Immediate Uptake of Phonological Corrective Feedback in Language Learning and Retention

As language learners’ phonological errors have attracted substantial attention, error treatment strategies have become an indispensable part of teachers’ repertoire. Research has found positive effects for corrective feedback on language learner uptake; however, the effect has not been proved to be sustained over time.,is quasiexperimental study sought to explore whether uptake can reflect language learning and retention through measuring the effectiveness of three common types of oral corrective feedback on Iranian EFL learners’ phonological errors. Fifty-four male intermediate-level learners received a nine-session treatment in the form of recast, elicitation, and metalinguistic feedback during story retelling tasks. Results of comparing and correlating uptake with posttest scores revealed that while recast was found to be the most effective feedback in inducing correct uptake, it was metalinguistic feedback that proved to be the most conducive in learning and retention. Besides, there was no significant relationship between the learners’ scores in uptake and their learning and retention in any groups. ,is suggests that EFL learners’ immediate reactions to teachers’ input-providing or output-prompting correction could not be a reflection of language development, and more consistent and continuous long-term assessment of the success of corrective feedback has to be envisaged in language teaching methodologies.


Introduction
Pronunciation is an integral component of language and a vital element in effective communication, gaining an increasingly pivotal role in language teaching and testing approaches and methodologies [1,2].Celce-Murcia et al. [3] propose various traditional techniques and materials for teaching pronunciation, including recitation, reading aloud, recording learners' production listening, and imitation.Many of such techniques are drawn on in novel approaches, most exclusively focusing on "getting the sound right at the word level," in isolation, and in "very controlled and contrived sentence-based environments" rather than at a discourse level (p.10) [3].
According to Brown [4], phonology appears to be the most reliable linguistic category to predict language learners' performance since other microlinguistic levels of language such as lexis or grammar considerably show more variation than psychomotor-based phonological interference among language learners."Even presumably simple grammatical categories like word order, tense, or aspect have been shown to contain a good deal of variation" (p.212).Pronunciation, however, has frequently been problematic for language learners and hence a source of concern for teachers.Learners usually have a hard time pronouncing words owing to the specific phonetic system of the target language, absence of a sound in their native language or lack of consistency between the ways a grapheme is written and the way it is pronounced [5].Navehebrahim [2] has shown that Iranian EFL learners' inability to differentiate between the English and Persian sound systems (e.g., in contrasting/θ/and/t/as two distinctive and phonetically realized sounds) is due to their feeble English language proficiency, whereas many like Ziahosseiny [6] ascribe this to the phenomenon of proactive inhibition (i.e., pull of mother tongue).
In fact, as language learners start speaking, they are bound to make errors.ese errors might take various forms and aspects, classified as phonological, lexical, syntactic, or pragmatic.ey may be oral or written, in production or in perception, local or global, and a result of L2 ignorance (intralingual) or due to L1 interference (interlingual).ey might be systematic errors or performance errors.ey can be linguistic or extralinguistic [4,[6][7][8][9][10].On the other hand, teachers are recommended to take different actions in response to errors.In more recent communicative approaches to language learning, errors have been considered as the outcome of the development of communication skills and hence tolerated especially when the focus is on fluency, while the major global ones are corrected unobtrusively during accuracy-based activities [11].In his interaction hypothesis, Long [12] claims that interaction and negotiation of meaning can help learners in noticing their nontarget output by which they can enhance their accuracy.
As the focus of language teaching methods has oscillated between accuracy and fluency, the view towards error correction has also been varying.According to Scrivener [13], "if the objective is accuracy, then immediate correction is likely to be useful; if the aim is fluency, then lengthy, immediate correction that diverts from the flow of speaking is less appropriate" (p.299).
Many theoretical and empirical studies were conducted on error treatment particularly in the 1970s, 1980s, and 1990s, and providing corrective feedback (CF), along with conscious-raising activities that proved useful in enhancing learners' creativity, self-confidence, and autonomy as well as language proficiency [14,15].CF (with a scope broader than the term "error correction") is most often used to address learners' ill-formed utterances, incomplete or ungrammatical sentences, functional incorrectness, inaccurate word choices, and inter alia, phonological errors.Many researchers have endeavored to determine the type, frequency, amount, various moves, recommended guidelines, and required contexts as mostly favored by teachers and learners, this being an undeniable reflection of the ever-changing trends in language teaching [16][17][18].
Literature has witnessed various stances regarding the practicality and effectiveness of CF.For Krashen [19], error correction is a "serious mistake."Chun et al. [20] demonstrated that CF rarely occurs in classroom settings.Nunan [21] prefers positive feedback to CF, arguing that it serves the two major functions, that is, informing the students that they have performed correctly and increasing their motivation through praise.Ur [22] puts a certain place for correction, yet warns against over estimating this contribution.According to Scrivener [13], students learn more when they do things themselves instead of being told about these things.Van-Patten [23] argues that error correction has a negligible effect on language system development.In UG-based acquisition, Ellis [24] believes that negative evidence does not play a substantive role, contributing only to explicit knowledge.
To the proponents of interventionist approaches and cognitive and sociocultural theories, however, CF usually has a facilitative role [25][26][27][28][29]. Ellis [25] has endorsed the effectiveness of CF in different linguistic and psycholinguistic approaches through fostering learner motivation and ensuring linguistic accuracy.Nassaji [17] lays great emphasis on the role and efficacy of interactional feedback in language learning in various settings, focusing on several key areas in research including feedback timing and training, learnerlearner interaction, and computer-assisted feedback.In his seminal metaanalysis of 33 studies on oral corrective feedback (OCF), Li [30] revealed that OCF had a medium effect on learning, the effect being far greater in FL settings than SL contexts, where the experiment was conducted in laboratories than in classrooms and when the feedback was more implicit.Chu [1] demonstrated that CF can certainly lead to more enhanced oral accuracy, yet the effectiveness varies in terms of the learners' language proficiency, with medium-level and low-level learners being more favored by the intervention.Based on his studies on the cognitive, interactionist, and sociocultural theories of learning, Ellis [31] proposed guidelines for teachers in providing CF, one of the most significant of which is the necessity of familiarity with and implementing various oral and written CF strategies, as well as creating appropriate space following the corrective move for learners so as to uptake the correction.
Various classifications of CF have been offered.Lyster and Ranta [32] have introduced six feedback types in two broad CF categories: reformulations and prompts.Reformulations include recasts and explicit correction, both supplying learners with target reformulations of their nontarget output.Prompts include a variety of signals other than reformulations that push learners to self-repair (i.e., elicitation, metalinguistic clues, clarification requests, and repetition).Nassaji [33] has introduced two major categories, namely, elicitations and reformulations.Based on whether CF strategy is aimed at providing input or prompting output, Ellis [31,34] has classified CF strategies into recasts, repetitions, and clarification requests as the implicit forms, and explicit correction, metalinguistic comments, and elicitation as the explicit types.Sheen [35] has distinguished between focused (intensive) and unfocused (extensive) CF.Hendrickson [9], Ur [22], Harmer [36], and Hedge [37] have offered a series of strategies for OCF, including modeling, discussing, echoing, using gestures, questioning, asking for clarification, requesting repetition, and direct indication.
Recast, elicitation, and metalinguistic feedback are regarded as three instructionally useful strategies in language teaching.Recast is the reformulation and rephrasing of an error, usually with a confirmatory tone, minus the error.Based on Braidi [38], a response can be coded as recast if it incorporates "the content words of the immediately preceding incorrect nonnative speaker's utterance" and also changes and corrects "the utterance in some way (e.g., phonological, syntactic, morphological, or lexical)" (p.20).Elicitation prompts learners to self-correct, a distinctive characteristic of which is that it elicits and encourages a self-repair without making any direct reference to the error [33].Asking questions, pausing, and asking for clarification, repetition, or reformulations are the most common types of elicitation techniques [32].Metalinguistic feedback is usually in the form of comments, questions, and information regarding the correctness of an utterance with no provision of the correct form.It involves "providing learners with some form of explicit comment about the nature of the errors they have made" (p.100) [32].
e teacher might 2 Education Research International provide some codes, enquiries, explanations, or descriptions to help the learner reflect upon the linguistic feature and correct his error.According to Nassaji [39], recasts appear to be more effective than elicitations in immediate effects.Lyster and Ranta [32], however, found that the use of elicitations to correct mispronunciations can better improve learning.Soleimani et al. [40] showed elicitation to be more effective than more explicit types in correcting Iranian EFL learners' phonological errors.Nicholas et al. [41] state that recasts are helpful devices particularly in laboratory studies rather than classroom settings.Vahdani and Khabbazi Alavi [42] found that recasts are the most frequently used type of CF, whereas direct and explicit corrections are regarded as the least popular feedback type among Iranian learners.Recasts are usually considered to be at the implicit end of CF spectrum [43], although they can constitute an implicit-explicit continuum [39,44].Long [43] argues that recasts are really crucial and that they can provide positive evidence for short-term language development.In the same vein, Lyster and Izquierdo [45] claim that recasts are as effective in pushing the learners to self-correct as other more explicit types.Based on Lyster [26], however, recasts have a greater frequency yet lower efficiency.According to Lyster and Ranta [32], explicit correction and metalinguistic clues outperform the more implicit types like recast.Rassaei et al. [46] also showed that lowerintermediate Iranian learners who received metalinguistic feedback outperformed those who received recast during task-based interactions.Loewen and Philp [47] found metalinguistic information to be highly conducive to correct uptake.Gitsaki and Althobaiti [48] also showed that repetition and metalinguistic feedback led to the highest rate of successful uptake in dealing with learners' errors.Yang [49] studied EFL learners' OCF preferences with regard to their proficiency level, cultural background, and types of error.Learners preferred metalinguistic feedback, explicit correction, and recasts for most error types, respectively.Besides, intermediate learners found clarification requests on phonological errors more effective as compared to those at the elementary level.Recast was also found to be more helpful in dealing with phonological errors.
To demonstrate the efficacy of CF, however, a line of demarcation needs to be drawn between immediate uptake and retention.Uptake is the learners' accurate recognition and correction of an erroneous utterance or linguistic feature right after the teacher's correction usually in the form of repetition or self-repair.Lyster and Ranta [32] define uptake 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 utterance" (p.49).According to Sheen [50], uptake does not mean learning, but it can be an indicator of whether the learner has noticed the targeted linguistic form.Retention, however, is the ensuing production or comprehension of a target language feature or pattern.According to Richards and Schmidt [51], retention is the ability of remembering things after a lapse of time.
e importance of successful uptake for learning is a matter of controversy and empirical questions.Lyster [26] asserts that immediate uptake leads to more profound processing of the linguistic feature which, in turn, is critical to language learning.Similarly, McDonough [52] and Nobuyoshi and Ellis [53] found that learners' successful error repairs in response to teachers' CF in the form of clarification request could significantly predict the learning of the past tense and development of question formation.Ohta [54] studied the oral language that adult FL learners of Japanese addressed to themselves in classroom activities and interactions which mainly focused on grammar and metalinguistic instruction.Analyzing the learners' private speech, she found that even if recasts did not sometimes result in immediate uptake, they were noticed by the learner, especially when it was directed to other learners or to the whole class rather than to the one who made the error.Ellis et al. [16] conducted a descriptive study on adult language learners' immediate responses to their teachers' different types of feedback.It was revealed that recasts were the most frequent teachers' responses to the errors, to most of which the learners immediately reacted, and that learners' noticing and reacting to such recasts positively contributed to their learning.Gitsaki and Althobaiti [48] studied the types and effectiveness of different interactional types of CF to 28 beginners and intermediate learners' lexical, phonological, and grammatical errors, and found that although explicit correction was the most frequent type in both classes, repetition and metalinguistic feedback were more effective since they led to the highest rate of uptake.Iraji et al. [55] studied the role of OCF in 100 Iranian male and female EFL learners' phonological uptake in classroom talks and interactions.A strong relationship was shown between the instructors' CF and the learners' uptake.Also, male learners appeared to be more successful than their female counterparts.
Efficacy of uptake in relation to language learning regarding different types of feedback in terms of degree of explicitness has been the focus of a number of studies.Nassaji [39] studied the immediate (i.e., uptake) and delayed effects of recasts (as implicit CF) and elicitations (as explicit CF) on 42 adult learners.Results showed that recasts were more effective than elicitations in immediate identification and correction of the linguistic forms (e.g., the use of to be verbs and gerunds) during interactions.In both CF types, the more explicit form was more effective than its implicit form.In another similar study, Nassaji [56] investigated the relationship between learner-and teacher-generated repair in response to elicitation and recast and the learning of targeted linguistic forms in dyadic interactions.e results revealed that learners could recognize and correct more than half of the errors that they had already repaired during interaction, and that learner-generated self-repair (i.e., elicitation) was more effective over time.Similarly, Saeidi and Raveshi [57] have also shown that whereas recast results in more repaired uptake, explicit elicitation leads to more durable and sustained acquisition.
In contrast, Loewen [58] could not find any relationship between L2 learners' error repair in response to teachers' recasts and their scores in the delayed posttests.Lyster and Ranta [32] argue that even if repair happens after correction, Education Research International it may not be very beneficial for learning.Long [43] also asserts that uptake should not be equated with language learning.According to Swain [59], uptake does not so much contribute to accuracy than to fluency.Mackey and Philp [60] studied the effect of intensive recasting on advanced learners' development of morphosyntactic forms in question formation.ey found that the learners' immediate response to recasts was not a reliable predictor of whether they could subsequently make further use of the CF.
ey regarded uptake as a methodological "red herring" fallacy and not a suitable outcome measure of L2 learning.
Despite the vastness and variety of all the studies conducted so far on the role of CF, the contribution of immediate uptake of CF to language learning has not been extensively probed in academic circles.e present study, therefore, aims at investigating the pedagogical value of uptake of phonological CF by posing the following research questions: (1) Do learners who score higher in the uptake of recast, elicitation, and metalinguistic feedback perform better in the immediate and delayed posttests?(2) Is there any relationship between uptake of oral corrective feedback and learning and retention of the targeted sounds among the learners?

Methodology
2.1.Participants.A total of 57 Iranian male EFL learners in three intact groups participated in this quasiexperimental study.e learners were first homogenized through the standard Preliminary English Test (PET) as a basic-level qualification.Eliminating three students with the most extreme scores on the test gave an equal number of 18 intermediate-level learners, with an approximate age range of 16 to 24, in each class.e learners were encouraged to seriously participate and engage in all class activities including the treatment sessions and the assessments.ree proficient English language teachers, who were previously instructed and familiarized with the study procedure and the type of required feedback, also participated in this study; each of them taking charge of one class.e teachers and students spoke with the American English accent, which is usually the norm in most Iranian EFL classes.

Materials and Instruments.
A PET test was administered to homogenize the learners, a pretest was employed to evaluate them before the treatment, and two posttests were drawn on to measure their learning and retention.Several checklists were used as the measurement instruments, in which the frequency of uptake/no uptake for each type of CF and correct/incorrect pronunciations were calculated and tabulated for the learners in each group.Further, to collect data on the elicited mispronunciations of the learners, two appropriate storybooks published by Oxford Bookworms Series with the same difficulty level were chosen, "Agatha Christie, Woman of Mystery" by Escott [61] ( e Flesch Reading Ease score: 59.1) to elicit the data for the pretest and the two posttests and "Hamlet" by McCallum [62] ( e Flesch Reading Ease score: 61.6) to be used during the treatment sessions.A sound recorder was also used in order to audio record the teacher's CF and the learners' errors and repairs during the pretest, the posttests, and the treatment sessions for further reanalysis and double-checking of the filled-out checklists.

Data Collection
Procedure.After the homogeneity of the learners was ensured, the classes were randomly assigned to three groups to receive three types of OCF, i.e., recast, elicitation, and metalinguistic feedback.Each class met twice a week and lasted for 105 minutes.Approximately, 50 minutes of each session was allocated to the study, and the rest of the class time was allotted to their course-book " e American English File" and follow-up activities.It took two and a half months for the whole study to be completed.e classes were first observed for two sessions, based on which three most frequent types of phonological errors that learners had made while participating in class activities were recorded.It is to be noted that, in this study, learner errors are referred to as incorrect deviancies from the standard American English pronunciation.Based on the frequency of the errors, the data gathered from groups during the observation became the target of the study and the basis for comparison and scoring.In fact, the observation helped preparing a small taxonomy of the most common pronunciation errors, which were directly selected on the basis of the learners' performance in their classes and not predetermined by the researcher.
e most frequent phonological errors were found to be (1) the incorrect pronunciation of the sound [w] as in "well" pronounced as [v], (2) mispronunciation of the lax vowel [i] as in "thin" pronounced as [i:], and (3) mispronunciation of consonant clusters as in "street" pronounced with an added vowel, usually [e] or [i], accounting for 59.8 percent of the overall errors.It is to be noted that many English short vowels (e.g., [ə] or [ʌ] are not present in Persian phonetic system.Besides, the syllable structure of the two languages are different; while English permits up to three consonants in its onset and up to four consonants in the coda (i.e., CCCVCCCC), the Persian language permits only one consonant its syllable onset and a maximum of two in the coda (i.e., CVCC). is difference causes numerous errors among Iranian learners of English [63,64].
In order to measure a starting point, a pretest was conducted in session three, right after the observation and after the three types of errors to be subjected to OCF were chosen.
e students were first instructed and then required to perform a retelling task which took two minutes.Meanwhile, the observer carefully took down the number of times each targeted sound was pronounced correctly or incorrectly by each learner during this focused pronunciation task.
e treatment then commenced which lasted nine sessions, during which all the learners were supposed to retell to the class the assigned sections of another storybook, while their preselected phonological errors were corrected by the teacher using a specific OCF, and the frequency of the pronunciation errors together with self-repairs or correct repetition after teacher's OCF was recorded by the observer For any incidence of an error, the feedback was only offered one time to check the learners' reaction.In case a teacher did not provide the required OCF for any reason, the instance was eliminated from the list.It is noteworthy that the term "treatment" has often been used for teachers' corrective practices in response to their learners' errors [1,30,39].
Afterwards, in order to further ascertain the effect of OCF on the students' learning and retention, the groups took two posttests: one posttest was administered in the session immediately after the treatments, and one delayed posttest was carried out after a one-month time interval.Both tests again involved a 2-minute story retelling task for each learner, recording the correct/incorrect pronunciation of targeted sounds, and the same scoring procedure was used in the pretest.
At the beginning of the study, the learners' informed consent was obtained, and they were told that they were participating in a research project centering on story retelling; however, they were unaware of the exact objectives, the teachers' remedial activities, and the true reason for the presence of the observer.Following the delayed posttest, and in order to address the ethical issues, they were given the opportunity to be thoroughly debriefed of the nature, purpose, and focus of the study.

Data Analysis.
e data analysis and scoring were performed by two raters (inter-rater reliability � 0.82).A percentage calculation of the variables (i.e., the frequency of the number of uptake and no uptake and also the number of correct and incorrect pronunciations in the pretest and posttests) was first conducted.To quantify the performance of the learners in uptake and the tests, a score was given to them out of 20, based on the ratio of the incidence of uptake to the total number of occurrences of the corresponding feedback and the ratio of correct pronunciations to the total number of occurrence of the targeted sounds (i.e., aggregate of correct and incorrect pronunciations).e scoring formula for uptake is as follows: Total number of teacher's CF � uptake + no uptake Raw score � number of uptake total number of CF Score out of 20 � number of uptake × 20 total number of CF e scoring formula for the pretest and the posttest is as follows: Total frequency of the targeted sounds Such scoring procedure has not been applied in previous studies, where the number of learner uptake has only been counted.Computing the frequency of learner uptake might seem to be simpler to conduct and more objective to consider.However, it might pose a critical research problem as well.A CF of a specific type leading to a higher number of uptakes (irrespective of the number of no uptakes) turns out to be more successful, yet this conceals the fact that the learner might have received a higher number of CFs and hence had more uptakes; a learner corrected 10 times with four successful uptakes does not outperform the one who has been corrected only four times and has three successful uptakes.As a result, considering the number of times there was no uptake of a CF type is also a crucial step to be taken in studying the effectiveness of CF in terms of uptake. is became the basis of the scoring procedure in this study.

Results
e first research question concerned whether learners who score higher in uptake of recast, elicitation, and metalinguistic feedback perform better in the immediate and delayed Education Research International posttests.After the required assumptions (i.e., normality and homogeneity of variances) were ensured, several one-way ANOVAs were administered in order to determine the differences among the groups in uptake and the posttests.Tables 1-3 display the descriptive statistics, results of the ANOVA, and Tukey post hoc test for the uptake scores.
e results of one-way ANOVA showed that F(2,51) � 6.73 and p < 0.05.Tukey post hoc test revealed that the recast (R) group received the highest score and outperformed the metalinguistic feedback (M) and elicitation (E) groups.e M group, in turn, performed better than the E group.Tables 4 and 5 illustrate the descriptive statistics and the results of ANOVA for the pretest scores.
e results of one-way ANOVA (F(2,51) � 1.99, p > 0.05) revealed that the there was no significant difference among the groups in their pretests (i.e., before the treatment sessions).Tables 6 and 7 illustrate the descriptive statistics and the results of ANOVA for the immediate posttest scores.
e results of one-way ANOVA (F(2,51) � 0.13, p > 0.05) revealed that there was no significant difference among the groups in the immediate posttest.Tables 8-10 present the descriptive statistics, results of the ANOVA, and Tukey post hoc test of the delayed posttest scores.
e results of one-way ANOVA (F(2,51) � 5.98, p < 0.01) and Tukey post hoc test showed that the M group significantly averaged higher than the recast and elicitation groups.e group means across the three testing times (i.e., uptake, immediate posttest, and delayed posttest) are also presented in Figure 1.
e plot visualizes the interaction among the variables of time and group.e scores increased form the pretest to the immediate posttest, and all the groups improved; Education Research International however, the rate then dropped in the delayed posttest.Time, therefore, seems to trade off the effect of the treatment especially in the case of recast and elicitation; the effect of the metalinguistic type of OCF came out to be more sustained.is indicates that although the R group averaged the highest as regards the uptake, it was the M group that had the best performance in taking in the teachers' CF.Regarding research question one, therefore, higher uptake did not lead to greater learning/retention as the learners who had a higher score in uptake of recast, elicitation, and metalinguistic feedback could not perform better in the immediate and delayed posttests.Language uptake does not seem to be effectively employed by learners in learning through phonological corrective feedback.
To determine the relationship between the learners' scores in their uptake and posttests, Pearson's product-moment correlation coefficients were calculated, after the assumptions of normal distribution of the data, linearity, and the homogeneity of variances were ensured.e results of the correlation tests for the three groups are presented in Tables 11-13 ( e scores for the uptake and posttests were all given on a scale of 0 to 20 and were hence comparable).
No significant correlation is found in the experimental groups between the scores in the uptake and those in the immediate posttest or between the scores in the uptake and those in the delayed posttest.Regarding research question two, there is no relationship between uptake of OCF and language learning or retention among the learners.

Discussion
As it was expected in this experimental research, the three groups first improved, and their scores increased form the pretest to the immediate posttest because of the structured and consistent nature of the intervention.However, while recast was found to be the most effective CF type in inducing correct uptake, it was metalinguistic feedback that proved to be the most conducive strategy for the learners in retention.e more explicit the type of correction, the more influential it is in leading to sustained learning [32].Saeidi and Raveshi [57] have also shown that Iranian upper-intermediate EFL learners find recast more useful in prompting correct uptake as compared with elicitation.e results are also in line with many other studies such as those by Allen and Mills [65], Engwall and Bälter [66], Lee [67], Lyster [26,68], Loewen and Philp [47], Panova and Lyster [69], and Rassaei et al. [46], who claim that CF would cause better results when it comprises metalinguistic information.Other researchers provide evidence for the usefulness of other types of CF.Doughty and Varela [70], for instance, found that recast can improve English pronunciation.
Although the present study, as in Lyster and Ranta [32], Lyster [26,68], and Ammar and Spada [71], corroborated the usefulness of explicit types of feedback in dealing with students' errors in language learning and use and in fostering accuracy in language development, others argue that CF has a place in both accuracy and fluency.Ellis [31], for example, recommends teachers "identify specific linguistic targets for correction in different lessons. is will occur naturally in accuracy work based on a structure-of-the-day approach but can also be usefully applied in fluency work" (p.14).
e results of comparing the groups' performance in uptake (i.e., repetition in response to recasts and self-repair following elicitation and metalinguistic feedback) and the posttests revealed that higher uptake, by no means, shows more sustained learning.e learners who had a higher score in uptake could not perform better in the posttests.In a similar vein, the results of the correlation analysis revealed the absence of a significant relationship between uptake and learning of the targeted sounds indicated in the posttests.
e students had seemingly forgotten the correct pronunciation of the sounds they were once corrected for and they pronounced correctly right after the teachers' feedback.
is revealed that uptake does not reflect or predict retention, immediate self-correction or repetition does not always guarantee learning, and equating uptake and learning can be a methodological fallacy.
Whatever reason one might seek for such inconsistency in immediate response and follow-up command (such as pull of mother tongue, attrition, or insufficient attention), it goes without saying that relying only upon the incidence of learners' uptake in gauging the effectiveness of CF would be inconclusive.Uptake does not show learning; early gains might be on the wane over time since learners need ample time to process and analyze the negative evidence they receive.
Lack of uptake does not show lack of learning either [28]; no immediate reaction by the learners might be due to conversational constraints and does not necessarily mean that they have not acquired a feature or pattern.In line with it, Mackey and Philp [60] found that although L2 learners showed little evidence of uptake (5%), they could internalize the knowledge provided by feedback.Focusing on the effectiveness of OCF, Panova and Lyster [69] claim that failure to uptake might be due to the learners' uncertainty in interpreting the feedback.According to Yoshida [72] and Nassaji [56], a learner might simply repeat the teacher's reformulation even without noticing or processing the error; on the other hand, he may grasp the erroneous part and learn from the feedback but does not immediately respond to it.
Russel [18] also believes that the social dynamics between teacher and learners might affect the interpretations of the feedback that may, in turn, affect the learners' uptake.Moreover, uptake can sometimes occur in the form of covert or private speech which is not easily identifiable but still beneficial [24].e results of this study, hence, are congruent with findings obtained by a number of researchers, [10,17,18,32,43,56,58,66,68,[72][73][74], who claim that uptake cannot be a plausible measure of language learning.According to Lyster and Ranta [32], language learning cannot be reliably measured by uptake; even if repair happens after correction, it may not be very beneficial for learning.Long [43] asserts that noticing the teachers' correction is far more important than the subsequent uptake.Lyster [10] found that recasts are not uptaken by learners since they attend to meaning, not from.According to Nabei and Swain [73], uptake is an important determiner in understanding the impact of CF; however, it cannot mean that learning has occurred since learners' uptake does not necessarily represent cognitive processing of CF.Williams [74] asserts that the most optimum way to appraise language knowledge is by the subsequent perception and production of a component or skill.
Although uptake might indirectly pave the way for learning, it is not always a reliable indicator, and the beneficial effects of CF need to be perused in light of its robustness and durability.Reviewing the literature on the association between feedback uptake and learning, it turns out that there is a dearth of research in this respect.Ellis [24] finds little evidence to show that uptake-with-repair enhances the impact of CF.If no such association is ultimately found between uptake and learning, then it might pose the question why they are sometimes regarded the same and how teachers are supposed to interpret and deal with their students' immediate reactions to their feedback.
e results might thus challenge the studies which equate phonological uptake with improvement in language attainment and examine the effect of OCF only in short periods of time, for example, [16,26,35,48,52,53,55,75], where it has been postulated (or implied) that the learners' immediate uptake of the teacher's feedback indisputably contributes or relates to language development.According to Nassaji [17], in many studies on interactional CF, the effectiveness has been measured merely through uptake whereas long-term effects have not been explored.

Conclusion
A large number of investigations have been conducted on the effectiveness of oral CF and the conditions under which it works better.Facets of such studies have been captured in various theoretical and practical stances.Nonetheless, the nature of error correction in regard to uptake, learning, and retention, and the corresponding types of discourse and strategies applied in EFL contexts have not been extensively studied [76,77].Nassaji [39] has shown the paucity of such comprehensive and thoroughgoing studies on the role of OCF in language learning contexts.e results of this study might pose a serious question:; if uptake of CF cannot reflect or contribute to language learning, then why should learner errors be immediately corrected at all?Should learners be interrupted by the teacher when they make serious errors or CF needs to be postponed to the end of each class activity?On the other hand, errors need to be corrected; otherwise the erroneous forms might become permanent and fossilize in the learners' transitional competence.
Having a clear idea of the nature, type, and frequency of errors and knowing the reason for their occurrence would help teachers adopt the best strategy in tackling their learners' learning problems.In language teaching classes, where the major goal is to foster communication abilities, microlinguistic errors have to be mostly tolerated in favor of attempts for negotiation and expression of meaning. is obviates the need for immediate corrective response of the teacher to an erroneous utterance.According to Keshavarz [8], teachers should not interrupt learners to correct their grammatical, lexical, or pronunciation errors unless the deviation hinders communication or distorts intelligibility.In later stages, however, where the class activities and exercises are oriented towards accuracy enhancement, teachers may apply various awareness-raising techniques to familiarize their students with their errors and ways to correction and prevention of linguistic deviations, bearing in mind that learners might get the same corrective feedback in different ways and therefore react to it differently.
In addition, the nature, type, and complexity of CF together with individual and psychological factors such as ethnicity, gender, age, language proficiency, personality types, learning styles and goals, attitude, preferences, and motivation might be acknowledged to affect learners' reactions to CF [73].According to Zhang et al. [78], for instance, language learners usually prefer to be immediately corrected for their phonological errors but not for the grammatical or lexical ones.Teachers need to be very careful when implementing a teaching or corrective strategy and longing for its effect on improving their students' language knowledge and skills.Nothing has to be prepackaged but needs to be tailored based on these factors, as one man's meat is another man's poison.
is is what dynamics and diversity mean in human beings.Recast, as a type of input providing might be the most commonly used type of OCF in one setting, the most beneficial one in another, and the least effective in the other.Even the nature of uptake might influence its relationship with language learning [17].
e obtained findings would therefore open up novel and interesting questions for further research agendas.e study might be replicated in broader educational contexts enjoying various data collection tools and tasks.New procedures need to be worked out for accurate operationalization and assessment of uptake.Future researchers are recommended to provide a well-defined taxonomy of types of L2 learners' errors at different levels of proficiency and the most frequent types of CF to these errors.Based on such sample-based categorizations, the effectiveness of immediate and delayed CF can be comprehensively studied, where certain types of CF might be found to be more effective in dealing with certain types of errors for certain learners at certain times.As many Education Research International

�Score out of 20 �
correct pronunciations + incorrect pronunciations Raw score � correct pronunciations of the targeted sounds × 20 total frequency of the targeted sounds correct pronunciations of the targeted sounds × 20 total frequency of the targeted sounds

Table 1 :
Descriptive statistics for the learners' scores in uptake.

Table 2 :
One-way ANOVA for the uptake scores.

Table 3 :
Tukey post hoc test for the uptake scores.

Table 4 :
Descriptive statistics for the learners' scores in the pretest.

Table 5 :
One-way ANOVA for the pretest scores.

Table 7 :
One-way ANOVA for the immediate posttest scores.

Table 8 :
Descriptive statistics for the learners' scores in the delayed posttest.

Table 9 :
One-way ANOVA for the delayed posttest scores.

Table 10 :
Tukey post hoc test for the delayed posttest scores.

Table 6 :
Descriptive statistics for the learners' scores in the immediate posttest.

Table 11 :
Correlations between measures of uptake and posttests for the M group.

Table 12 :
Correlations between measures of uptake and posttests for the E group.

Table 13 :
Correlations between measures of uptake and posttests for the R group.