to a reformulated version.. Revise a clean copy of the original essay.. Reformulation + Think-Aloud. Write a 30-minnute picture description.. Compare the essay to a reformulated version while thinking aloud.. Revise a clean copy of the original essay.. 政治大. In this study, each of the fifteen participants had to go through the three writing. 立. conditions by turns. In every condition, each wrote one story describing a set of comic. ‧ 國. 學. strips; that is, every participant composed three different stories. Two raters coded every story and found 3481 errors, categorized into 40 types. After the error coding. ‧. and correction (grammatical errors were corrected, style and cohesion improved, and. y. Nat. sit. some more accurate vocabulary introduced), reformulations of the original stories. n. al. er. io. were typed on separate sheets of paper and given back to the participants in the. i n U. v. reformulation and think-aloud sessions. The next day, based on clean copies of the. Ch. engchi. originals, the participants rewrote the same comic strips without any notes. For data analysis, Sachs (2003) divided each original, as well as its revision, into T-units, and each T-unit (the shortest unit which a sentence can be reduced to, and consisting of one independent clause together with whatever dependent clauses are attached to it) was coded for evidence of noticing, i.e. “ revision accuracy,”to observe whether there was any change in the revised version--at least partially changed (+), completely corrected (0), completely unchanged (-), or not applicable (n/a). Her research results suggested, “ error corrections were the most effective in promoting changes in accuracy at the level of T-units, followed by reformulations, and finally 14.

(24) think-alouds”(2003, p. 69). Sachs (2003) implemented the same research procedures in Study 2, where forty-four participants were either high intermediate or advanced ESL university students, and the other ten came from an ESL class at a local community college. The results again indicated that error correction outperformed reformulation in light of linguistic accuracy in revisions. In other words, both Sachs’study results came opposed to her initial assumption--reformulation would be more helpful than explicit. 政治大 have affected the study results. As the participants in the Error Correction condition 立. error correction. Therefore, Sachs further discussed that “ memory”and “ time”could. did not have to search for differences or talk about what they were doing, they. ‧ 國. 學. probably had more time to understand and to remember the corrections.. ‧. Research Questions. sit. y. Nat. Reformulation has been investigated in the L2 context where learners have. io. er. frequent use of English for communication and have access to an abundance of English input (Cohen, 1982, 1983; Qi, & Lapkin,2001;Sachs, 2003). At the same time,. al. n. v i n C hthe participants atUhigh intermediate L2 level or above. it has been found effective with engchi Nevertheless, little literature, if any, has targeted at the effects of reformulation on. EFL high school low-achievers. These English learners are unable to write an accurate English sentence though exposed to formal English instruction for at least five years (Lee, 2009). This may be owing to the fact that they have little access to communicative English use (Fotos, 2001). These students may need to become more conscious of their English use in written form. Therefore, further research should be undertaken to determine whether reformulation can benefit EFL low-achievers. Considering these students’poor proficiency, the present study is conducted to investigate the effects of reformulation on their paragraph writing, and is intended to 15.

(25) focus on the grammaticalness (form) of their writing products. It is expected that these participants may boost their consciousness of form-meaning correspondence in their use of the target language, English. The questions to be investigated in this research are listed as follows: 1. Is reformulation, compared to direct error correction, more helpful in Taiwan c hi e ve r s ’paragraph writing? EFL high school low-a 2. Among Taiwan EFL high school low-achievers, do those with better English. 政治大 lower English writing proficiency? 立. writing proficiency benefit more from reformulation in writing than those with. 3. Is there any change in the errors made by EFL low-achievers after. ‧ 國. 學. reformulation?. ‧. 4. What do the students think of reformulation as a way to improve writing?. n. er. io. sit. y. Nat. al. Ch. engchi. 16. i n U. v.

(26) 17. CHAPTER 3 METHODOLOGY This chapter describes how the research is designed to investigate whether reformulation can make any positive effect on senior high school EFL low-achievers in paragraph writing in Taiwan. Participants The participants in the present study were sixty Ta i wa n’ s freshmen from a private senior high school in northern Taipei. They were selected from two classes of. 政治大. 40 according to the test results from the English subset of the Basic Competence Test. 立. for Junior High School Students (EBCT) in July, 2009. Based on the EBCT scores,. ‧ 國. 學. when the participants were assigned, ten students from each class who might affect the homogeneity of grouping were excluded from the list of the participants before the. ‧. treatment even though they still took part in all the class activities.. y. Nat. io. sit. The participants, 32 boys and 28 girls, were 15 to 16 years old. They had. n. al. er. received formal English instruction in the EFL context for at least five years (2 years. i n U. v. in elementary school and 3 years in junior high). While the full marks for EBCT were. Ch. engchi. 80, the mean of the participants’scores appeared relatively low (25.78). This suggested that the participants were low-achievers. In the meantime, the standard deviation (11.335) indicated that the scores were widely distributed among the participants. Ten of them gained EBCT scores between 1 and 14; fourteen scored from 15 to 23; sixteen had scores ranging from 24 to 30; twelve received scores from 32 to 42; and, eight scored between 44 and 51. Then, based on the results of the writing pre-test (see Table 3.2), these students were divided into two groups, one experimental group and one control group. The experimental group consisted of the students whose writing scores were similar to.

(27) those in the control group. In addition, two participants in the control group were absent on the day when the post-test was administered. Therefore, two experimental participants with pre-test writing performances similar to the control absentees were excluded from the statistics to make the experimental and control groups identical both in student number (28) and in writing ability. In addition, all the participants were instructed by the same English teacher. None of them had received formal English writing instruction before. During the. 政治大 while the control group received direct error correction. 立. experiment, the experimental group detected and recorded their own errors in writing,. Instruments. ‧ 國. 學. In order to answer the four research questions, the following instruments were. ‧. employed in the present study: (a) the 2009 EBCT; (b) a GEPT elementary level. sit. y. Nat. writing test; (c) the GEPT holistic scoring guide; (d) the reformulation technique; (e). io. al. n. interview.. er. an error record form; (f) the error classification system; (g) a semi-structured. i n C h EBCT in 2009 engchi U. v. The present study aimed to investigate EFL low-achievers’paragraph writing in their first semester in senior high. Therefore, the participants’EBCT scores were examined to determine their English performance before the fall semester. The EBCT, held every July, was a national standardized test designed to examine how Taiwan junior high school graduates perform in English after three years of formal English instruction in junior high. It was constituted of 45 multiple-choice questions testing vocabulary, idioms and phrases, grammatical usages, and reading comprehension. In the present study, the EBCT scores were referred to as the participants’English achievements. Considering 80 as the full marks, the test takers who scored far below 18.

(28) it were defined as low-achievers. GEPT Elementary Level Writing Test To investigate improvement in writing, a paragraph writing test, adopted from an elementary level test of General English Proficiency Test (GEPT), was employed twice with an interval of ten weeks. Namely, the same writing test was conducted as the pre-test and the post-test. The GEPT was a standardized test developed by the Language Training and Testing Center (LTTC) in Taiwan. With the support of the. 政治大 indication of its test takers’general English proficiency. Therefore, GEPT was 立. Ministry of Education, its test results had been accepted widely in Taiwan as an. adopted in the present study to test the participants’paragraph writing ability.. ‧ 國. 學. The results from the pre-test were used to determine the participants’initial. ‧. ability in paragraph writing before any treatment. Meanwhile, the scores were used as. sit. y. Nat. reference to divide the participants into one experimental group and one control group. io. er. so that the two groups appeared homogeneous in terms of initial writing ability. Then four weeks after the treatment, the same test was administered as the post-test to. al. n. v i n C h between the two compare the extent of improvement groups. engchi U. This test lasted for thirty minutes and required the test takers to write 50 English. words to describe the four pictures on the test paper (see Appendix A ). The test takers were allowed to write as many words as they could in order to convey their ideas completely. Additionally, those at really poor English levels were encouraged to write as much as possible during the thirty-minute test. GEPT Holistic Scoring Guide The GEPT writing scoring guide, researched and developed by the LTTC, was adopted in the present study to evaluate the participants’writings in the pre-test and post-test. As displayed in Table 3.1, a writing product would be assessed as level 0 to 19.

(29) level 5 in consideration of topic development, grammar, wording, spelling, capitalization, and punctuation, as a whole. Table 3.1 LTTC Writing Scoring Guide (translated from Chinese) Level. Criteria. 5. Topic is richly and fully developed. Grammar and wording appear nearly error free.. 4. Topic is generally clearly and completely developed. Grammar and wording errors are not distracting.. 3. Topic is developed clearly, but not completely. Grammar and wording errors are sometimes distracting.. 2. Topic development is present, but limited by incompleteness, lack of clarity, or lack of focus. Grammar and wording errors are distracting.. ‧ 國. ‧. Topic development has only one point of view or one dimension. Grammar and wording errors are overarching and seriously distracting.. y. Nat. io. n. al. sit. Nothing or little is written.. er. 0. 學. 1. 立. 政治大. Ch. i n U. Reformulation. engchi. v. To probe the effect of noticing (Schmidt, 1990), the researcher reformulated, or rewrote, the writing products from the experiment group in correct English form without changing the authors’original ideas. Reformulation, used in several studies (Cohen,1982, 1983; Qi, & Lapkin, 2001; Sachs, 2003), originally involved both rhetorical and grammatical factors in writing. However, in the present study, given the low achievers’language ability, the technique of reformulation was adapted for the participants to target at their use in English form, and was expected to decrease the grammatical errors in sentence construction. In addition, with the limitation of resources (hiring a qualified native speaker can cost a fortune), the reformulater was 20.

(30) the non-native English teacher-researcher instead of an English native speaker. Error Record Form The error record form (see Appendix B), designed by the researcher, was used by the experimental group to record their errors found through comparison between their original sentences and the reformulated ones. The records were regarded as evidence that they had noticed the differences in form between their interlanguage and the target language. In recording the errors, the experiment group might observe how they. 政治大. improved in accuracy of English sentences and could become reinforced to write more accurate sentences.. 立. This form was given and demonstrated to the experimental group when the. ‧ 國. 學. reformulated versions were given back to them. The list of directions on the form,. Error Classification System. er. io. sit. y. Nat. identification.. ‧. along with the teacher’ s oral instructions, led the students to work on individual error. To analyze the change in the participants’errors, the error classification system. al. n. v i n C hSachs (2003, p. 112-114), (see Appendix C), adapted from had been extensively engchi U. simplified to enhance its feasibility in the present study. To make the participants fully perceive what error they made, the simplified classification system retained only the most common 18 error types found in the participants’English writing. Moreover, the grammatical terms were presented in Chinese, and were limited to those familiar to the participants. Such modification was intended, firstly, to match up the participants’ limited cognition of the form of the target language; secondly, to avoid the participants’pressure and panic from difficult grammatical classification terms. That is, the students were expected to focus their attention only on the errors themselves rather than on the terminology. This modified error classification system was 21.

(31) employed to examine the errors produced in the pre-test and post-test. It aimed to classify those errors and to help indicate the change in the distribution of each type of error before and after the treatment. Interview A semi-structured interview was employed to investigate how the experimental students felt about reformulation. The questions were listed as below: 1. What do you think of reformulation as a way to improve your writing?. 政治大 3. In reformulation, what is difficult for you? 立 2. In reformulation, what is easy for you?. 4. Do you have any suggestions about reformulation?. ‧ 國. 學. 5. When writing the picture description in the post-test, what do you find about. ‧. your writing?. sit. y. Nat. During the interview the students’answers were tape recorded for later transcription.. io. al. Procedures. er. The transcripts were coded into several categories for further analysis.. n. v i n The research procedureC comprised six steps: sampling and orientation, pre-test, hengchi U. rating and grouping, treatment, post-test and rating, and interview. As to the treatment, it was made up of five writing sessions (see Appendix D). Totally speaking, it took fifteen weeks to finish the six steps: Steps 1, 2, and 3 respectively for one week, Step 4 five weeks; Step 5 conducted four weeks after the treatment took two weeks, and the final step one week. The ten-week interval between the pre-test and post-test were devised to minimize the influence of memory, as discussed in Sachs’studies (2003), and to serve the purpose of examining whether the participants noticed and decreased erroneous use of their original output.. 22.

(32) Step 1: Sampling and Orientation Before the fall semester, based on the EBCT scores, sixty low-achieving freshmen were selected from two classes of forty to participate in the present study (See the Participants section). In the first English writing class, both classes were informed that they would have a writing class every week. At this moment, the participants did not know that they would take a picture-description writing test in the following period. Step 2: Pre-test 政治大 In the second English class, the pre-test was administered to collect the data of 立. students’initial ability in writing. The students were asked to write 50 English words. ‧ 國. 學. to describe four pictures on the test paper. Before the 30-minute picture description,. ‧. the whole class spent 3 minutes discussing the content of the pictures of a story. By. sit. y. Nat. doing this, everyone might have a clear idea of what to write about the pictures. Then. io. er. they could focus on the use of English rather than think about the plot in the pictures. With the help of one colleague of the teacher-researcher, the two participating. al. n. v i n C h All the test takersUwere instructed to read the classes took the test concurrently. engchi. instructions very carefully before they began to write any sentences. Dictionaries, translators, and peer help were not allowed. The writing test papers were not returned to them until the end of the experiment. Step 3: Rating and Grouping The products of the picture description test were first rated by the teacher-researcher and one fellow English teacher before the next writing session. Rating was divided into two stages--holistic scoring and error counting. Such device was implemented after both the pre-test and the post-test. On one hand, the results of holistic scoring were used to answer the first and second research questions 23.

(33) concerning the comparison between the experimental group and control group. On the other hand, error counting dealt with the third research question regarding error change after the treatment. The method of error counting was explained as follows. Since research question three targeted at the change in error between the pre-test and the post-test, the error counting method was devised by referring to Weltig (2004), and the procedure went as follows. First, the numbers of and the types of error in each piece of writing were recorded and counted. Second, the error total was tallied by. 政治大 group. Next, the percentage of each error type was calculated with the following 立. adding up the numbers of the 18 types of errors (see Appendix C) in the experimental. formula: the number of one type of error was divided by the number of all the errors.. ‧ 國. 學. For example, the error total was 933 and the errors of verb formation appeared 54. ‧. times. The error density was thus calculated as (54/933)*100%, which was 5.79%.. sit. y. Nat. The error percentage gained in this way would be used later for observation of error. io. Intra-rater Reliability. er. change before and after the treatment.. al. n. v i n To check the intra-raterC reliability, ten of these h e n g c h i Uwritings were assessed two weeks. later, which was the second rating. Each of the two raters independently gave scores. based on the GEPT scoring criteria, as she had done for the first rating. All the scores were recorded on separate sheets of paper; that is, the ten writings were kept clean without any marks. When one rater scored the same ten papers, she did not well remember what levels they were assessed as by her for the first rating (since it was two weeks ago). As for each rater, two sets of grades produced from the first rating and the second rating were respectively calculated with reliability analysis (Cronbach's Alpha value of SPSS 12.0). For the researcher’ s rating, there was a strong and positive 24.

(34) correlation between the two ratings since the correlation coefficient (r) was .945 (-1≦ r ≦ 1). For the other rater, the reliability reached .953 (-1≦ r ≦ 1). Inter-rater Reliability By referring to the GEPT scoring guide, two clean photocopies of the same ten pre-test writings were scored by the two independent raters individually on the day after the pre-test. The two sets of scores were later analyzed for inter-rater reliability by means of Cronbach's Alpha value of SPSS 12.0. The correlation coefficient (r). 政治大 correlation between the ratings of the raters. Namely, there was great agreement 立 was .947 (-1≦ r ≦ 1). This result indicated there was a close and positive. between the raters in operating the GEPT scoring guide, which led the rating to be. ‧ 國. 學. highly reliable.. ‧. Next, the participants’pre-test scores were gained by calculating the means of. sit. y. Nat. the scores from the two raters. For instance, student A was rated as level 4 by one. io. er. rater and as level 3 by the other, so his pre-test writing scored 3.5. These scores were then used to select 56 participants out of the 80 students into the experimental group. al. n. v i n of 28 and the control group C of 28. Both were at a similar h e n g c h i U level of writing. Grouping. Before the experiment, two senior high freshman classes of forty had taken the pre-test where each student was asked to write a 50-word English paragraph describing four pictures (see Appendix A). However, only the 56 participants’pre-test results were compared by using independent-sample t-test (SPSS 12.0) and listed as below:. 25.

(35) Table 3.1 T-test Result for Pre-test Group. Means. SD. T-value. df. Class A (n=28) Class B (n=28). 3.83 3.45. .77 .57. 2.17. 54. ** = p < .005. T-value (2.17, p < .005) indicates no significant difference in writing performance between the two groups of students, so no adjustment was necessary to. 政治大. form one experimental and one control group. Thus, randomly, Class A was the. 立. 學. ‧ 國. experimental group and Class B the control group. Step 4: Treatment. ‧. All the paragraphs produced by the experimental group were rewritten by the teacher in correct English grammar without changing the writers’original ideas. There. y. Nat. io. sit. was no correction or comment marked on the experimental group’ s originals. In. n. al. er. contrast, those in the control group were given direct error corrections on their writings.. The Experimental Group. Ch. engchi. i n U. v. In the second writing class, the original paragraphs and the reformulated versions, as well as blank error record forms, were returned to the experimental group. Each participant in this group was asked to fill out the form while they were comparing the sentences in the originals with those in the reformulated versions. In this session, the students only had to examine the sentences related to the first picture. They were asked to identify and record their errors in the column labeled as observed errors. In the process, the teacher answered the questions from those who were unable to find out their errors or those who did not know how to correct their errors. In addition, all 26.

(36) the participants revised their problematic sentences based on their findings of the errors and their knowledge of English grammar. This whole activity took ten to fifteen minutes. After that, the originals, reformulations, and error record forms were collected by the teacher. Before the next session of writing, the teacher read through the error record forms from the experimental group. When she found that one participant had not discovered all of his or her errors on the error record form, she made such comments. 政治大 find them!”on the column labeled as Teacher’ s comments. These comments from the 立. as “ It’ s great for you to find one error, but the other two were left unnoticed. Try to. teacher could help make sure that every student in the experimental group found out. ‧ 國. 學. each error which he or she had made.. ‧. In the third session, all the sheets were given back to the participants. First, the. sit. y. Nat. students were asked to read the comments from the teacher. Then, some of them had. io. er. to perfect their error hunt for the first picture based on the teacher’ s comments, while the others examined the sentences describing the second picture and kept records of. al. n. v i n C experimental errors. Finally, when all in the completed the error hunt with the h e n g c group hi U. second picture, they gave all their sheets back to the teacher.. The same procedure continued until these students finished examining the sentences concerning the fourth picture. (See Appendix D) The Control Group In the second writing session, the control group received their corrected paragraphs. The teacher asked each student to read through his or her paragraph. Then she targeted at one serious error shared by most of the students. For example, many of the students might mistakenly put the accusative case of a pronoun where its genitive case should be. The sentence might go like “ The ball hit him eye.”After such 27.

(37) discussion, all the students were asked to revise where they found the discussed error in another sheet of paper. Five minutes later, all the sheets were returned to the teacher. In the third class, the teacher gave back all the sheets from the control group. Then she discussed another serious error made by many of the students. Soon after that, the students revised the discussed point. Finally all the written results, including the originals and revisions, were collected by the teacher.. 政治大. The same procedures went on in the following two weeks, as explicated in Appendix D.. 立. Step 5: Post-test and Rating. ‧ 國. 學. Four weeks after the treatment (Appendix D), the thirty-minute post-test was. ‧. administered to examine whether the participants had made any progress in accuracy. sit. y. Nat. of sentence construction. The test content was identical to the pre-test in order to. io. er. control the writing content for later comparison. This device was expected to suit the purpose of investigation into the extent to which the participants noticed the errors. al. n. v i n C h both classes tookUthe test at the same time. Before which they had ever made. Again, engchi. the test, all the test-takers were guided to read the instructions carefully. No dictionary, translator, or peer help was allowed in the process. Then the sentences were scored by the teacher-researcher with the same rating method. Step 6: Interview Two weeks after the post-test, the pre-test and post-test results, including the error rates and numbers, were given back to the experimental group. In this way, each experimental participant had a clear idea of how they performed in both tests and of whether they made any progress in writing. Immediately after that, a semi-structured interview was conducted with the experimental group students who scored top 5 and 28.

(38) those who were rated bottom 5. The talks were tape recorded and transcribed for further analysis. Data Analysis The methods for data analysis in the present study came by referring to Cohen’ s in 1983, Sachs’in 2003, and Weltig’ s in 2004. In Cohen’ s study, only the first 100 words of each essay were corrected and reformulated for comparative analysis. Then four respects of each 100-word excerpts--vocabulary, syntax, cohesion, and. 政治大 5-6/average; 7-8/good; 9/excellent). Finally, a t-test was employed to compare the 立. total--were holistically rated according to a nine-point scale (1-2/very poor; 3-4/poor;. writing performances in the four aspects between a reformulation group of 8 and a. ‧ 國. 學. correction group of 8, both made up of advanced learners of Hebrew as a second. ‧. language. However, it was problematic for the present study to assess the writings in. sit. y. Nat. the four categories as in Cohen’ s paper. Firstly, over 95 % of the participants wrote. io. er. less than 100 words and many of the sentences were in incomplete structure. Secondly, considering the participants’low proficiency, this present research targeted at change. al. n. v i n C h only the t-test method, in grammatical errors. Therefore, as with Cohen’ s study, was engchi U adopted in the present study.. As to Sachs’study in 2003, all the errors were coded in 40 types so that the observation would focus only on whether one participant noticed and corrected any of the 40 types of errors (namely, any newly produced error type was excluded from discussion). This device facilitated comparison between the originals and reformulated versions. Sachs’data analysis centered on comparison of revision accuracy between groups, and abandoned any further discussion on individual error types simply because that was problematic. The reasons were as follows:. 29.

(39) 1.. Newly produced error types could be introduced and then repeated through the revision.. 2.. Major or related errors could be memorized, spotted, and then corrected immediately.. 3.. Statistical analysis became difficult owing to individual participants’ differences in number of errors and in distribution of error type.. In agreement with Sachs, the present research focused on error change in group rather. 政治大 stage of linguistic development in mind (judged from their performances in writing), 立. than in individual participants. When it comes to error coding, with the low-achievers’. the 40 types observed in Sachs’research were simplified to 18 types, which were. ‧ 國. 學. common and understandable to the present participants with low proficiency.. ‧. Furthermore, the method of essay coding in Weltig (2004) was given special. sit. y. Nat. consideration, for it followed Sachs’coding of 40 error types, but concerned error. io. er. density, or frequency of errors of each type. For instance, error totals were 933 in Weltig’ s collected essays and the error of verb formation appeared 54 times. The error. al. n. v i n frequency of verb formationC was 54/933, which was h e n g c h i U 0.058. This statistical method. suited the need of the present study to count and compare the density of grammatical errors appearing in the participants’pre-test and post-test writings. By adaptation and integration of the data-analysis methods above, the data collected for the present research were analyzed as explained below in order of the four research questions. To answer the first research question, the data of the participants’scores were calculated by means of SPSS 12.0. That is, independent-sample t-test was employed to compare the mean scores of the two groups in the post-test. Research question two aims to compare the extent of improvement between the 30.

(40) students with better writing proficiency and those with lower proficiency. For the purpose, top ten and bottom ten participants in the experimental and control group were decided according to the pre-test results. Next, independent-sample t-test was used to examine whether there was significance between the pre-test and post-test results of the top ten students in both groups. The same procedure was done with the bottom ten. To answer the third research question, the number of each kind of error made by. 政治大 between the pre-test and the post-test. The result of comparison could indicate the 立. the experimental group was recorded and compared, quantitatively and descriptively,. change in error between the two tests.. ‧ 國. 學. To deal with research question four, all the experimental participants were. ‧. interviewed about how they felt about the treatment, about what difficulties they. sit. y. Nat. encountered in the process, and about what they found about their writing before and. io. er. after the treatment. Their replies were then categorized and summarized to help the researcher assess the reformulation technique and consider its practicality in the future. n. al. writing teaching.. Ch. engchi. 31. i n U. v.