Sie sind auf Seite 1von 14

Available online at www.sciencedirect.

com

Journal of Second Language Writing 17 (2008) 292305

Error correction, revision, and learning


John Truscott *, Angela Yi-ping Hsu
Department of Foreign Languages and Literature, National Tsing Hua University, Hsinchu 30013, Taiwan

Abstract Previous research has shown that corrective feedback on an assignment helps learners reduce their errors on that assignment during the revision process. Does this nding constitute evidence that learning resulted from the feedback? Differing answers play an important role in the ongoing debate over the effectiveness of error correction, suggesting a need for empirical investigation. In this study, learners rst wrote an in-class narrative and then revised their writing during the next class. Half the students had their errors underlined and used this feedback in the revision task while the other half did the same task without feedback. Results matched those of previous studies: the underline group was signicantly more successful than the control group. One week later, all students wrote a new narrative as a measure of (short-term) learning. On this measure, change in error rate from the rst narrative to the second, the two groups were virtually identical. Thus, successful error reduction during revision is not a predictor of learning (at least for the uncoded corrective feedback that has typied studies in this area), as the two groups differed dramatically on the former but were indistinguishable on the latter. Improvements made during revision are not evidence on the effectiveness of correction for improving learners writing ability. # 2008 Elsevier Inc. All rights reserved.
Keywords: Error correction; Revision; Learning; Correction debate

Introduction It is generally agreed that revision plays a central role in good writing, in terms of both content and form. Not surprisingly then, a considerable amount of research has been devoted to exploring issues regarding the revision process (see, for example, Ferris, 2006; Goldstein, 2006; Sachs & Polio, 2007). Our present concern, however, is with only a small subset of these studies, those that investigated the effects of teachers form-based feedback on students success in the revision process (Ashwell, 2000; Fathman & Whalley, 1990; Ferris & Roberts, 2001). In these experiments learners were asked to revise their writing, some with the benet of written error

* Corresponding author. E-mail addresses: truscott@mx.nthu.edu.tw (J. Truscott), angela@mx.nthu.edu.tw (A.-p. Hsu). 1060-3743/$ see front matter # 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.jslw.2008.05.003

J. Truscott, A.-p. Hsu / Journal of Second Language Writing 17 (2008) 292305

293

correction and some without, and the effects of the feedback were measured by the extent to which they managed to improve the accuracy of their essays. An additional study, that of Lee (1997), is sometimes included in this category, though it differed from the others in that the task did not involve revision of the students own writing but rather identication, classication, and correction of errors implanted in a newspaper article. One clear conclusion from this research is that teachers corrections do indeed help learners reduce their errors: the revised manuscripts of students who received it showed signicantly more improvement in accuracy than those of students who did not receive it. This research also produced a number of interesting ndings on related issues, such as the relation between feedback on form and feedback on content or the timing of the two types. Our concern, however, is not with these results but rather with the question of how research on revision relates to broader issues regarding the role of error correction in writing classes. In particular, we are concerned with the differing positions that have appeared in the literature on the use of grammar correction as a teaching device and with the place of the revision research in evaluations of these positions. One view (see especially Truscott, 1996, 1999, 2007) holds that correction makes little or no contribution to the development of accuracy in writing, possibly even harming the learning process, and therefore has no place in writing classes. The other (see especially Ferris, 1999, 2003, 2004) takes a more favorable view of correction, recommending its use in writing instruction. Not surprisingly, the two sides in this debate have offered very different interpretations of the revision research, and these differences have become an important part of the debate. The disagreement is about whether the ndings of this research constitute evidence of learning, which we will dene as improvements in learners ability to write accurately. We will not be concerned here with the question of whether such improvements represent new knowledge or simply priming of existing knowledge; the issue is simply whether learners become better writers as a result of the treatment. View #1: error reduction during revision is not a measure of learning Truscott (1996, 1999, 2007) took the position that the revision research had no implications for the issue he was addressing, because the point of his case against grammar correction was that correction is not useful as a teaching device, while the revision research was about its usefulness specically as an editing tool, a way to improve a particular manuscript. Evidence of its value in this function does not constitute evidence of its value for learning, the argument goes, because evidence on learning necessarily involves a comparison between two independently written works. So the revision research is not relevant to the case against grammar correction. A similar view was expressed by Ashwell (2000) in discussing the relation between his study of revision on the one hand and the case against grammar correction on the other. He argued that the two were unrelated, because the goals of correction in the two cases were entirely different: he was studying its role in improving a particular manuscript, not as a teaching device. View #2: error reduction during revision is a measure of learning A sharply contrasting view has been offered by a number of other authors, for whom the revision research is an integral part of efforts to understand the value of correction as a teaching tool. The rst was Sheppard (1992), who carried out an experiment comparing a class of ESL students who had all their formal errors corrected with a similar class that received feedback only on content and clarity. After 10 weeks of this treatment, he found an advantage for the

294

J. Truscott, A.-p. Hsu / Journal of Second Language Writing 17 (2008) 292305

no-correction group. He then contrasted this negative nding with the positive outcome of Fathman and Whalley (1990), discussing the possible reasons why the two studies obtained such different results. In doing so, he took it for granted that Fathman and Whalleys ndings, like his own, were in fact a measure of learning, and he looked elsewhere for explanations of the contrasting outcomes. This view was echoed in Chandlers (2003, 2004) discussion of the literature. She suggested that the two studies might have yielded contrasting results because Sheppard (1992) looked only at selected errors while Fathman and Whalley (1990) were more comprehensive. In her general discussions of the grammar correction debate, she gave a prominent place to the revision studies. Like Sheppard, she saw this research as signicant evidence regarding the effects of correction on learning. Similarly, Fazio (2001) discussed the negative ndings of three experiments on the effects of correction on students writing ability (Kepner, 1991; Semke, 1984; Sheppard, 1992) and presented Fathman and Whalleys (1990) results as contradictory evidence. The explanation she suggested for this contrast was that Fathman and Whalley studied a process-oriented writing class while the others looked at skill development classes. It was taken for granted that the revision study, like the experiments with which it was contrasted, did in fact measure learning. The same position was adopted by Russell and Spada (2006), who carried out a meta-analysis of research on the effects of error correction, with the goal of determining how it affects learning. Their sample of 15 studies included three that focused on improvements in accuracy made during revision (Ashwell, 2000; Fathman & Whalley, 1990; Lee, 1997), the results of which were averaged in with work that uncontroversially measured learning, to produce general conclusions about the effect of correction on learning. These three studies yielded very large effect sizes and therefore played an important role in the overall ndings. Thus, the question of whether they actually measured learning is quite important for interpretation of this meta-analysis. Dana Ferris, the most prominent advocate of grammar correction, has discussed this research more extensively. Her main point appears to be that it provides good evidence regarding shortterm learning that results from error correction. Ferris (1999) treated Fathman and Whalleys (1990) ndings as problematic for Truscotts (1996) case against grammar correction. Ferris (2003) argued that the signicance of Fathman and Whalleys ndings was limited by the shortterm nature of the experiment, but that it was nonetheless the best evidence available that correction is effective, the issue being whether the observed benets would hold up in the long run. This concern with short-term versus long-term effects of the treatment again contradicts View #1, on which the observed benets were limited to changes in the manuscripts themselves, ruling out any issue of whether they would endure. So this discussion can be taken as an implicit rejection of View #1. Ferris (2004) discussed criticisms that had been made of revision studies, suggesting that the point of these criticisms was that the studies were not longitudinal. More recently, Guenette (2007) echoed Ferriss view, presenting these studies as useful evidence while leaving open the question of whether the effects they found were only short-term. For our purposes, the essential point here is that the revision research is taken as evidence regarding the impact of correction on learning. Thus, two fundamentally opposed views of the revision research have been put forward, with important consequences for the debate over the effectiveness of error correction. One view is that this research does not measure learning at all. The other is that it provides at least a valid measure of short-term learning resulting from correction and possibly an indicator of long-term improvement as well.

J. Truscott, A.-p. Hsu / Journal of Second Language Writing 17 (2008) 292305

295

There are no signs of any resolution to this clash of interpretations, and a reasonably safe conclusion at this point is that further discussion will not produce one. What might do so is empirical evidence regarding the relation between error reduction in revision on the one hand and improvements in writing ability on the other. If the studies that measured improvements in accuracy during the revision process had also included an uncontroversial measure of learninga comparison between accuracy on the initial writing and that on a new writing taskwe would be able to compare the two and draw conclusions about this relation. A nding that they yield very similar results would constitute evidence that the revision studies can indeed be taken as evidence of learning. On the other hand, a nding that they are entirely unrelated would imply that ndings obtained entirely during revision cannot be taken as evidence of learning. But past studies did not include any uncontroversial measure of learning, so no such comparison is possible. What is needed then is research that looks at the effects that error correction has during the revision process, in essentially the same way previous authors have done, but which also includes a second writing task. A comparison can then be made between gains made during revision and gains made from one independent writing task to the other, to determine if success during revision is or is not a genuine indicator of learning. The second task should probably not be carried out immediately after the revision, as this approach could introduce a fatigue factor. However, it should not be so long after the rst that additional experience with the target language could contaminate the results, possibly overwhelming the effects of the single episode of error correction and revision. Our study is an effort to ll the need for research of this sort. We began with an in-class narrative writing task, followed the next week by an in-class revision of the narrative, both tasks comparable to those used by Fathman and Whalley (1990). Half the participants had their errors on the initial writing underlined and were able to use this information while revising, while the other half did their revisions without the benet of any feedback. A week later, all were asked to write a new narrative on a different subject, providing a measure of any differential learning that had occurred as a result of the presence or absence of correction. Methods Participants and setting Forty-seven EFL graduate students, 38 males and 9 females, from a large public university in Taiwan participated in this study. Twenty were from the College of Electrical Engineering and Computer Science, 10 from the College of Science, 10 from Engineering, 5 from Life Sciences, 1 from Nuclear Science, and 1 from Technology Management. These students were enrolled in three sections of an elective basic writing seminar, which oriented them to the use of writing conventions and standards including general attributes of writing (clarity and conciseness) as well as basic structure (thesis statement, specic support, organization, unity, coherence, and cohesion). Specic genres covered in this course were narration, description, argumentation, comparison/contrast, and process. Each class met for 3 hours once a week for 18 weeks. Prior to registering for the basic writing course, students were required to take a diagnostic test extracted from the guided writing section of the high-intermediate General English Prociency Test (GEPT). Test results were used to place students with similar writing prociency in appropriate levels of the course in order to maximize the value of teaching. The GEPT, a criterion-referenced test, was developed by the Language Training and Testing Center (LTTC) in Taiwan. It has been widely used as a means of determining learners English abilities for admission and placement, as well as a criterion to evaluate whether or not a student has met certain graduation requirements.

296

J. Truscott, A.-p. Hsu / Journal of Second Language Writing 17 (2008) 292305

Students were given 40 minutes to write an essay of 150180 words on whether they supported or opposed the Public Welfare Lottery, and they also discussed its impact on society as a whole. Each student essay was then graded by two experienced writing instructors based on the grading criteria developed by LTTC. Students who scored from 30 to 42 (out of 60) on their essays were placed into this course. These students demonstrated skills of basic writing structure and fair use of highfrequency vocabulary but failed to produce cohesive sentences and coherent paragraphs. Two instructors with many years of experience teaching EFL writing taught the course sections, one of them teaching two sections. The other instructor, one of the researchers for this study, taught the third section of the course. This instructor-researcher provided support and ensured consistency among the three sections. Working closely together, the two instructors maintained the same instruction (the course objectives, content, activities, and materials were consistent), and data collected for this study were collected in the same manner. In order to retain objectivity in the marking and rating of written data, all scoring was done by the other researcher, who did not teach any of the sections. Data collection Students enrolled in these courses were divided equally into two groups (a control and an experimental group) based on their Narrative 1 error rates. To form the two groups, students in each section of the course were ranked from high to low according to these error rates. Those with odd-numbered rankings were placed into the experimental group, whereas even-numbered students were placed in the control group. The oddeven number designation was used ONLY for creating the two groups and for determining the level of feedback students would receive on the two in-class writing assignments used in this study, which are described in the following paragraphs. These two writing samples were not graded. Throughout the rest of the semester, there was no such division. All students received the same in-class instruction and assignments; written feedback on all their regular assignments was given in the same manner. Students were informed at the beginning of the semester that their writing products would be analyzed and used for research and teaching-improvement purposes. After 11 weeks of instruction in the genres of narration, description, and argumentation, students were quite comfortable with and familiar with the procedures, pacing, and activities in the course, including writing during class time. Therefore, data were collected from an in-class writing assignment conducted during weeks 1214. Students viewed a sequence of eight pictures provided by the instructors. The pictures which provided the prompt for Narrative 1 showed the story of two school boys who met in front of their school after class. They then went to a bookstore to buy a birthday card for their friend. While walking to their friends place, they ran into their other friends who were also heading to the birthday party. The ending of this story is a joyful birthday celebration. For Narrative 1, the students in all three sections were given 30 minutes to write a guided narrative story based on the pictures. A sample student narrative is provided in Appendix A, with errors marked. In week 13, students Narrative 1s were returned, and they were given 30 minutes to revise them (see Appendix B). Students were specically told not to write a new essay; instead, they should improve the one they had already produced in week 12. Students who were in the experimental group received their Narrative 1s back with errors underlined; students in the control group received their drafts without any marks on them. Neither group had grades issued on their drafts. In week 14, students again were provided 30 minutes to write a guided narrative story (Narrative 2) based on a new series of eight pictures. This story was about a little boy who

J. Truscott, A.-p. Hsu / Journal of Second Language Writing 17 (2008) 292305

297

ran out of the house after being scolded by his mom. The boy called a friend of his and went shopping for jeans. His friend put a pair of jeans in his backpack without paying for them. The sensors set off the alarms as the two boys were leaving the store. They then ended up in jail for shoplifting. A sample of Narrative 2 is given in Appendix C, written by the same student whose Narrative 1 and revision are provided in Appendices A and B, respectively. All three writing exercisesincluding Narrative 1, Narrative 1 revisions, and Narrative 2 were collected for further analysis. Only those students who completed all three tasks were included in the data analysis, and one additional student was excludedbefore the Narrative 1 revisions were donedue to an exceptionally high error rate on Narrative 1. The experimental group ultimately consisted of 21 students and the control group 26. Marking of errors On each of the three writing exercises errors were underlined in red immediately after students completed the exercises. Students in the experimental group did their Narrative 1 revisions using the markings, while those in the control group were instead given unmarked copies of their original narratives. After the errors were marked, each piece of writing was assigned an error rate: the total number of errors divided by the total number of words written. Marking was done by the researcher who did not teach any of the sections, and the procedure was entirely blindat no time did the scorer have any knowledge of whether a writing sample was produced by a student in the experimental group or by one in the control group. In order to ensure the reliability of the scoring, 10% of the writings were graded a second time, 6 months later, by the same scorer. The intra-rater correlation for the two markings of the same narrative was .967, with S.D. = .066. The corresponding 95% condence interval is (.837, 1.000). Selection of error types to be marked was based on the need for both consistency and broad coverage. All grammatical errors were included, but errors in word choice were not, except when they involved the choice of a function word (particularly determiners, prepositions, and transitions) and could therefore be considered grammatical problems. Errors in mechanics in general were not counted; exceptions were those associated with fragments and run-on sentences and failure to use a question mark instead of a period. Spelling errors were marked. This approach yielded a substantial number and range of marked errors while maintaining a very high level of consistency, as was shown by the second scoring 6 months later. Results Descriptive statistics for the two groups on each of the three measures are shown in Table 1. In each case, standard deviations are given in parentheses. In order to ensure that students in the two groups began the study with similar writing prociency, an analysis of variance (ANOVA) was performed on the error rates for Narrative 1. Results are shown in Table 2.
Table 1 Mean error rates for experimental and control groups Groups Experimental Control n 21 26 Narrative 1 .0799 (.0367) .0763 (.0389) Revision .0474 (.0255) .0680 (.0447) Narrative 2 .1130 (.0472) .1095 (.0537)

298

J. Truscott, A.-p. Hsu / Journal of Second Language Writing 17 (2008) 292305

Table 2 Analysis of variance for error rates: Narrative 1 Source of variation Between groups Residuals Total (corrected) d.f. 1 45 46 SS .00015 .06472 .06487 MS .00015 .00144 .00141 F .1045 p-Value .7480

Table 3 Analysis of variance for error reduction: Narrative 1 to Revision Source of variation Between groups Residuals Total (corrected) d.f. 1 45 46 SS .0068 .0361 .0429 MS .0068 .0008 .0009 F 8.4998 p-Value .0055

The difference between the error rate for the experimental group (.0799) and that for the control group (.0763) fell far short of signicance (F = .1045, p = .7480). We thus conclude that the groups were equal in their initial writing prociency. To examine the effect of error feedback on students subsequent revisions, a comparison was made between the two groups on their reduction in error rates from Narrative 1 to the revisions. The results are shown in Table 3. The difference in error reduction between the two groups is signicant (F = 8.4998, p = .0055). Students who had their errors underlined performed better on the revisions than those who did not. In other words, error feedback had a signicant effect on students rewrites. The error rate for Narrative 2 was then compared with that for Narrative 1. Results are shown in Table 4. As can be seen there and in Table 1, the two groups were virtually identical on this measure (F = .0002, p = .988), indicating that the corrections did not have an effect on students writing development. Whether students received corrections on their drafts did not seem to inuence their writing performance on the subsequent assignment. The error rate in Narrative 2 was considerably higher than that in Narrative 1, suggesting that Narrative 2 was harder for students to write. To preclude the possibility that this difference inuenced the results, the scores were standardized and analyzed a second time. (For discussion of standardization and its uses, see Hatch & Lazaraton, 1991.) Table 5 shows the results of this additional analysis. As can be seen, the test results based on the standardized scores are consistent with those based on the original scores. The above tests were performed under the assumption of normality. However, because this study used proportion data, the normality assumption may have been violated. Therefore, in addition to ANOVA, the Wilcoxon rank-sum test, a nonparametric testing procedure for which
Table 4 Analysis of variance for error reduction: Narrative 1 to Narrative 2 Source of variation Between groups Residuals Total (corrected) d.f. 1 45 46 SS .00000039 .07753157 .07753196 MS .000000390 .001722924 .001722932 F .0002261744 p-Value .9880675

J. Truscott, A.-p. Hsu / Journal of Second Language Writing 17 (2008) 292305 Table 5 ANOVA for error reduction (standardized): Narrative 1 to Narrative 2 Source of variation Between groups Residuals Total (corrected) d.f. 1 45 46 SS .0091 36.9650 36.9741 MS .0091 .8214 .8038 F .0111

299

p-Value .9165

Table 6 Rank-sum test statistics Test Error rate: Narrative 1 Error reduction: Narrative 1 to Revision Error reduction: Narrative 1 to Narrative 2 Z .396 2.4719 .0963 p-Value .6921 .0134 .9233

the normality assumption is not required, was conducted. The results, shown in Table 6, were consistent with the ndings reported in Tables 25. Once again, no meaningful difference was found between the two groups on Narrative 1 or on the change in error rates from Narrative 1 to Narrative 2, while a signicant difference was found in error reduction from Narrative 1 to the revision (favoring the experimental group). That is, the test results are reliable with or without the assumption of normality. Discussion First of all, our ndings conrm once again that correction does help students reduce their errors on the writing on which they receive the corrections, and that the effect is substantial. Our primary nding, though, is that the benets of error correction found on the revision task did not extend to a new writing task performed a week later; the students who had received correction on Narrative 1 and therefore were more successful in reducing their errors during revision did not differ from the students who had received no correction and therefore did not perform as well on the revision. In other words, no relation was found between success on the revision task and learning as measured by performance on a new writing task. The implication is that successful error reduction during revision is not a predictor, even a very weak predictor, of learning. These ndings, in themselves, do not constitute evidence against the effectiveness of error correction. They are entirely consistent with the view that feedback of this sortif provided over an extended period of timehas benecial effects, just as they are consistent with the view that it does not. Our goal here is not to make a direct contribution to the debate over the effectiveness of correction but rather to clarify the issue of what does and does not count as relevant evidence. The point to be drawn from our ndings is that studies which looked specically at error reduction during the revision process and did not include a second, independent writing task do not provide evidence on the value of error correction as a teaching device. Several possible limitations should be noted, though we do not believe that any of them offer a strong challenge to our conclusions. First, the possibility exists that learning occurred but was too

300

J. Truscott, A.-p. Hsu / Journal of Second Language Writing 17 (2008) 292305

short-term to be detected by our Narrative 2; perhaps the correction group learned a great deal but then forgot. But if this is the case then the learning was not a very interesting variety, as no traces of it could be found just 1 week after the treatment. Similarly, the possibility exists that extraneous variables differentially inuenced the two groups during the 1-week gap between the revision and the second narrative. Perhaps a substantial difference existed between the writing abilities of the two groups immediately after the corrections but was entirely eliminated by extraneous factors during the 1-week period that followed. But it is difcult to imagine what factors might produce such an effect, particularly given that the two groups had no existence outside our essentially random designation of certain learners as experimental subjects and others as control subjects. Another possible limitation on our conclusions is that we used only one type of feedback, marking location of errors but not giving information about their nature. Fathman and Whalley (1990) and Ashwell (2000) also used uncoded correction, so this point has no bearing on our conclusion that their results do not constitute evidence of learning. Ferris and Roberts (2001) tested both coded and uncoded correction, so the possibility exists that half of their ndings (those involving coded correction) are related to learning. In other words, it is conceivable that successful corrections made during revision result in short-term learning specically when they are based on coded corrective feedback rather than uncoded. But it is difcult to see why such a contrast would exist. Ferris and Roberts found the two correction types nearly identical in their value for reducing errors during revision (64% vs. 60% success), and the literature contains no apparent reason to think that one is superior to the other for learning. We conclude that while limitations exist, as they do in any single experiment, our ndings make a strong case for View #1: The revision studies reviewed offer no evidence regarding the effect of correction on learning. The debate over the effectiveness of error correction should thus be carried on either without reference to such studies or with a sharp distinction maintained between the value of correction for learning on the one hand and for improving a particular piece of writing on the other. Acknowledgements We wish to thank Professor Nan-Jung Hsu for extensive help with the statistical analysis, Chia-yu Joy Lin for making her two writing sections available to us, Melody Shen for drawing the pictures used in the writing tasks, and of course the students who participated in the study. References
Ashwell, T. (2000). Patterns of teacher response to student writing in a multiple-draft composition classroom: Is content feedback followed by form feedback the best method? Journal of Second Language Writing, 9, 227257. Chandler, J. (2003). The efcacy of various kinds of error feedback for improvement in the accuracy and uency of L2 student writing. Journal of Second Language Writing, 12, 267296. Chandler, J. (2004). A response to Truscott. Journal of Second Language Writing, 13, 345348. Fathman, A. K., & Whalley, E. (1990). Teacher response to student writing: Focus on form versus content. In B. Kroll (Ed.), Second language writing: Research insights for the classroom (pp. 178190). Cambridge: Cambridge University Press. Fazio, L. L. (2001). The effect of corrections and commentaries on the journal writing accuracy of minority- and majoritylanguage students. Journal of Second Language Writing, 10, 235249.

J. Truscott, A.-p. Hsu / Journal of Second Language Writing 17 (2008) 292305

301

Ferris, D. (1999). The case for grammar correction in L2 writing classes: A response to Truscott (1996). Journal of Second Language Writing, 8, 111. Ferris, D. (2003). Response to student writing: Implications for second language students. Mahwah, NJ: Erlbaum. Ferris, D. (2004). The grammar correction debate in L2 writing: Where are we, and where do we go from here? (and what do we do in the meantime?). Journal of Second Language Writing, 13, 4962. Ferris, D. (2006). Does error feedback help student writers? New evidence on the short- and long-term effects of written error correction. In K. Hyland & F. Hyland (Eds.), Feedback in second language writing: Contexts and issues (pp. 81 104). Cambridge: Cambridge University Press. Ferris, D., & Roberts, B. (2001). Error feedback in L2 writing classes: How explicit does it need to be? Journal of Second Language Writing, 10, 161184. Goldstein, L. (2006). Feedback and revision in second language writing: Contextual, teacher, and student variables. In K. Hyland & F. Hyland (Eds.), Feedback in second language writing: Contexts and issues (pp. 185205). Cambridge: Cambridge University Press. Guenette, D. (2007). Is feedback pedagogically correct? Research design issues in studies of feedback on writing. Journal of Second Language Writing, 16, 4053. Hatch, E., & Lazaraton, A. (1991). The research manual: Design and statistics for applied linguistics. New York: Newbury. Kepner, C. G. (1991). An experiment in the relationship of types of written feedback to the development of secondlanguage writing skills. Modern Language Journal, 75, 305313. Lee, I. (1997). ESL learners performance in error correction in writing: Some implications for teaching. System, 25, 465 477. Russell, J., & Spada, N. (2006). The effectiveness of corrective feedback for the acquisition of L2 grammar: A metaanalysis of the research. In J. M. Norris & L. Ortega (Eds.), Synthesizing research on language learning and teaching (pp. 133164). Amsterdam: Benjamins. Sachs, R., & Polio, C. (2007). Learners uses of two types of written feedback on a L2 writing revision task. Studies in Second Language Acquisition, 29, 67100. Semke, H. D. (1984). Effects of the red pen. Foreign Language Annals, 17, 195202. Sheppard, K. (1992). Two feedback types: Do they make a difference? RELC Journal, 23, 103110. Truscott, J. (1996). The case against grammar correction in L2 writing classes. Language Learning, 46, 327369. Truscott, J. (1999). The case for The case against grammar correction in L2 writing classes: A response to Ferris. Journal of Second Language Writing, 8, 111122. Truscott, J. (2007). The effect of error correction on learners ability to write accurately. Journal of Second Language Writing, 16, 255272.

302

J. Truscott, A.-p. Hsu / Journal of Second Language Writing 17 (2008) 292305

Appendix A. Narrative 1 sample

J. Truscott, A.-p. Hsu / Journal of Second Language Writing 17 (2008) 292305

303

Appendix B. Narrative 1 revision sample

304

J. Truscott, A.-p. Hsu / Journal of Second Language Writing 17 (2008) 292305

Appendix C. Narrative 2 sample

J. Truscott, A.-p. Hsu / Journal of Second Language Writing 17 (2008) 292305

305

John Truscott is a Professor in the Department of Foreign Languages and Literature at National Tsing Hua University in Hsinchu, Taiwan. His research interests include the effectiveness of grammar instruction and correction and applications of research and theory in linguistics and cognitive psychology to second language acquisition. His current research is primarily concerned with developing a processing-based account of language development, rst and second. Angela Yi-ping Hsu is an Assistant Professor in the Department of Foreign Languages and Literature at National Tsing Hua University in Hsinchu, Taiwan. Her research interests include plagiarism issues in ESL and EFL student writing, bridging the gap of research and teaching, as well as the correlation between teacher commentary and student writing development.

Das könnte Ihnen auch gefallen