Microsoft Word - brain_2_1.doc 17 The Effect of Recast vs. Self Correction on Writing Accuracy: The Role of Awareness Parviz Maftoon Islamic Azad University (Science and Research Branch) Faculty of Foreign Languages and Literature, Hesarak, Tehran, Iran, Postal Code: 1477893855 P.O.Box 14515 – 755, Tel. +98 311-6692696 p_maftoon@iust.ac.ir Masoumeh Ahmadi Shirazi University of Tehran, Faculty of Foreign Languages and Literatures North Kargar Avenue, Tehran, Iran. Postal Code: 14174 shirin_shirazi@yahoo.com Parisa Daftarifard Islamic Azad University (Science and Research Branch) Faculty of Foreign Languages and Literature, Hesarak, Tehran, Iran, Postal Code: 1477893855 P.O.Box 14515 – 755, Tel. +98 311-6692696 pdaftaryfard@yahoo.com Abstract Studies in support of the facilitative role of correction in either forms of recast, repetition, confirmations, compression checks, or clarification request abound (Ellis, 2003). Among these studies recast has proved to be the most frequent type of error correction in classroom [30] [32]; however, self-correction or elicitation might make students notice the gap more than recast. To illuminate the difference, two methods of recast and self correction were applied to two groups of students to see if there would be any difference between two methods on the students' writing accuracy regarding the use of past tense. Man Whitney, along with Wilcoxon, was run as the main statistical techniques to examine the effectiveness of methods of error correction on students' writing accuracy. The result of this study showed that there is no significant difference between the two methods of correction although students improve after treatment; however, self correction outperformed recast and recast did not improve students' accuracy in posttest. Keywords: self correction, recast, awareness, error correction. 1. Introduction Literature is replete with the studies in support or rejection of the facilitative role of error correction in second language teaching [11]. The inception of communicative language teaching, in its purest form especially, and its failure in producing accurate but fluent speaker [7] escalated this debate. The debate was especially highlighted in immersion programs, as Swain and Lapkin [33] (2007) state, what was missed in immersion classrooms was not the facilities as dictionaries, books, or computers but "the opportunities to interact with all these artifacts by verbalizing understanding that arise from such interactions" (p. 85). The debate still continues; the inclusion of instruction in either forms of preemptive or reactive is not a one-horse race. A series of modular theories in favor of natural development of second language are pervasive. These include Pienemann's [26] [27] processability theory, and Clahsen and Muysken's [3] strategy theory, as well as early communicative language teaching (CLT) perspectives in the late 1970s and the 1980s. According to early CLT proponents and authorities, errors are signs of language development and cannot be avoided or prevented. Therefore, in Han's [11] sense "in CLT paradigm, form focus instruction is deemed detrimental" (p. 1). On the other hand, the pure meaning focus instruction could not function very well not in terms of fluency but accuracy. Han [11] goes on saying that error correction-free instruction leads to "false fluency," "fossilization," inclination to a "low quality input," and learners' handicap in the case of overgeneralization failure. Through "skill-acquisition perspective" analogy, she emphasizes BRAIN. Broad Research in Artificial Intelligence and Neuroscience Volume 2, Issue 1, January 2011, ISSN 2067-3957 (online), ISSN 2068 - 0473 (print) 18 the important role of error correction as an extrinsic feedback which is more effective than intrinsic feedback in early stages of language acquisition. She further states: The classroom offers much space for extrinsic feedback on learner output. In the naturalistic setting, on the other hand, the learner has to rely almost exclusively on intrinsic feedback, i.e., feedback that springs from the situation itself …and from the learner himself. (p. 8) In SLA literature, extrinsic feedback or error correction, as was previously stated, is claimed to be more effective than intrinsic one or positive input [4]. Studies on different types of correction like recast, repetition, confirmations, reformulations, compression checks, confirmation checks, and request for clarification abound. One may refer to Ellis [5], Lyster [19], Loewen and Philp [17], Mackey [22], Mackey and Silver [23], Robinson [28], and Sheen [31, 32] for a comprehensive review on error correction, types, differences, and effectiveness. Among different kinds of correction, recast seems to be the most frequently used one [5] although it may lead to the least uptake [21]. Recast was originally defined in first language acquisition; Bohannon et al. (1996, as cited in Sheen, [32]) define recast as a kind of correction "that expands, deletes, permutes, or otherwise changes the platform while maintaining overlap in meaning" (p. 434). Ellis [5] states that "recast involves rephrasing an utterance" by changing one of its components while meaning is kept unchanged (p. 168). In this sense, recast is the "reformulation" of learners' non-target-like utterances. Mackey [22] defines recast as a target-like model signaled to students' erroneous sentences. This way, recast is considered as positive evidence (information about which forms are grammatical in the target language) for learners than negative one. The following example shows how recast works. Student: She is a good cooker Teacher: a good cook. Is She? Recast Student: Yes she is a good cook Modified output Being similar in its nature, recast has not appeared in a single form. Gass [9] classifies recast into "partial recast," "full recast," "recast in response to single error," and "recast in response to multiple errors". Partial recast is the reformulation of the erroneous segment of an utterance [32], or in Lyster's [20] terms, it is a recast through reduction or "a reduction of the learner’s utterance with added stress for emphasis" (p. 271). Elsewhere, Lyster [20] made a distinction between "isolated" and "incorporated" recast. While the former refers to recast without any change in meaning (through addition or deletion), the latter occurs when corrected reformulation added to the meaning of utterance. He also referred to isolated interrogative and incorporated interrogative recasts. Also, Mackey and Philp [23] differentiate "intensive" recast from "complex and simple" recasts. While intensive recast focuses on a particular linguistic item during communication, complex recast focuses on multiple linguistic items in one discoursal move. Moreover, Sheen [32] distinguishes conversational from didactic recast; the former is concerned with meaning, the latter with form. Another way of correction is self correction. Self correction is an indirect feedback where the teacher provides students with the choices so that they themselves discern the correct form [2]. Self correction or "self repair" [20], in either form of self or peer, happens when the teacher makes the erroneous point salient through repetition (in speaking), underlining the erroneous sentence (in writing) or writing both the correct and erroneous form on the board so that students can choose from. What is unique in self correction is the learner's conscious attention to his or her erroneous sentence. It seems that self correction is more effective than recast since it makes the learner notice the gap (Schmidt, 2001). To Lyster and Ranta [21] and Mckay [2006], elicitation is an explicit correction where the teacher provides the students with the wrong form, along with the correct form. Sometimes, two forms of correct and the original sentence are provided, and students should find the correct form. P. Maftoon, M. Ahmadi, P. Daftarifard - The Effect of Recast vs. Self Correction on Writing Accuracy 19 Self correction has been found to be effective in several studies. For example, Kubota [14] worked on lower intermediate university students learning Japanese as a foreign language. She found out that the number of errors of different categories in students' writing diminished when they use self correction through self-help resources. Elsewhere, Makino (1993, as cited in Lee and Ridley [15]) states student correction in either form of self correction and peer correction is more effective than teacher correction in the sense that "it allows learners to be the ’architects’ of their own learning"(p. 26). Self correction and recast are different from each other in terms of the types of evidence each provide students with. Error correction can provide the learners with both negative and positive evidence; positive evidence is usually referred to the natural input the learners are exposed to. In second language learning, there is no consensus on whether negative evidence is more effective or positive evidence. Negative evidence is usually referred to as correction and reactive instruction, whereas positive evidence entails both natural and reactive instructional setting. There is no doubt that self correction is self-negative directed evidence. However, to some of the researchers, recast can function both as positive and negative evidence. It functions as positive evidence when its saliency is internally created, and learners unconsciously uptake the correct form [11]. It can function as negative evidence when learners' reflect their own erroneous utterance [18]. Such explicit and implicit differentiation is not always regarded as dichotomous; to Loewen and Philp [17], such differentiation is a matter of degree. To Oliver and Mackey [25], implicitness and explicitness of recast is more context bound depending on how much salient they become to learners. Elsewhere, Sheen [32] believes that recast saliency is a psycholinguistic phenomenon: Linguistic phenomenon is determined by the linguistic characteristics of the recasts, whereas saliency is a psycho-linguistic construct. Recasts that are linguistically explicit are likely to be salient but explicitness cannot guarantee saliency as individual learner factors will influence the extent to which learners find particular recasts salient. In this respect, an investigation of learners’ perceptions about different characteristics of recasts is needed. (p. 386) With whatever role they might play either as positive or negative evidence or either as an explicit focus on form or implicit focus on form, error corrections vary in their degree of effectiveness. It is believed that the results on the effectiveness of feedback type change across different contexts and forms [24]. Although Lyster's study of French immersion classroom learners did not show the effectiveness of recast in children, Mackey and Silver [24] found out that the adult ESL students' linguistic output developed through recast even when the students did not repeat recasts or modify their output as a result of recasts. 2. Awareness and learning There is a paradigm war as to the nature, role, function, time, and importance of learners’ attention to input either internally or externally [35]. The concept has been discussed in different school of thought; while some delineate it within cognitive psychology [29], others approach it from more linguistic perspective [34]. Schmidt [29] attributed the possibility of SLA to focal attention and noticing. He reconceptualizes the concept of attention as a limited capacity. To him attention is a must in language learning. Some other scholars believe that attention can be both internally and externally driven [30]. Sharwood Smith [30] refers to input enhancement which is a means of drawing learners’ attention to the specific feature of input. Accordingly, internally driven attention or internal input saliency is more important than external as a means of highlighting certain language areas for learners with the goal of drawing their attention to those areas. Elsewhere, some [9] refers to attention as one of the factors that affect input processing so believes in attention as externally driven phenomena. Gass [9] criticizes Krashen’s simple model of conversion of input to intake subconsciously. Instead, she attributes language learning to attention in several stages. BRAIN. Broad Research in Artificial Intelligence and Neuroscience Volume 2, Issue 1, January 2011, ISSN 2067-3957 (online), ISSN 2068 - 0473 (print) 20 3. The purpose of this study Research into the effectiveness of error correction in language learning sometimes follows pretest posttest grammar test, or the number of corrected forms in the learners' uptake. Truscott [37] questioned the efficacy of such studies because "studies that look at the actual speech/writing of learners who have undergone correction have consistently failed to find any benefits" (p. 123). Elsewhere, Lyster and Ranta [21] cast doubt on the appropriateness of uptake as an indicator of feedback efficacy. Scholars have usually supported the use of oral or written tasks for studying the effectiveness of feedback on learners' uptake. Adams [1] believes that writing and speaking can more accurately measure feedback efficacy than pre and post grammatical test. For the same token and because not much research has been done using writing accuracy as a criterion of measuring feedback efficacy, this study aims at comparing the effectiveness of recast and self correction in classroom conversation through measuring the accuracy of the writing task. To do so, this study addresses the following questions: Q1. Is there any difference between the method of error correction and the correct use of past tense of the Iranian students at the elementary level? Q2. Does error correction in speaking affect the correct use of past tense in writing of the Iranian students at the elementary level? 4. Method 4.1. Participants Participants of this study were 16 students ranging from 19 to 24 years old. They enrolled at Marefat institute to study General English. They were customarily placed in the elementary level at which they studied New Interchange book. They were randomly grouped into two classes to receive two different corrective feedbacks as the treatment. Group A (8 students) received self correction as the correction strategy in the way that the teacher provided the correct form for the erroneous sentence the students produced, and the students needed to decide which one was grammatically correct. Group B (8 students) received recast with single move reformulation. Correction of either types of self correction or recast can vary in terms of different constituents or "moves" it takes. Correction, as Sheen [32] delineates, consists of a trigger utterance (elicit or error), the teacher's corrective feedback (sometimes recast, sometimes other corrective feedbacks), and either uptake (learner's immediate response to feedback) or no response. In the latter, when learner does not uptake the correct form, the topic continues by either learners or the teacher. These moves can happen in a multi-move or single-move. Multi-move correction contains more than one teacher- feedback-move, whereas simple move correction may happen in the form of declarative a recast or interrogative. As the purpose of this study was to investigate the impact of two types of feedback on the accuracy of writing in terms of using past tense, students' writing sample was selected as the instrument as for both pretest and posttest. In the first session, the first prompt that is, "Last holiday" was given to the students to write. The students were free to write about the topic in the sense that they could write at will. The posttest writing prompt was "What did you do last week?" and it was given in the last session. No meta-linguistic tests or grammatical tests were used in this study because in the previous studies [37] the use of such tests was criticized, for they are the artificial way of checking competence. Moreover, because Marefat institute frequently measures its students' competency in grammar, the researchers relied on the institution's assumption about the level and homogeneity of the students. Also, as was mentioned earlier, writing and spontaneous speech are considered to be more natural in comparing with discrete grammar tests [1]. The participants in this study were free in writing the compositions at any length. To neutralize the length difference the number of errors were divided by the number of verbs used. In P. Maftoon, M. Ahmadi, P. Daftarifard - The Effect of Recast vs. Self Correction on Writing Accuracy 21 doing so, the mean of erroneously used verbs is obtained for each participant. Table 1 summarizes the students' writing features as number of words, verbs (past tense) both in pretest and posttest. Table 1. Data Features Summary Name Method Verbpre Wrongpre Verbpost Wrongpost Meanpre meanpost 1 1.00 recast 14.00 1.00 12.00 0.00 0.07 0.00 2 12.00 recast 9.00 1.00 9.00 4.00 0.11 0.44 3 16.00 recast 9.00 4.00 19.00 3.00 0.44 0.16 4 15.00 recast 12.00 4.00 12.00 1.00 0.33 0.08 5 10.00 recast 10.00 2.00 9.00 1.00 0.20 0.11 6 13.00 recast 7.00 0.00 11.00 1.00 0.00 0.09 7 17.00 recast 10.00 1.00 12.00 0.00 0.10 0.00 8 14.00 recast 8.00 1.00 8.00 0.00 0.13 0.00 9 8.00 Self-correction 20.00 4.00 6.00 3.00 0.20 0.50 10 11.00 Self-correction 10.00 1.00 12.00 1.00 0.10 0.08 11 5.00 Self-correction 11.00 1.00 6.00 1.00 0.09 0.17 12 2.00 Self-correction 13.00 2.00 17.00 0.00 0.15 0.00 13 7.00 Self-correction 14.00 4.00 6.00 0.00 0.29 0.00 14 3.00 Self-correction 25.00 3.00 8.00 1.00 0.12 0.13 15 9.00 Self-correction 18.00 1.00 10.00 0.00 0.06 0.00 16 4.00 Self-correction 10.00 2.00 6.00 0.00 0.20 0.00 4. 2. Procedure Both classes covered five lessons from New Interchange book (nine to thirteen). Each unit includes two conversations, grammar focus lessons, and listening parts, as well as one writing, and reading task. There is also a conversation task separating writing and reading tasks from each other which can be considered as both post writing, and pre reading activity. One unit of the book was covered every three sessions. For the purpose of this study, students were assigned to read three short stories on their own. The short stories were all selected to meet their level of proficiency. The students were supposed to retell the story. The teacher corrected the erroneous use of tenses in any forms of simple present, simple past and present perfect. However, because students at this elementary level are supposed not to have any error in the form of past tense, the focus of the pretest and posttest was on the correct use of past tense. 4. 3. Instructional Treatments Each session lasted for 90 minutes twice a week, and for seven running weeks taught by the same teacher, who was an MA graduate of TEFL. Both groups were matched for the amount of instructional time. The first group was provided with recast in the form of immediate feedback - feedback without any move, i.e., immediately after the occurrence of the erroneous statements. The second group was provided with the self correction feedback following the occurrence of the erroneous statements. The self correction feedback was vocalic; the teacher repeated the student's erroneous sentence, along with the correct form, and asked the student to choose the correct form. Both classes usually started with a warm up in the form of questions and greetings, and then the following procedure was approached: conversation tasks, grammar focuses, pronunciation part, listening, post listening (some questions about students' opinion on the topic discussed in listening), and snap shots on the equivalent topic, conversation, grammar, listening, writing, speaking, and then reading. As was previously mentioned, one unit was covered every three sessions. Both groups were provided with feedback on the participants' erroneous sentences as far as past tense was concerned. For example, indirect speech, which has not been covered for this level, was not corrected. Besides retelling the story, some participants were called to answer questions in class. Whenever the participants produced a grammatically incorrect sentence, feedback was provided on the spot (in one group in the form of recast and in another group in the form of self correction). BRAIN. Broad Research in Artificial Intelligence and Neuroscience Volume 2, Issue 1, January 2011, ISSN 2067-3957 (online), ISSN 2068 - 0473 (print) 22 5. Result To answer the questions of this study, first the test of homogeneity, Kolmogorov-Smirnov, was run for both pre and post tests to see if the data met the criteria needed for parametric test. If p value of the test is significant (p<0.05), the distribution is not normal. As is shown in Table 2, not all of the values conform to the parametric norms; self correction in pretest and posttest showed significant p value (p<0.05). Therefore, the nonparametric forms of data analyses (Mann Whitney Test and Wilcoxon) were run for the analysis of the data. Table 2. Tests of Normality of Kolmogorov-Smirnov METHOD Statistic df Sig. MEANPRE Recast .253 8 .140 Self correction .120 9 .200* MEANPOST Recast .250 8 .151 Self correction .279 9 .042** * This is a lower bound of the true significance. ** This is an upper bound of the true significance. a Lilliefors Significance Correction The second assumption to be tested was that the two groups, in this study, should not be significantly different in pretest. This is checked through Mann Whitney Test. As is shown in Table 3, there is no significant difference between these two groups (z = 0.053, p = 0.958). The probability index shows that there are only 0.04 chances for groups to be different from each other. Table 3. Pretest Man-Whitney Test b a Not corrected for ties. b Grouping Variable: METHOD To answer question one that is, "is there any difference between the effects of method of error correction on the correct use of past tense of students at the elementary level?" the nonparametric version of t test (Mann Whitney Test) was run again. . As is shown in Table 4, there is no significant difference between the two groups of recast and self correction (z = 0.165, p = 0.869). Table 4. Posttest Man-Whitney Test b METHOD N Mean Rank Sum of Ranks MEANPOST Recast 8 8.69 69.50 Self correction 8 8.31 66.50 Total 16 Mann-Whitney U 30.500 Z -.165 Asymp. Sig. (2-tailed) .869 Exact Sig. [2*(1-tailed Sig.)] .878 a a Not corrected for ties. b Grouping Variable: METHOD METHOD N Mean Rank Sum of Ranks MEANPRE Recast 8 8.56 68.50 Self correction 8 8.44 67.50 Total 16 Mann-Whitney U 31.500 Z -.053 Asymp. Sig (2-taild) .958 Exact Sig. [2*(1-tailed Sig.)] .959a P. Maftoon, M. Ahmadi, P. Daftarifard - The Effect of Recast vs. Self Correction on Writing Accuracy 23 Comparing Tables 3 and 4, we notice some differences between posttest and pretest although they are not significant. As is shown in Table 3, mean rank for recast in pretest is 8.56, but this value increases in posttest to 8.69. However, mean rank for self correction in pretest is 8.44 and this value decreases in posttest, 8.31. Although these changes are not significant, the results show that self correction outperform recast, and recast seems to have no effect at all because the number of errors increases. This finding is supported by the results reported in the literature of error correction [19], [17], [22], and [24]. It seems that recast cannot have a delay effect on the learners' uptake. This might be due to the fact that sometimes recast is not taken as error correction but repetition. Many problems associated with recast might be the result of the mismatch between teacher's intention and students' interpretation and perception on what is actually corrected. This mismatch between teachers' intention and students' interpretation, of course, might not be limited to one type of error correction as Han [11] states that "what is made salient by the teacher may not be perceived as salient by the learner" and "hence will have no effect on development" (p. 11). In case of self correction, however, the students are asked to find the erroneous sentences and this active engagement of the students might result in a better performance, though insignificantly, in the posttest. To answer the second question of this study, "does error correction in speaking affect the correct use of past tense in writing of students at the elementary level?" the nonparametric Wilcoxon Test was run for each group separately. Table 5. Recast Wilcoxon Signed Ranks Test N Mean Rank Sum of Ranks MEANPOST - MEANPRE Negative Ranks 6a 4.17 25.00 Positive Ranks 2b 5.50 11.00 Ties 0c Total 8 Ze -.980d Asymp. Sig. (2- tailed) .327 a MEANPOST < MEANPRE b MEANPOST > MEANPRE c MEANPRE = MEANPOST d Based on positive ranks. e Wilcoxon Signed Ranks Test As is shown in Table 5, the mean ranks reported for the recast group in the pretest and posttest were 4.17 and 5.50 respectively at the probability of 0.327. Therefore, the change or improvement through recast correction does not occur although this change is not statistically meaningful (z = 0.980, NS). However, the mean rank for self correction group in the pretest and posttest were 4.60 and 4.33 respectively at the probability of 0.484 (z = 0.700, NS). The result indicates that through self correction students have improved although this change is not statistically meaningful. Table 6. Self Correction Wilcoxon Signed Ranks Test N Mean Rank Sum of Ranks POSTMEAN - PREMEAN Negative Ranks 5a 4.60 23.00 Positive Ranks 3b 4.33 13.00 Ties 0c Total 8 Zd -.700e Asymp. Sig. (2- tailed) .484 a POSTMEAN < PREMEAN BRAIN. Broad Research in Artificial Intelligence and Neuroscience Volume 2, Issue 1, January 2011, ISSN 2067-3957 (online), ISSN 2068 - 0473 (print) 24 b POSTMEAN > PREMEAN c PREMEAN = POSTMEAN d Based on positive ranks. e Wilcoxon Signed Ranks Test Although the results are not statistically meaningful, they should be taken into consideration for two reasons. First, the non-significant result might be because of the fact that the number of participants in this study, due to the nature of research, i.e., classroom research, is not statistically satisfactory; there were eight participants in each group. Second, this population has not formed normal distribution (See Table 2). However, as Hatch and Lazaraton [12] indicate, the practical and interpretable significance worth considering even if the result is not statistically significant, especially when the result is fully supported by the literature. the concluding section would further elaborate on this issue. In second language acquisition (SLA), learners are different in terms of rate, route, and the final attainment. As Han [11] states, "the time a learner takes in arriving at a correct hypothesis can vary from a few seconds to a lifetime" (p. 9). This could be the result of many factors such as age, amount of exposure to language, gender, or probably the days of participation in class. For this reason, the researchers decided to qualitatively explore probable individual differences through the available historical background. Table 6 summarizes the available information about participants. As is shown in Table 6, some students in the self correction group pass this course for the second time (students 5, and 7). Two students of the same group passed the previous term (Inter1) with the minimum score of 73 and 74 (students 8 and 11). Student number eight, in the self correction group, seems not to have functioned very well. This might be due to the fact that he has passed the previous course by minimum score and therefore, he might not be ready for learning past tense yet. The interesting point is that, the recast group seems to be more qualified than the self correction group; they passed their previous term with higher scores. However, the recast group has not outperformed self correction group. Table 7. Exploring Individual Differences: Qualitative Analysis Recast group Posttest Pretest Name present/absent Absent days previous score Previous level .00 .07 1 6/10 (5, 8, 9, 10) 84 inter2 .44 .11 12 10 95 inter2 .16 .44 16 9/10 (7) 92 inter2 .08 .33 15 9/10 (9) 81 inter2 .11 .20 10 9/10 (2) 81 inter2 .09 .00 13 9/10 (2) 92 inter2 .00 .10 17 9/10 (7) 92 inter2 .00 .13 14 9/10 (2) Self correction group .50 .20 8 10 73 inter2 .08 .10 11 9 (8) 74 inter2 .17 .09 5 7/10 (1, 8, 9) 0 .00 .15 2 7/10 (3, 6, 8) 89 inter2 .00 .29 7 10 0 .13 .12 3 7/10 (1, 2, 3) 91 inter2 .00 .06 9 8/10 (1, 2) 93 inter2 .00 .20 4 9/10 (7) 97 inter2 6. Conclusion There is an unending debate between two extreme ways of teaching in second language setting: intrusive or instructional and non-intrusive or natural [16]. Instructional or focus on form and focus on meaning happen when the learners are exposed to negative evidence in either forms of P. Maftoon, M. Ahmadi, P. Daftarifard - The Effect of Recast vs. Self Correction on Writing Accuracy 25 preemptive and reactive [8]. On the one hand, non-intrusive language teaching or focus on meaning [6] mainly emphasizes the importance of natural exposure to language input through meaningful, natural or pedagogical interaction or task either within (early CLT) or outside of classroom [11]. Pienemann's [26] [27] processability theory, Krashen's [13] input hypothesis, and Clahsen and Muysken's [3] strategy theory are endorsements for such a claim. In an extreme reaction to the failure of audio-lingual method and as a result of many modular theories and research inquiries, the early CLTs totally prevented and condemned any use of error correction. On the other hand, the failure of purely meaning-based language teaching in developing accurate but fluent learners has led scholars to question the effectiveness of teaching activities without any focus on form. A series of articles that are published in [8] are significant pieces of evidence on the felt importance of instruction either in the form of preemptive or reactive language teaching. There are several studies supporting the facilitative role of correction of any type [5], [9], [19], [17], [22], [24], [28], and [31. As was mentioned earlier among different kinds of correction, recast seems to be the most frequently used one (Ellis, 2003; Sheen, 2004, 2006), however, as Lyster and Ranta's [21] study shows, it led to the least uptake. On the other hand, it is mostly mentioned that self correction or elicitation may be more effective than recast because in the former case, students notice the gap [29]. The result of this study should be considered cautiously. The sample size was small, so the power of test is definitely low. Bearing this in mind, this study cannot take sides of either of extremes. The study showed that there was no significant difference between both methods of correction and pretest and posttest result. However, the interesting point in this study was the trend difference data showed. As Hatch and Lazaraton [12] mention the significant difference is a "convention" set by the statistician "not a hard-and-fast rule" (p. 233). In the present study, the differences between pretest and posttest did not show a significant difference; however, the mean rank of wrong used verbs in recast group in the posttest was more than that of self correction group, considering that less mean rank indicated a better result. For sure, a classroom with no error correction is far from reality; all teachers would like to create a change in their students. When there is less possibility for exposure to positive evidence, especially in case of foreign language context, the best way to prevent fossilization is utilizing different methods of error correction. However, as Guenette [10] states this is a "decisive factor in the attainment of language fluency and accuracy" (p. 41). Truscott [37] questions the efficacy of correction. He states that "studies that look at the actual speech/writing of learners who have undergone correction have consistently failed to find any benefits" (p. 123). This study has not shown any statistically significant result; however, two methods of error correction come to a mixed result. While self correction bettered the result of the posttest, recast worsened it. This can be interpreted from several perspectives each of which is explained below. First, although error correction has won the supports of many scholars, Pienemann's processability theory states that no change or learning would occur if it is beyond the learners' level of interlanguage. The participants of this study were at the elementary level. They have recently learned the use and concept of past tense through New Interchange book, which is based on the communicative language teaching syllabus. Therefore, it is not surprising that students have not fully acquired the concept and use of past tense yet. Therefore, this study seems to support Pienemann's theories, i.e., teachability, learnability, and processability. Accordingly, no learning would happen beyond the developmental level of interlanguage, no matter how hard we tried to create it in our learners. Scrutinizing the data, it was found out that the self correction group included those students who have failed the previous terms, and this was the second time they were taking this course. Maybe, this can explain the improvement of these students over the recast group. More research enquires are required to explore this possibility to see if the better performance is the result of retaking of the same course, or recast as a method of error correction is easier and less effective method of error correction than self correction. BRAIN. Broad Research in Artificial Intelligence and Neuroscience Volume 2, Issue 1, January 2011, ISSN 2067-3957 (online), ISSN 2068 - 0473 (print) 26 Second, self correction may result in "discovery learning" which is supported from different theoretical positions as "experiential learning," "deep processing," "self-investment," "noticing," and "psychological readiness" [36]. Experiential learning refers to the situation where learners first experience a feature before studying it. Discovery learning implies that learning is more internally driven than externally driven. In this case, the whole person is ready to learn and acquire the feature. Deep processing is semantic based that is, learners attend to the meaning and significance of the intake. Self-investment also supports the importance of discovery learning. Tomlinson [36] asserts that "learners will only learn if they need and want to learn and are willing to invest time and energy in the process (p. 179). Maybe, noticing through self correction gives "salience" to the feature in the future input and this helps learners be more "psychologically ready" [36] to uptake the input in the future. Guenette [10] states "experts have demonstrated that under certain conditions, with certain student populations, and in some contexts, error correction is effective" (p. 41). The data of this study show that students did not uptake the erroneous points in the same way. Some students improved in the posttest, but some did not. This should be explained through more information obtained about each participant. We have not had much information on students' learning style, strategy, and specific demographic background. Therefore, another research may study the effectiveness of these two methods of error correction more meticulously by probing the relationships between different learning style, motivation, or learning strategy and the occurrence of uptake. Moreover, as this research lacks control group, we are not sure that the methods have no effect at all. Another research with the same design, i.e., searching for accuracy of writing by giving feedback through speaking out the grammatical points, might consider this problem and include the control group as well. This study focused on writing composition. Another research might focus on more natural mode of writing as writing letters or emails, or writing in forums because they are more natural and spontaneous. Also, this study explored the effect of giving feedback in speech on writing accuracy. Another research can study the effect of giving feedback in writing on speech accuracy. And last but not least is that this study did not consider time very much. Another research can focus on time as an important factor in remembering and up-taking the correct form of erroneous sentences. This can include gathering the writing samples of students who receive feedback and those who listen to feedback. This can help us understand the importance of both time and attention over attendees and audiences. 7. References [1] Adams, R. (2003). L2 output, reformulation and noticing: Implications for IL development. Language Teaching Research, 7, 347–376. [2] Bitchener, J., Young, S., and Cameron, D. (2005). The effect of different types of corrective feedback on ESL student writing. Journal of Second Language Writing, 14, 191-205. [3] Clahsen, H., and Muysken, P. (1989). The UG paradox in L2 acquisition. Second Language Research, 5, 1-29. [4] Elder, C., Erlam, R., and Philp, J. (2007). Explicit language knowledge and focus on form: Options and obstacles for TESOL teacher trainees. In S. Fotos and H. Nassaji (Eds.), Form-focused instruction and teacher education: Studies in honor of Rod Ellis (pp. 225-240). Oxford: OUP. [5] Ellis, R. (2003). Task-based language learning and teaching. Oxford: OUP. [6] Ellis, R. (2008). Explicit form-focused instruction and second language acquisition. In B. Spolsky and F. M. Hult (Eds.), the handbook of educational linguistics (pp. 43-455). Oxford: Blackwell Publishing. [7] Ellis, N. C. (2007). The weak interface, consciousness, and form-focused instruction: Mind the doors. In S. Fotos and H. Nassaji (Eds.), Form-focused instruction and teacher education: Studies in honor of Rod Ellis (pp. 17-34). Oxford: OUP. P. Maftoon, M. Ahmadi, P. Daftarifard - The Effect of Recast vs. Self Correction on Writing Accuracy 27 [8] Fotos, S., and Nassaji, H. (2007). Form-focused instruction and teacher education: Studies in honor of Rod Ellis. Oxford: OUP. [9] Gass, S. M. (2003). Input and Interaction. In C. Doughty, and M. Long (Eds.), The handbook of second language acquisition (pp. 224-255). New York: Blackwell. [10] Guenette, D. (2007). Is feedback pedagogically correct? Research design issues of feedback on writing. Journal of Second Language Writing, 16, 40-53. [11] Han, Zh. H. (2002). Rethinking the role of corrective feedback in communicative language teaching. RELC Journal, 33, 1-34. [12] Hatch, E. M., and Lazaraton, A. (1991). The research manual: Design and statistics for applied linguistics. New York: Newbury House. [13] Krashen, S. D. (1988). Second language acquisition and second language learning. New York: Prentice-Hall. [14] Kubota, M. (2001). Error correction strategies used by learners of Japanese when revising a writing task. System, 29, 467–80. [15] Lee, W., and Ridley, A. (1999). What implications does English globalization have for treatment of students' spoken errors? RELC Journal, 30, 18-37. [16] Lightbown, P. M., and Spada, N. (2006). How languages are learned (3rd ed.). Oxford: OUP. [17] Loewen, S., and Philp, J. (2006). Recasts in the adult English L2 classroom: Characteristics, explicitness, and effectiveness. The Modern Language Journal, 90, 536-556. [18] Long, M., Inagaki, S., and Ortega, L. (1998). The role of implicit negative feedback in SLA: Models and recasts in Japanese and Spanish. The Modern Language Journal, 82, 357-71. [19] Lyster, R. (1998a). Recasts, repetition, and ambiguity in L2 classroom discourse. Studies in Second Language Acquisition, 20, 51-81. [20] Lyster, R. (1998b). Negotiation of form, recasts, and explicit correction in relation to error types and learner repair in immersion classrooms. Language Learning, 48, 183-218. [21] Lyster, R., and Ranta, L. (1997). Corrective feedback and learner uptake: Negotiation of form in communicative classrooms. Studies in Second Language Acquisition, 19 (2), 37-66. [22] Mackey, A. (2006). Feedback, noticing and instructed second language learning. Applied Linguistics, 27(3), 405-430. [23] Mackey, A., and Philp, J. (1998). Conversational interaction and second language development: Recasts, responses, and red herrings? Modern Language Journal, 82, 338-56. [24] Mackey, A., and Silver, R. E. (2005). Interactional tasks and English L2 learning by immigrant children in Singapore. System, 33, 239-260. [25] Oliver, R., and Mackey, A. (2003). Interactional context and feedback in child ESL classrooms. Modern Language Journal, 87, 519-533. [26] Pienemann, M. (1998). Language processing and second language development: Processability theory. Philadelphia: John Benjamin B. V. [27] Pienemann, M. (2003). Language processing capacity. In C. Doughty and M. H. Long (Eds.), The handbook of second language acquisition (pp. 679-714). New York: Blackwell. [28] Robinson, P. (2001). Conditions in second language acquisition: Individual differences, cognitive abilities, aptitude complexes and learning. Second Language Research, 17 (4), 368-392. [29] Schmidt, R. (2001). Attention. In P. Robinson (Ed.), Cognition and second language instruction (2nd ed.) (pp. 3-32). Cambridge: CUP. [30] Sharwood Smith, M. (1994). Second language learning: Theoretical foundations. London: Longman. [31] Sheen, Y. (2004). Corrective feedback and learner uptake in communicative classrooms across instructional settings. Language Teaching Research, 8 (3), 263-300. [32] Sheen, Y. (2006). Exploring the relationship between characteristics of recasts and learner uptake. Language Teaching Research, 10 (4), 361-392. [33] Swain, M., and Lapkin, S. (2007). The distributed nature of second language learning: Neil's perspective. In S. Fotos and H. Nassaji (Eds.), Form-focused instruction and teacher education: Studies in honor of Rod Ellis (pp. 73-86). Oxford: OUP. BRAIN. Broad Research in Artificial Intelligence and Neuroscience Volume 2, Issue 1, January 2011, ISSN 2067-3957 (online), ISSN 2068 - 0473 (print) 28 [34] Talmy, L. (2008). Aspects of attention in language. In P. Robinson & N. C. Ellis (Eds.), Handbook of cognitive linguistics and second language acquisition (pp. 27-38). NY: Routledge. [35] Tomlin, R. S., & Villa, V. (1994). Attention in cognitive science and second language acquisition. SSLA, 16, 183-203. [36] Tomlinson, B. (2007). Using form-focused discovery approaches. In S. Fotos and H. Nassaji (Eds.), Form-focused instruction and teacher education: Studies in honor of Rod Ellis (pp. 177- 192). Oxford: OUP. [37] Truscott, J. (1998). Noticing in Second Language acquisition: A critical review. Second Language Research, 14 (2), 103-135.