607 Studies in Second Language Learning and Teaching Department of English Studies, Faculty of Pedagogy and Fine Arts, Adam Mickiewicz University, Kalisz SSLLT 9 (4). 2019. 607-631 http://dx.doi.org/10.14746/ssllt.2019.9.4.3 http://pressto.amu.edu.pl/index.php/ssllt Patterns of uptake and repair following recasts and prompts in an EFL context: Does feedback explicitness play a role? Reza Shirani University of Victoria, Canada https://orcid.org/0000-0003-4746-7719 rezashirani70@gmail.com Abstract This study sought to examine the effectiveness of two categories of feedback, namely recasts and prompts. Also, the study focused on the relationship be- tween subsets of each feedback type and the extent to which they led to learner uptake and repair in an EFL context. Data were collected through non-participant observations of three intact upper-intermediate EFL classes where 36 hours of interactions among 59 students and three teachers were audiotaped, tran- scribed, and analyzed in terms of pre-specified coding systems that addressed four different subtypes of prompts – clarification requests, repetitions, elicita- tions, and metalinguistic clues – and two recast subtypes – explicit and implicit recasts. Data analysis showed that among prompts, clarification requests led to the highest percentage of uptake whereas elicitations were associated with the highest repair percentage. As for recasts, more explicit ones led to higher per- centages of uptake and repair. The results of the study may contribute to a more in-depth understanding of the patterns of uptake and repair in an EFL context. The study confirms the role of feedback explicitness in such a context. Keywords: corrective feedback; prompts; recasts; repair; uptake Reza Shirani 608 1. Introduction Recently, classroom interaction has been a major research topic due to its im- portant role in language learning. Interaction provides learners with opportuni- ties to receive input and generate output as they attend to linguistic form and negotiate for meaning. Whereas some underscored the role of comprehensible input as the only necessary source of language learning (e.g., Krashen, 1982; Schwartz, 1993; Truscott, 1996), there was a growing consensus among other researchers that input is not sufficient by itself if learners are to achieve favora- ble mastery of an L2 (e.g., Lightbown & Spada, 1990; Long, 1996; Mackey, 2012; Norris & Ortega, 2000). Most of these researchers have shown that, besides re- ceiving input, learners need to produce output and receive feedback when nec- essary. Defined as feedback on erroneous language production, corrective feed- back (CF) is a focus-on-form (FonF) technique that draws learners’ attention to linguistic forms and features (Nassaji & Fotos, 2011). CF falls into four broad categories of prompts, recasts, explicit feedback, and direct correction. Nicholas, Lightbown, and Spada (2001) define recasts as “utterances that repeat a learner’s incorrect utterance, making only the changes necessary to produce a correct utterance, without changing the meaning” (p. 733). Prompts, on the other hand, are elicitation strategies which “withhold cor- rect forms (and other signs of approval) and instead offer learners an opportunity to self-repair by generating their own modified response” (Lyster, 2004, p. 405). To assess the effectiveness of CF, it has been suggested that researchers consider learner uptake, “a student’s utterance that immediately follows the teachers’ feedback and that constitutes a reaction in some way to the teachers’ intention to draw attention to some aspects of the student’s initial utterance” (Lyster & Ranta 1997, p. 49). Uptake then may be either successful or unsuccessful. To date, many studies (e.g., Ellis, Basturkmen, & Loewen 2001; Fu & Nas- saji, 2016; Llinares & Lyster 2014; Lyster & Ranta, 1997; Panova & Lyster, 2002) have explored the relationship between different types of CF and learner up- take. However, further research is needed to examine the relationship between implicit and explicit feedback types and learner uptake and repair in a foreign language (FL) context, where the provision of appropriate types of CF is of vital importance. Therefore, this study aimed to investigate the patterns of uptake and repair following implicit and explicit recasts and prompts in three FL classes. 2. Recasts, prompts, and feedback explicitness To date, most of the previous studies (e.g., Llinares & Lyster, 2014; Lyster & Mori, 2006; Lyster & Ranta, 1997; Panova & Lyster, 2002) have considered recasts a Patterns of uptake and repair following recasts and prompts in an EFL context: Does feedback. . . 609 single implicit feedback type. Some other studies (Erlam & Loewen, 2010; Nas- saji, 2007; Sheen, 2006), however, have demonstrated that recasts fall on a con- tinuum of explicitness, depending on their features. On the whole, the following features have been found to influence the explicitness of recasts: · Prosodic emphasis: If a recast is provided with added intonational em- phasis, it will be less implicit and therefore more salient to the learner (e.g., Nassaji, 2007). · Length: Shorter recasts are more explicit than longer ones because they fo- cus on the part of the utterance that contains the error (e.g., Sheen, 2006). · Intensity of focus: Recasts provided intensively on a particular linguistic form are generally more explicit than the ones given extensively on any ran- dom form (e.g., Erlam & Loewen, 2010). · Number of feedback moves: Recasts accompanied by a second feedback move become more salient to the learner (e.g., Erlam & Loewen, 2010; Sheen, 2006). Moreover, a controversial feature that may affect the explicitness of recasts is whether they are declarative or interrogative. Whereas some (e.g., Loewen & Philp, 2006; Sheen, 2006) have provided evidence that recasts with declarative intonation are more salient and thus more explicit, others have argued the op- posite (e.g., Nassaji & Fotos, 2011). All in all, previous research has shown that not all recasts are implicit. On the other hand, according to Lyster (2004), clarification requests, repe- titions, metalinguistic clues and elicitations are the four types of CF that are clas- sified as prompts, in that they push or prompt the learner to self-correct. Theo- retically, prompts are believed to be highly effective as they provide negative feed- back, draw learners’ attention to form, and provoke self-repair. Prompts can, for example, lead to noticing the hole (i.e., when learners realize that they are not able to generate the output they desire to [Swain, 1993]). Moreover, Swain (2005) posited that prompts can result in noticing the gap (i.e., when learners discern the differences between their interlanguage and the target language). Although it is generally suggested that both recasts and prompts contribute to L2 acquisi- tion, there is controversy over which type can contribute more. 3. Previous studies Many experimental and descriptive studies have investigated the efficacy of CF. However, because the main focus of the current study is on the relationship be- tween feedback and uptake, only the previous studies on uptake are reviewed here. A widely-cited observational study into CF by Lyster and Ranta (1997) inves- tigated the frequency of CF types and uptake in four elementary French immersion Reza Shirani 610 classes and reported that recasts were the most frequent type (55%) whereas elicitations (14%), clarification requests (11%), metalinguistic feedback (8%), and repetitions (5%) occurred less frequently. Also, it was found that prompts led to more uptake and repair vis-à-vis recasts. However, the repair percentage of recasts (18%) was calculated according to the overall number of recasts and not based on the total number of uptake moves succeeding those recasts. Clearly, not all recasts did lead to uptake in their observation. As noted by Oliver (1995), uptake opportunity has a decisive effect on the number of uptake moves. Uptake opportunity was not, however, considered in the above study. Panova and Lyster’s (2002) study also explored CF and uptake patterns in an L2 classroom setting and found similar results to those obtained by Lyster and Ranta (1997). Analyzing 10 hours of interaction, they observed that although recasts occurred more frequently than other kinds of feedback, they led to fewer uptake moves (40%) and repair (13%) than prompts. Ellis et al. (2001) conducted another study into feedback, uptake, and repair in an English as a second language (ESL) context and also found that recasts were the most dominant type of feedback, but in contrast to Lyster and Ranta (1997), Ellis et al. (2001) discovered that recasts could be highly facilitative as they led to a great deal of uptake (71.6%) and repair (76.3%). Unlike Lyster and Ranta, Ellis et al. computed the percentage of repair based on the total number of uptake moves following recasts, and not according to the overall number of recasts. Nassaji’s (2007) study also examined whether uptake and repair can be influenced by how a feedback type is provided. Forty-two ESL learners were ran- domly paired up with a teacher and engaged in task based interactions in which they received CF in different ways. Six types of recasts and five subtypes of elic- itations were identified and coded on the basis of the extent to which they were accompanied by certain signals that could make the feedback more salient. The analyses revealed that whenever a feedback move was provided more explicitly, learners were more likely to generate uptake with repair. Loewen and Philp (2006), too, reported that the degree to which recasts are accompanied by ex- plicit features can have a positive influence on uptake and repair. In another study, Lyster and Mori (2006) compared recasts and prompts in French immersion (FI) and Japanese immersion (JI) classes. First, they exam- ined the relationship between different types of feedback and learner uptake and repair in each context. Second, they compared the overall communicative orientation of the two contexts. As for the first comparison, in the FI context, prompts resulted in higher percentages of uptake (62%) and repair (53%) whereas in the JI context, recasts were found to be more effective (61% uptake; 68% repair). As for the second comparison, the JI context demonstrated some extent of form-focused orientation, which was not observed in the FI context. Patterns of uptake and repair following recasts and prompts in an EFL context: Does feedback. . . 611 More recently, Llinares and Lyster’s (2014) study of feedback and uptake across content and language integrated learning (CLIL), French immersion (FI), and Japanese immersion (JI) instructional settings revealed that teachers ap- plied prompts and recasts at a similar rate in all three instructional settings, with recasts being the most dominant type of feedback. Moreover, whereas recasts led to a higher percentage of repair (77%) in the CLIL setting, prompts resulted in more repair (53%) in the FI instructional setting. Fu and Nassaji’s (2016) investigation of a Chinese as a FL class also reported that recasts were the most prevalent type of feedback (56.5%). One intriguing infer- ence from the findings of this study is that although recasts failed to lead to a large number of uptake moves (49.6%), whenever they did result in uptake, they led to a high number of repair moves. However, similar to Lyster and Ranta (1997), compu- tations in this study were based on recasts-repair ratio, rather than uptake-repair ratio. As for prompts, clarification requests were found to lead to larger amounts of uptake (100%) and repair (66.7%) than other types of elicitation strategies. Research into CF in relation to uptake patterns has reported conflicting findings. Although most of the previous studies have, as discussed earlier, agreed on the higher frequency of recasts compared to prompts, they have re- vealed conflicting results with regard to which type may lead to more uptake and repair episodes, with some favoring prompts and some favoring recasts. Moreover, most of the previous studies of CF and uptake (cf. Sheen, 2006) have ignored how the explicitness and implicitness of prompt and recast subtypes may af- fect the patterns of uptake and repair in language classrooms. These studies (e.g., Llinares & Lyster, 2014; Lyster and Ranta, 1997; Panova and Lyster, 2002) have, for example, coded all of the recasts as one single type regardless of their explicitness or implicitness. Yet, as discussed earlier, recasts may vary in terms of explicitness under the influence of certain factors such as intonation, the number of moves and length. Finally, further research is needed to broaden the understanding of the nature of CF, its types, and patterns of uptake and repair in a FL context because, to date, the studies of CF in ESL contexts have comprised a larger proportion of research than those in FL classrooms. In FL contexts, where the classroom may be the only place for learners to learn, use, and practice a foreign language (Brown & Lee, 2015), it is important for teachers and practitioners to provide the most efficient types of CF to facilitate learning. 4. This study In light of the need for further research, this study examined discourse patterns of uptake and repair following different types of prompts and recasts during interactions in three FL classrooms. Also, a more specific aim of the study was to explore the Reza Shirani 612 effects of recasts that are more or less explicit. Hence, this study addressed the following questions: 1. What is the frequency of recasts and prompts and their subtypes during classroom interactions in a FL context? 2. To what extent do recasts, prompts, and their subtypes lead to learner uptake in a FL context? 3. To what extent do recasts, prompts, and their subtypes lead to learner repair in a FL context? 5. Method 5.1. Participants Participants were 59 students and three teachers from three intact upper-interme- diate classes (class A: 25, class B: 16, class C: 18) at a language school in Iran. All were present during the observations and audio recordings of the classes investi- gated in the study. The classes were true reflections of a FL context as all of the students shared the same linguistic background, Farsi. Also, none of them had ever stayed in an English-speaking country. The participants were placed in this level af- ter passing oral and written tests at the end of the previous semester. With the ex- ception of five students (three from class A; two from class C), who had just joined this language school after leaving another school, all other participants had been studying English for more than three years at the language school investigated in this study. Finally, it should be noted that all of the participants were either univer- sity freshmen or school seniors who were going to improve their English skills. In- formation about the students and teachers in each class is presented in Table 1. Table 1 Participants of the study Classes Teachers Students Class A One 29 year-old male language teacher; spoke Farsi as his native language; had obtained an overall score of 8.5 on the IELTS Academic Module; had taught English for nine years; held an MA in applied linguistics. Fifteen female and 10 male upper- intermediate students; ranged in age from 16 to 19; had been learn- ing English for three to four years. Class B One 26 year-old male language teacher; spoke Farsi as his native language; had obtained an overall score of 8 on the IELTS Academic Module; had taught English for four years; held a BA in TEFL. Seven female and nine male upper- intermediate students; ranged in age from 18 to 21; had been learn- ing English for three to four years. Class C One 31 year-old female language teacher; spoke Farsi as her native language; had obtained an overall score of 8.5 on the IELTS Academic Module; had taught English for seven years; held an MA in TEFL. Ten female and eight male upper- intermediate students; ranged in age from 16 to 20; had been learn- ing English for three to four years. Patterns of uptake and repair following recasts and prompts in an EFL context: Does feedback. . . 613 5.2. Context The EFL classes in this study constituted 22 90-minute sessions in total, and they were held three days a week over approximately two months. The aim of these classes was to develop all linguistic skills of language learners with a primary emphasis on oral and aural abilities, and a secondary emphasis on writing and reading skills, within the communicative orientation of language teaching and learning. The language institute where the classes were observed expected the teachers to follow task-based teaching together with some form-focused in- struction. To achieve this, in addition to employing teachers who were inspired by task-based and form-focused instruction, the institute held a number of in- structional meetings and workshops in order for the teachers to stay au courant with the latest advances. Over the course of such workshops, the teachers were encouraged to utilize adequate explicit instruction to make sure that together with the fluency of the learners, their accuracy also improves. Aside from these training sessions, the teachers were observed and given feedback twice a year by a more experienced professional. Also, every teacher was required to ob- serve at least 630 minutes of two other teachers’ classes each year to maintain consistency in teaching methodology. The policy of following task-based and form-focused principles was also implemented by the teachers of the classes in this study. Thus, although most of the time was spent on completing and working on meaning-focused tasks, at times, the teachers were observed to teach some of the linguistic features (es- pecially L2 grammar) explicitly. The textbook used in these classes was American English File 4 (Latham- Koenig & Oxenden, 2014). This textbook is task-based in design with the majority of the tasks focusing on meaning. It also provides learners with opportunities for both dyadic and group interactions and with a decent deal of written and oral input. Although the teachers used different tasks and activities (and sometimes different materials) from time to time, the basic teaching routine in these clas- ses did not seem to change considerably during the study. It was observed that the teachers usually divided the class time into three main sections. The first 15- 20 minutes of the class was devoted to reviewing the major topics that were worked on during the previous session. This section involved asking learners questions related to the linguistic features that were taught before. For instance, to check that conditionals had been learned, students were asked questions such as “what would you buy if you had one million dollars?,” “where will you go tonight if the weather is good?” and so on. When a student failed to provide a targetlike answer, the teachers usually provided CF. Reza Shirani 614 During the next 50-60 minutes of the class, the focus was on the topics and features that were introduced in the new lesson. This section usually began with some explicit instruction (e.g., an explicit grammar explanation, or a short lecture in L1) concerning particular linguistic features, together with some awareness-raising activities (e.g., input flood, input enhancement, and CF). It then continued with assigning learners to pairs and groups to do meaning- and form-focused tasks (as required by the textbook) collaboratively. Meanwhile, the teachers tended to walk around the classroom, monitor the students, and provide feedback when necessary. When all of the students were done with the task, each pair/group was asked to present the task to the whole class. Again, CF was provided, especially when an error was associated with the linguistic fea- tures that were supposed to be learned in that session. At the end of each session, the teachers tended to involve learners in whole-class discussions about real-world topics. This kind of activity was usually related to the linguistic features that were covered in that very session, and it contained a good deal of CF provided during teacher-student interactions. The teachers tended to avoid any kind of explicit explanation of grammatical rules at this point and also prevented the learners from using L1 for any purpose. As with previous sections, all the interactions in this section were recorded. 5.3. Data collection After debriefing sessions with the teachers, each class was observed and audio- recorded for eight sessions (36 hours of classroom time in total; 12 hours from each class). During observations, the observer sat unobtrusively at the back of the classrooms with no involvement in any of the activities and events. The ob- server wrote down notes to make real-time records of paralinguistic, nonverbal, and contextual events that might transcend the audio recordings. For example, it was observed that some feedback moves were followed by specific gestures and facial expressions instead of uptake. Example 1 (S = student, T = teacher) S: There are many people who overlook at the role of women in the society. T: They overlook the role of women in the society? According to the notes, the feedback in Example 1 was followed by the student nodding his head, which could be a sign of either understanding the correct form or providing a “yes” answer to the teacher’s interrogative utterance. Thus, the role of observing and taking notes was to ease the codification procedures, which are explained in the next section. Also, it should be noted that the classes Patterns of uptake and repair following recasts and prompts in an EFL context: Does feedback. . . 615 could not be video-recorded because permission to do so was not obtained from both the institute and the participants. To capture every possible interaction and feedback move such as un- planned and spontaneous interactions and feedback, the interactions were au- dio-recorded from the beginning of each session to the end. 5.4. Data codification Thirty-six hours of audio-recorded data were transcribed by the researcher and checked and confirmed by a research assistant to make sure that the transcrip- tion was clear and inclusive. Lyster and Ranta’s (1997) model was used to iden- tify the sequences of error treatment in the transcribed data. Lyster and Ranta explain their model as follows: The sequence begins with a learner’s utterance containing at least one error. The errone- ous utterance is followed either by the teacher’s corrective feedback or not; if not, then there is topic continuation. If corrective feedback is provided by the teacher, then it is either followed by uptake on the part of the student or not (no uptake entails topic con- tinuation). If there is uptake, then the student’s initially erroneous utterance is either re- paired or continues to need repair in some way. If the utterance needs repair, then cor- rective feedback may again be provided by the teacher; if no further feedback is provided, then there is topic continuation. If and when there is repair, then it is followed either by topic continuation or by some repair-related reinforcement provided by the teacher. Fol- lowing the reinforcement, there is topic continuation. (p. 45) The data were then coded for error, feedback types, and uptake based on the definitions below. 5.4.1. Error Learners’ utterances that included one or more phonological (e.g., incorrect pronunciation), grammatical (e.g., subject-verb agreement), or lexical (e.g., use of a wrong word) nontargetlike form(s) were coded as erroneous. For instance, in the following interaction move, the student’s utterance was coded as errone- ous because it contained a grammatically nontargetlike form. Example 2 S: My father used to went to the gym when he was younger. T: He used to go to the gym when he was younger. In some cases, the students’ utterances contained more than one nontargetlike form. Lyster and Ranta referred to these types of errors as multiple. However, in Reza Shirani 616 this study utterances with more than one error were coded the way utterances with one error were coded: Both were categorized as erroneous. 5.4.2. Feedback types Initially, three major categories of feedback, namely recasts, prompts, and oth- ers were identified. Recasts were coded as a broad type of CF that reformulated all or parts of a student’s erroneous utterance and that provided the targetlike form. Prompts were coded as feedback that pushed learners to correct their er- roneous utterances, and then were categorized as elicitations, metalinguistic clues, clarification requests, and repetitions (Lyster, 2004). Because the study fo- cused mainly on recasts and prompts, and because other types of feedback rarely occurred in the interactions, any other kinds of feedback were categorized as others (e.g., explicit correction, nonverbal feedback, and using L1). Moreover, two main types of recasts were recognized, namely explicit re- casts and implicit recasts, based on certain factors which will be discussed thor- oughly in Section 5.4.2.2. The data were thus coded for six types of feedback – four types of prompts and two types of recasts. 5.4.2.1. Prompts Clarification requests occur when the teacher uses such interrogative phrases as pardon me? or I’m sorry? to push the learner to reformulate or modify their utterance because there is either an error in form or lack of comprehensibility in meaning (Panova & Lyster, 2002). Example 3 S: You looked that movie? T: Excuse me? Elicitation refers to overt strategies aimed at eliciting the targetlike form from the students by indicating to them that their linguistic production is erroneous. Elicitation may occur in the form of overt questions like Would you come again? or requests for reformulation of an utterance. Also, the teacher may repeat the erroneous utterance up to the nontargetlike form and then pause in hopes that the student will fill in the blank with the correct form. Example 4 S: She’ll goes to the pool to tomorrow. I’m sure. T: Hmm, she’ll …? Patterns of uptake and repair following recasts and prompts in an EFL context: Does feedback. . . 617 Repetitions occur when the teacher repeats all or part of the student’s errone- ous utterance usually with a rising intonation to signal that the student’s utter- ance is erroneous. Example 5 S: We discussed about our school subjects and then had dinner. T: Discussed about? Metalinguistic clues are metalinguistic comments on the correctness of the student’s utterance in a declarative or interrogative form without providing the correct form. Example 6 S: I was pretty sure she had took the man to the hospital. T: In the past perfect tense, you need the past participle form of the verb. 5.4.2.2. Recasts: Explicit and implicit recasts Explicit and implicit recasts were distinguished based on four inherent features suggested by different researchers: length (Nassaji & Fotos, 2011; Sheen 2006), prosodic emphasis (Loewen & Philp, 2006; Nassaji & Fotos, 2011), intensity of focus on a particular linguistic feature (Ellis 2001; Erlam & Loewen, 2010), and the number of feedback moves (see Doughty & Varela, 1998; Loewen & Philp, 2006). These factors were considered for three reasons. First, among other char- acteristics (e.g., contextual, social, and cultural issues), the above-mentioned factors are considered inherent features of recasts. Second, despite controversy over other features (e.g., interrogative and declarative intonation), these char- acteristics have been widely agreed on in the literature. Third, an analysis of the data revealed that categorizing recasts based on these four factors was the most appropriate because there was no trace of any other characteristics of explicit recasts in the data. First, coding explicit recasts will be discussed. Short recasts are a single word or a short phrase with one content word (Sheen, 2006). Because short recasts reformulate only the erroneous part of the utterance, they are more salient. Example 7 S: I think it is really necessary that one washes his hands regularly. T: Wash. Recasts with prosodic emphasis reformulate the erroneous utterance by putting an intonational stress on the correct form. This may occur in either a declarative or an interrogative form. Reza Shirani 618 Example 8 S: Once upon a time, everyone used to enjoyed the clean weather in our city. T: Everyone used to enjoy the clean weather. Example 9 S: If they forbidded smoking, we would have a less polluted city. T: If they forbade smoking? Intensive recasts “focus on one or two particular linguistic structures” (Erlam & Loewen, 2010, p. 880). Intensive recasts are considered more explicit than ex- tensive ones (Ellis, 2001; Lyster, 1998b). The teachers in this study provided in- tensive recasts, particularly after they had taught the grammar and pronuncia- tion sections of the textbook. For instance, when the textbook concentrated on a specific grammar feature, say conditionals, they tended to constantly provide students with intensive feedback. Example 10 S: If I had had a car, I have not gone to work by bus. T: If I had had a car, I wouldn’t have gone to work by bus. Multi-move recasts contain at least one recast accompanied by another feed- back move in a single turn. Multi-move recasts are considered explicit because they offer a double emphasis on the nontargetlike form (see Doughty & Varela, 1998; Erlam & Loewen, 2010; Loewen & Philp, 2006). Example 11 S: I’m quite sure happiness was more rampant in the past. T: Happiness was rampant? It was everywhere. In Example 11, the teacher uses two feedback moves to address the same error (i.e., a repetition and a recast). As noted by Sheen (2006), multi-move recasts may occur (a) after the repetition of feedback (i.e., corrective recasts), (b) after another full or partial recast move (i.e., repeated recasts), and (c) with any other types of feedback other than explicit correction (i.e., combination recasts). The second category of recasts were implicit ones. Long recasts consist of more than two words (Erlam & Loewen, 2010, p. 880), and therefore their cor- rective nature may be implicit to the learner. Example 12 S: Animals protect their babies from dangerous. T: They protect their babies from danger. Patterns of uptake and repair following recasts and prompts in an EFL context: Does feedback. . . 619 Recasts without prosodic emphasis implicitly reformulate (with either a declar- ative or an interrogative intonation) the student’s erroneous utterance without putting an intonational stress on the correct form. Example 13 S: There are many people who overlook at the role of women in the society. T: They overlook the role of women in the society? Extensive recasts are concerned with a range of linguistic items any time during interaction without a deliberate focus on one or two particular linguistic features. Explicit recasts were coded as possessing at least one of the characteris- tics of explicitness discussed above. Inevitably, there were particular circum- stances where a recast entailed both an implicit feature (e.g., it was long) and an explicit feature (e.g., it was intensive). On such occasions, because the recast move contained at least one explicit feature, it was coded as explicit. Moreover, on a few occasions, explicit recasts exhibited more than one of the explicit fea- tures. For example, a recast could be simultaneously intensive and short. As with recasts entailing one feature of explicitness, recasts with more than one prop- erty of explicitness were coded as explicit recasts. On the other hand, implicit recasts were coded as possessing all implicit features and no explicit features. In other words, every single-move extensive long recast without any prosodic emphasis – whether interrogative or declarative – was coded as implicit. 5.4.3. Uptake Uptake was defined as “a student’s utterance that immediately follows the teacher’s feedback and that constitutes a reaction in some way to the teacher’s intention to draw attention to some aspect of the student’s initial utterances” (Lyster & Ranta, 1997, p. 49). In this study, two types of uptake were identified, namely repair and needs repair. Repair referred to a student’s successful correc- tion of a nontargetlike form. Example 14 S: She’s much more smarter than I expected. T: She’s much…? S: Much smarter. Needs repair referred to a partial correction or off-target reformulation of an erroneous utterance. Reza Shirani 620 Example 15 S: I had no idea what did he use to do as a child. T: I had no idea what he used to do as a child. S: No idea. Such acknowledgments as yes, ok, or yeah were also coded as needs repair because the student did not provide any correction. Moreover, the term no uptake referred to the cases when the student did not produce any verbal response to CF. 5.5. Reliability After codification of the data (from all of the classes) by the researcher, the re- search assistant recoded a random sample of 50% of the data to ensure the re- liability of the codification. Inconsistencies in coding were then negotiated. Ta- ble 2 shows the inter-rater reliability scores for erroneous utterances, broad cat- egories of CF, subsets of CF, and learner uptake. Table 2 Inter-rater reliability scores % agreement Erroneous utterances 87 Recasts 90 Explicit recasts 83 Implicit recasts 86 Prompts 91 Clarification requests 92 Repetitions 89 Metalinguistic clues 96 Elicitation 87 Others 89 Uptake and uptake types 93 5.6. Data analysis The study used frequency data and simple percentages to describe the frequen- cies of different types of CF as well as the relationship between each feedback type and uptake/repair. It should be noted that for the same purposes, some of the prior studies (e.g., Sheen, 2006) have employed chi-square tests. However, as Nassaji (2007, p. 532) correctly points out, One of the assumptions of the chi-square is that the data must be independent within and across cells. Although chi-square has sometimes been used in SLA re- search in such conditions, the independent assumption is violated if one participant contributes more data to one cell than the others. Patterns of uptake and repair following recasts and prompts in an EFL context: Does feedback. . . 621 Thus, like in many other studies (e.g., Lyster & Ranta, 1997; Nassaji, 2007; Panova & Lyster, 2002), simple percentages were used to analyze the data. 6. Results 6.1. Feedback frequency The numbers and percentages of the occurrence of the general types of CF are presented in Table 3. Following 678 erroneous utterances, prompts constituted more than half of the feedback moves and were the most frequently used type of feedback, whereas recasts occurred after 36.6% of the erroneous utterances. Table 3 The distribution of broad categories of CF Prompts Recasts Others N % N % N % Class A 234 57.3 157 38.5 17 4.2 Class B 201 61.2 121 36.8 6 1.8 Class C 243 60.9 142 35.5 14 3.5 Total 678 59.2 420 36.6 37 3.2 To compare the distribution of different subsets of recasts and prompts, further analyses were conducted. As can be seen in Table 4, the most frequently used subtypes of prompts in all of the classes were elicitations and repetitions, which occurred after 229 and 213 erroneous utterances respectively. It should be noted that the difference in the distribution of elicitations and repetitions was too small. The least frequent subtype of prompts, however, was metalin- guistic clues, which constituted 10.4% of all prompt moves. Table 4 The distribution of different subsets of prompts Elicitation Repetitions Clarification requests Metalinguistic clues N % N % N % N % Class A 76 32.5 74 31.6 51 21.8 33 14.1 Class B 61 30.0 60 29.8 54 26.8 26 12.9 Class C 92 37.8 79 32.5 60 24.6 12 4.9 Total 229 33.7 213 31.4 165 24.3 71 10.4 Within recasts, as shown in Table 5, whereas explicit ones were provided after 255 erroneous utterances, 165 recast moves possessed implicit features. Reza Shirani 622 Table 5 The distribution of different subsets of recasts Explicit recasts Implicit recasts N % N % Class A 90 57.3 67 42.7 Class B 66 54.5 55 45.5 Class C 99 69.7 43 30.3 Total 255 60.7 165 39.3 6.2. Feedback and uptake Further, the numbers and percentages of learner uptake after each broad cate- gory of feedback were calculated. The results are shown in Table 6. As for prompts, 91.0% led to uptake and only 9.0% did not. As for recasts, 62.6% led and 37.4% did not lead to uptake. As can be seen, prompts led to much greater percentages of uptake than did recasts. Table 6 The number and percentage of uptake following general categories of CF Uptake No uptake N % N % Prompts (N = 678) 617 91.0 61 9.0 Recasts (N = 420) 263 62.6 157 37.4 Total 880 80.1 218 19.9 The frequencies of uptake following subtypes of recasts and prompts were also compared. Table 7 presents the results. Within prompts, clarification led to the greatest percentage of uptake (97.6%). Likewise, elicitation (91.7%) and repetitions (91.5%) led to great percentages of uptake. Thus, the difference in uptake was quite insignificant when elicitations, repetitions, and clarification requests were provided. However, metalinguistic clues led to a smaller percent- age of uptake compared to other types of prompts. Table 7 The relationship between subtypes of feedback and learners’ uptake Uptake No uptake N % N % Prompts Elicitation (N =229) 210 91.7 19 8.3 Repetitions (N = 213) 195 91.5 18 8.4 Clarification requests (N = 165) 161 97.6 4 2.4 Metalinguistic clues (N = 71) 51 71.8 20 28.2 Recasts Explicit recasts (N = 255) 193 75.7 62 24.3 Implicit recasts (N = 165) 95 70 42.4 57.6 Patterns of uptake and repair following recasts and prompts in an EFL context: Does feedback. . . 623 Furthermore, as Table 7 shows, whereas explicit recasts led to 75.7% of uptake, implicit recasts resulted in 42.4% of uptake. Therefore, as the above analyses suggest, the implicitness and explicitness of recasts might have influ- enced the frequencies of uptake. Apparently, explicit recasts were more likely to provoke a learner response. 6.3. Feedback and repair The frequencies of leaner repair following each broad category of CF were cal- culated based on the total moves of (a) each feedback type and (b) immediate uptake. As Table 8 shows, 41.6% of prompts led to repair and 49.4% led to needs repair. Also, 45.7% of uptake following prompts contained repair and 54.3% in- cluded needs repair. This seems to suggest that whenever prompts did lead to uptake, the possibility of repair was less than half of the whole uptake moves whereas the likelihood of needs-repair was more than half of the total uptake. As for recasts, 46.6% led to repair and 15.9% led to needs repair. Moreover, 74.5% of uptake following recasts contained full repair whereas 25.5% of uptake needed further repair. In total, recasts led to a larger repair percentage with re- gard to both the number of feedback moves and the frequencies of uptake. Table 8 The relationship between general categories of feedback and learners’ repair Repair Needs repair N % of feedback % of uptake N % of feedback % of uptake Prompts 282 41.6 45.7 335 49.4 54.3 Recasts 196 46.6 74.5 67 15.9 25.5 Total 478 43.5 54.3 402 36.6 45.7 The relationship between different subtypes of feedback and learner repair was also analyzed. As can be seen in Table 9, among prompts, elicitation led to the greatest amount of repair (51.5% of feedback and 56.2% of uptake). The smallest per- centage of repair, however, was associated with repetitions (35.7% of feedback and 39% of uptake). Repetitions and uptake following them, therefore, offered the highest percentage of needs repair vis-à-vis other prompts. Moreover, as far as the ratio of repair to feedback moves was concerned, clarification requests (43.4%) were found to result in a higher percentage of repair compared to metalinguistic clues (39.4%). However, when the ratio of uptake and repair was considered, whereas 44.7% of up- take following clarification requests brought about repair, 54.9% of uptake following metalinguistic clues was the cause of learners’ repair. Therefore, the extent to which prompts led to repair could noticeably vary according to whether the total numbers of feedback moves or the overall frequencies of uptake were considered. Reza Shirani 624 As for recasts, the explicit ones were associated with a larger repair per- centage than the implicit ones. Whereas 58.2% of explicit recasts and 67.9% of uptake following such feedback led to repair, 21.8% of implicit recasts and 51.4% of uptake following them were related to full repair. This implies that when an implicit recast did lead to uptake, on more than half of the occasions, the uptake was likely to contain the correct form. Moreover, the above analyses seem to suggest that when recasts are accompanied with some characteristics of explic- itness, they are more likely to provoke repair. Table 9 The relationship between subsets of feedback and learners' repair Repair Needs repair N % of feedback % of uptake N % of feedback % of uptake Prompts Elicitation 118 51.5 56.2 92 40.2 43.8 Repetitions 76 35.7 39.0 119 55.9 61.0 Clarification requests 72 43.4 44.7 89 54.0 55.3 Metalinguistic clues 28 39.4 54.9 23 32.4 45.1 Recasts Explicit recasts 131 58.2 67.9 62 27.5 32.1 Implicit recasts 36 21.8 51.4 34 20.6 48.6 In summary, according the analyses, learner repair might be, to some extent, influenced by the way feedback was provided. When the students were given the correct form, and when they responded to such feedback, the possibility of repair was stronger than when the learners were invited to self-repair. Moreover, it should be noted that although prompts led to a higher percentage of uptake, recasts, es- pecially explicit ones, might lead to a higher percentage of repair than prompts. This might be particularly true when the ratio of uptake to repair is considered. 7. Discussion The first research question in this study concerned the frequency of recasts and prompts and their subtypes in the Iranian EFL context. In all of the three classes, prompts occurred more frequently than recasts. These findings might imply teachers have a tendency to encourage students to self-repair rather than providing the correct form instantly. The findings stand in contrast to those of many studies (e.g., Ellis et al., 2001; Fu & Nassaji, 2016; Nabei & Swain, 2002; Panova & Lyster, 2002) in which the higher frequency of recasts has been ac- counted for in terms of the assumption that recasts perform a more communi- cative function than prompts and that they result in fewer communication breakdowns as they keep the interlocutors’ attention primarily on meaning. Yet, Patterns of uptake and repair following recasts and prompts in an EFL context: Does feedback. . . 625 some studies (e.g., Kennedy, 2010; Lyster & Ranta, 1997) have shown that for more proficient learners, teachers usually provide prompts more than recasts because such learners are believed to have adequate linguistic resources to rely on when they are prompted to self-repair. Accordingly, the higher frequency of prompts in this study might be attributed to the participants’ proficiency level, which was rated as upper-intermediate by the language school, and to their con- siderable language learning experience, which ranged from three to four years. Furthermore, data analysis revealed that among prompts, elicitation, an explicit feedback type (Nassaji & Fotos, 2011), was the most dominant type of feedback before repetitions, clarification requests, and metalinguistic clues. Similarly, explicit recasts occurred more frequently than implicit ones. The teachers’ tendency to provide more salient and explicit feedback types may be attributed to the context of the study. Unlike students in many other contexts, Iranian learners have minimum exposure to English outside the classroom, and to compensate for this, most of the language schools, including the one in this study, hold training courses and workshops for their teachers to make sure that students receive appropriate instruction. Such training courses are usually taught by an invited professor or researcher who provides a theoretical and practical presentation of appropriate CF techniques. During the training courses at the language school in this study, the teachers were encouraged to use suffi- cient explicit feedback to make sure that students would notice the corrective nature of feedback in some way. In one of the between-semester workshops, for example, it was observed that the invited researcher made frequent refer- ences to the studies that have confirmed the positive role of explicit instruction and explicit feedback in the Iranian EFL context. Thus, the teachers’ training background might play a role in how CF was provided in this study. Another factor that might have encouraged the teachers to use explicit feedback frequently might have to do with Iranian students’ preferences for dif- ferent types of CF. The present researcher did not conduct any formal surveys to explore the participants’ preferences and perceptions of CF, but previous re- search (e.g., Kaivanpanah, Alavi, & Sepehrnia, 2015) has shown that the major- ity of Iranian EFL learners view CF as positive and even necessary. Very casual post-study interviews with a few random participants also revealed their prefer- ence for explicit feedback. A 19-year-old male student from Class A, for example, said: “I think we deserve to get feedback from teacher. I want to know where I am wrong. I just want to correct it.” Another 17-year old-female student from Class C stated: “When I do not realize my mistakes in conversation, I will repeat them again and again. I think teachers should let us know.” The second research question concerned the extent to which recasts, prompts, and their subtypes lead to learner uptake in a FL context. Similarly to Reza Shirani 626 previous studies (e.g., Fu & Nassaji 2016; Lyster & Ranta, 1997; Panova & Lyster, 2002), prompts were found to provide more uptake opportunities vis-à-vis recasts. This may be related to the efficacy of prompts in pushing learners to produce out- put; according to Swain’s (1993) output hypothesis, pushing learners to produce output helps them notice gaps in their interlanguage as they struggle to convey their meaning. To compensate for their gaps, learners are pushed to make use of their linguistic resources, engage in hypothesis testing, and reflect on their output more carefully. The greater uptake opportunity following prompts might be at- tributed to their greater saliency and their elicitative force (Yang & Lyster, 2010). Moreover, it was found that within prompts, clarification requests led to the largest uptake percentage. This might be explained by the assumption that clarification requests were used after either miscomprehension of the message or emergence of an error (Panova & Lyster, 2002). As the observer took notes, even when the teachers provided clarification requests to encourage the stu- dents to self-repair an error, the students sometimes responded to such re- quests as if there were lack of message comprehensibility. The following extract from the data may exemplify that kind of situation: Example 16 S: All he had to do was went running to lose weight. T: Sorry? S: Running! He could lose weight if he did sports. T: So… All he had to do was go running. S: Yes. As can be seen, in the first sentence, the learner uses the word went when the base form of the verb should have been used. When the teacher makes a clari- fication request, the learner, according to the researcher’s notes, along with his oral language production, gesticulates and imitates running with his body lan- guage, which is probably an attempt to get the message across. The teacher’s second attempt to provide CF (in the form of a recast), however, indicates that the intention behind the first feedback is probably corrective. The third research question concerned the extent to which recasts, prompts, and their subtypes lead to repair in a FL context. Compared to prompts, recasts led to a higher percentage of full repair both when the ratio of feedback to repair and the ratio of uptake to repair were considered. Of course, when examining the efficacy of recasts on measures of repair, it should be noted that recasts may not elicit student-generated repair other than the repetition of the feedback, and therefore, those feedback types that give learners more lan- guage production opportunities might be more effective (Lyster 1998a, 1998b; Lyster & Ranta 1997; Panova & Lyster 2002). Patterns of uptake and repair following recasts and prompts in an EFL context: Does feedback. . . 627 Moreover, it was found that within prompts, although clarification requests led to many uptake moves, they were not as effective as elicitations and metalin- guistic clues – the more salient types of prompts – as far as learner repair was con- cerned. The role of feedback saliency in generating a higher percentage of repair is even more tangible within recasts, among which those with explicit features were associated with a larger number of both uptake and repair moves when compared to implicit ones. These findings confirm Loewen and Philp’s (2006) finding that the efficacy of recasts could depend considerably on their characteristics. They were found to be more effective if combined with more explicit features like stress, mul- tiple feedback moves, shortened length, and interrogative intonation (see also Sheen, 2006). Nassaji (2007), too, concludes that the extent to which a feedback type results in learner repair may depend on how explicitly it is provided and that whenever a feedback move is combined with salient intonational or verbal cues, learners are more likely to generate uptake with repair. 8. Conclusions, implications, and directions The findings of this study suggest that the efficacy of prompts and recasts in an EFL context may depend on how explicitly they are provided. In other words, these find- ings and the findings of some other studies (e.g., Doughty & Varela, 1998; Loewen & Philp, 2006; Nassaji, 2007; Sheen, 2006) seem to imply that the effectiveness of CF may depend on how successfully it draws learners’ attention to form. This con- firms Schmidt’s noticing hypothesis (1995), whereby in order for CF to be effective, it is important for learners to notice the corrective nature of such feedback so that they can make use of it to improve their interlanguage. The fact that more explicit feedback types led to higher percentages of repair verified the effective roles of ex- plicitness, salience, and noticing – three interrelated concepts. Moreover, it should be noted that in developing an in-depth understand- ing of CF, uptake and repair should not be ignored. Chaudron (1997, p. 440), for instance, claimed that “the main immediate measurement of effectiveness of any type of corrective reaction would be a frequency count of the students’ cor- rect responses following each type.” Accordingly, although providing uptake does not indicate language acquisition (Loewen, 2004; Panova & Lyster, 2002), it may contribute to L2 development as it indicates that the learner has noticed the feedback and used it in some way (Mackey & Philp, 1998; Sheen, 2004). In addition, immediate repair has been believed to contribute to language devel- opment to some extent (McDonough, 2005; Nassaji, 2007; Pica, 1994). Lastly, even if one assumes that uptake does not play a major role in acquisition, it is still quite clear that it influences language use. Reza Shirani 628 Now that learner uptake and repair seems to be fair criteria to assess CF, it should be further noted that investigating learner uptake in relation to different fac- tors (explicitness/implicitness of feedback in this study) is of vital importance in an EFL context, particularly in Iran, where there is usually minimum exposure to L2 out- side the classroom, and where a variety of instructional techniques and curricula are used across language schools. Reporting that in an EFL context the effectiveness of different types of CF may be positively influenced by how explicitly they are pro- vided, this study may have practical implications for language teachers and praction- ers who seek to provide their students with the most beneficial kind of feedback. When considering the findings of this study, one, however, should exer- cise due caution. For one thing, this study was conducted with Iranian upper- intermediate students in one single language school, and therefore generalizing the findings to other students with different levels of proficiency might be prob- lematic, especially given that previous studies (e.g., Mackey & Philp, 1998; Philp, 2003) have found that the efficacy and distribution of different types of CF vary significantly across different proficiency levels. Thus, an open issue for future research could be how the effect of CF explicitness and implicitness vary under the influence of learner proficiency. Also, as mentioned earlier, some (e.g., Loewen, 2004; Panova & Lyster, 2002) have suggested that uptake should not be equated with language acqui- sition. Thus, a more appropriate approach to CF may be illustrated by carefully controlled experimental or quasi-experimental studies (e.g., Lee & Lyster, 2016; Nassaji, 2017) that address the relationship between CF and the acquisition of particular linguistic forms. Such studies may provide a more tangible insight into the relationship between CF and actual language learning. Finally, this study investigated CF effectiveness in terms of how explicitly it is provided. However, there might be other factors that influence the efficacy of CF. For example, such learner-related variables as motivation, anxiety, atti- tude towards an L2, and so forth may all play a role in how learners benefit from CF. Another open issue for future research, therefore, is how the usefulness of CF varies under the influence of such factors. Acknowledgements I would like to thank all of the people who have contributed to this research. I am grateful to the participants of the study, and the two anonymous reviewers whose insightful comments helped improve the work. My special thanks also go to Dr. Jakub Bielak, the editor, and Dr. Mirosław Pawlak, the editor-in-chief of Studies in Second Language Learning and Teaching. All of the errors in the manuscript are mine. Patterns of uptake and repair following recasts and prompts in an EFL context: Does feedback. . . 629 References Brown, H. D., & Lee, H. (2015). Teaching by principles: An interactive approach to language pedagogy. White Plains, NY: Pearson Education. Chaudron, C. (1977). A descriptive model of discourse in the corrective treat- ment of learners’ errors. Language Learning, 27, 29-46. Doughty, C., & Varela, E. (1998). Communicative focus on form. In C. Doughty & J. Williams (Eds.), Focus on form in classroom second language acquisition (pp. 114-138). Cambridge: Cambridge University Press. Ellis, R. (2001). Introduction: Investigating form-focused instruction. Language Learning, 51(s1), 1-46. Ellis, R., Basturkmen, H., & Loewen, S. (2001). Learner uptake in communicative ESL lessons. Language Learning, 51, 281-318. Erlam, R., & Loewen, S. (2010). Implicit and explicit recasts in L2 oral French in- teraction. Canadian Modern Language Review, 66, 877-905. Fu, T. & Nassaji, H. (2016). Corrective feedback, learner uptake, and feedback perception in a Chinese as a foreign language classroom. Studies in Second Language Learning and Teaching, 6(1), 159-181. Kaivanpanah, S., Alavi, S. M., & Sepehrinia, S. (2015). Preferences for interac- tional feedback: differences between learners and teachers. The Lan- guage Learning Journal, 43, 74-93. Kennedy, S. (2010). Corrective feedback for learners of varied proficiency levels: A teacher’s choices. TESL Canada Journal, 27, 31-50. Krashen, S. (1982). Principles and practice in second language acquisition. Ox- ford: Pergamon Press. Latham-Koenig, C., & Oxeden, C. (2014). American English file 4. Oxford: Oxford University Press. Lee, A. H., & Lyster, R. (2016). The effects of corrective feedback on instructed L2 speech perception. Studies in Second Language Acquisition, 38, 35-64. Lightbown, P. M., & Spada, N. (1990). Focus-on-form and corrective feedback in communicative language teaching. Studies in Second Language Acquisi- tion, 12, 429-448. Llinares, A., & Lyster, R. (2014). The influence of context on patterns of corrective feedback and learner uptake: A comparison of CLIL and immersion class- rooms. The Language Learning Journal, 42, 181-194. Loewen, S. (2004). Uptake in incidental focus on form in meaning-focused ESL lessons. Language Learning, 54, 153-188. Loewen, S., & Philp, J. (2006). Recasts in the adult English L2 classroom: characteristics, explicitness, and effectiveness. Modern Language Journal, 90, 536-556. Reza Shirani 630 Long, M. (1996). The role of the linguistic environment in second language ac- quisition. In: W. Ritchie, & T. Bhatia (Eds.), Handbook of second language acquisition (pp. 413-468). San Diego, CA: Academic Press. Lyster, R. (1998a). Recasts, repetition, and ambiguity in L2 classroom discourse. Studies in Second Language Acquisition, 20, 51-81. 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. Lyster, R. (2004). Differential effects of prompts and recasts in form-focused in- struction. Studies in Second Language Acquisition, 26, 399-432. Lyster, R., & Mori, H. (2006). Interactional feedback and instructional counter- balance. Studies in Second Language Acquisition, 28, 269-300. Lyster, R., & Ranta, L. (1997). Corrective feedback and learner uptake: Negotia- tion of form in communicative classrooms. Studies in Second Language Acquisition, 19, 37-66. Mackey, A. (2012). Input, interaction and corrective feedback in L2 classrooms. Oxford: Oxford University Press. Mackey, A., & Philp, J. (1998). Conversational interaction and second language development: Recasts, responses, and red herrings?. Modern Language Journal, 82, 338-356. McDonough, K. (2005). Identifying the impact of negative feedback and learn- ers’ perceptions on ESL question development. Studies in Second Lan- guage Acquisition, 27, 79-103. Nabei, T., & Swain, M. (2002). Learner awareness of recasts in classroom inter- action: A case study of an adult EFL student's second language learning. Language Awareness, 11, 43-63. Nassaji, H. (2007). Elicitation and reformulation and their relationship with learner repair in dyadic interaction. Language Learning, 57, 511-548. Nassaji, H. (2017). The effectiveness of extensive versus intensive recasts for learning L2 grammar. Modern Language Journal, 101, 353-368. Nassaji, H., & Fotos, S. S. (2011). Teaching grammar in second language class- rooms: integrating form-focused instruction in communicative context. New York: Routledge. Nicholas, H., Lightbown, P. M., & Spada, N. (2001). Recasts as feedback to lan- guage learners. Language Learning, 51, 719-758. Norris, J. M., & Ortega, L. (2000). Effectiveness of L2 instruction: A research syn- thesis and quantitative meta-analysis. Language Learning, 50, 417-528. Oliver, R. (1995). Negative feedback in child NS-NNS conversation. Studies in Sec- ond Language Acquisition, 17, 459-481. Patterns of uptake and repair following recasts and prompts in an EFL context: Does feedback. . . 631 Panova, I., & Lyster, R. (2002). Patterns of corrective feedback and uptake in an adult ESL classroom. TESOL Quarterly, 36, 573-595. Pica, T. (1994). Research on negotiation: What does it reveal about second-lan- guage learning conditions, processes, and outcomes? Language Learning, 44, 493-527. Philp, J. (2003). Constraints on “noticing the gap.” Studies in Second Language Acquisition, 25, 99-126. Sheen, Y. (2004). Corrective feedback and learner uptake in communicative classrooms across instructional settings. Language Teaching Research, 8(3), 263-300. Sheen, Y. (2006). Exploring the relationship between characteristics of recasts and learner uptake. Language Teaching Research, 10, 361-392. Schmidt, R. (1995). Consciousness and foreign language learning: a tutorial on the role of attention and awareness in learning. In R. Schmidt (Ed), Atten- tion and awareness in foreign language learning (pp. 1-65). Honolulu: University of Hawai’i Press. Schwartz, B. D. (1993). On explicit and negative data effecting and affecting competence and linguistic behavior. Studies in Second Language Acquisition, 15, 147-163. Swain, M. (1993). The output hypothesis: Just speaking and writing aren't enough. Canadian Modern Language Review, 50, 158-164. Swain, M. (2005). The output hypothesis: Theory and research. In E. Hinkel (Ed.), Handbook on research in second language teaching and learning (pp. 471- 483). Mahwah, NJ: Erlbaum. Truscott, J. (1996). The case against grammar correction in L2 writing classes. Language Learning, 46, 327-369. Yang, Y., & Lyster, R. (2010). Effects of form-focused practice and feedback on Chinese EFL. learners’ acquisition of regular and irregular past tense forms. Studies in Second Language Acquisition, 32, 235-263.