The Role of Error Types and Feedback in Iranian EFL Classrooms

To facilitate successful language learning, teachers need to establish positive affect among students yet also engage in the interactive confrontational activity of error correction (Magilow, 1999). To shed more light on the issue, this study aims at the investigation of the error types, corrective feedback moves, and learner uptake, i.e., responses to feedback in Iranian communicatively-oriented EFL classrooms. The database is drawn from transcripts of audio-recordings of the elementary and high intermediate classes of a language institute including almost 12 hours of interaction among the students and teachers. Following the analysis, the errors were coded as grammatical, lexical, phonological or unsolicited use of L1 (first language) and corrective feedback moves as explicit correction, recast, clarification request, metalinguistic clues, elicitation, or repetition. Moreover, the suggested breakdown for uptakes included student-generated repair, repetition, and needs repair. Grammatical errors were the most frequent error type in the entire database (50.5%); however, phonological (26%) and lexical errors (22%) had lower rank error type. Moreover, the results indicated an overwhelming tendency for the teachers to use recasts (50.5%) in spite of their complete ineffectiveness at eliciting student-generated repair. Repetition (96%), metalinguistic feedback (86%), elicitation (67.5%), and clarification request (44%) –the negotiation of form feedback moveswere instead supposed to fulfill this aim.


Introduction
Classroom interaction has been widely studied in the field of Second Language Acquisition (SLA).Inside the field, different viewpoints on classroom interaction and discourse have been examined which have mainly concentrated on either teachers or students and their speech.When the focus is set on second language teaching and learning, and more specifically on English as Foreign Language (EFL) classrooms, the main issue is the language itself and how it is used in the interaction between participants in the classroom.
One of the areas studied inside SLA is corrective feedback, which occurs when a student produces an oral error, or an incorrect utterance of some sort.This erroneous utterance is usually followed by the teacher's reaction resulting in some type of corrective feedback.Furthermore, after the feedback move on the part of the teacher, the student may show signs of learning or understanding that implies the student's reaction to the teacher's feedback.
Contributing in this field of study, Lyster and Ranta (1997) focused on error, feedback, and uptake in their observations and analysis of four French immersion classrooms.They found that the teachers used six different feedback moves, of which the most frequent one was recast (correction without additional information), followed by elicitation ('milking' the answer from the students) and clarification requests.The other feedback types (metalinguistic feedback, explicit correction, and repetition) were all used during less than 10% of the feedback situations.Furthermore, the results showed that the least likely feedback type to result in any kind of uptake was recasts, whereas elicitation was the most successful feedback type leading to learner uptake.The findings proved that recasts are not a good technique to use, if teachers wish to engage students during lessons and achieve learner uptake.Lyster (2001) examined corrective feedback and its relationship with different error types.Furthermore, he wanted to discover the immediate effects of the feedback types by examining learner repair (uptake).The database in this study comprised of audio-recordings and transcripts which were then closely examined to find the different error types, feedback moves, and learner repairs.Lyster concluded that grammatical and phonological errors seemed to provoke recasts more than lexical errors.Moreover, the results showed that learner repair occurred in the form of repetitions after recasts, and peer-and self-repairs followed negotiation of form (which includes elicitation, metalinguistic feedback, clarification requests, and repetition feedback moves).Panova and Lyster (2002) conducted a similar study to that of Lyster and Ranta (1997) in which they focused on adult English as Second Language (ESL) classroom.The researchers categorized corrective feedback moves under seven different terms: recasts, elicitation, metalinguistic feedback, repetition, translation, clarification request, and explicit correction.The results showed that the most frequently used feedback moves were recasts and translations.Panova and Lyster concluded that the low proficiency level of the students and their incapability in correcting their own errors may have been the reason why the teacher used recasts so frequently.In an attempt to examine the rates of uptake, the lowest rates of learner uptake occurred when the teacher used a recast or explicitly corrected an error.Moreover, explicit corrections received no learner repair and recasts and translations obtained low rates of learner repair.
The importance of studying and examining corrective feedback can be seen when actual classroom discourse events are looked at more closely.Most of the interaction that takes place during foreign language lessons is guided by teachers.Thus, it is important to improve the teachers' knowledge of their own actions so that they get aware of the corrective feedback techniques they can use.Furthermore, since the students can learn from their own errors, correction of those errors plays a crucial role in their learning.
Since there were numerous studies on the different corrective feedback types and learner uptake (Lyster & Ranta, 1997;Lyster, 2001;Panova & Lyster, 2002;Lyster & Mori, 2006;Surakka, 2007), the present study is hardly a pioneering study in this particular area.However, most of these studies have been conducted in immersion classrooms or adult ESL settings, and not many studies have focused on EFL classrooms and on Iranian EFL ones.Furthermore, in the previous studies the data have been collected from the same grade level or from the same classroom.The present study; however, includes a comparative study of different error types, corrective feedback and learner uptake between elementary and high intermediate communicatively oriented classrooms.The present data will also show the differences in the occurrence of the type of errors and corrective feedback moves between the levels.
To enlighten teachers, this research also tries to look at the students' side whether there are clear signs of student uptake following the different corrective feedback moves.In an almost significant way, this study tries to show the different ways in which a teacher can correct a student's oral errors considering their proficiency level and the error type.Hendrickson (1978) goes on to claim that most teachers know nothing about the types of corrective feedback they provide.He claims that "most teacher training programs fail to prepare teachers to handle the variety of errors that occur inevitably in students ' speech" (p. 392).This is a gap in teacher education that should not be disregarded as something unimportant, since students may need corrective feedback when "they are not able to discover, through exposure alone, how their interlanguage differs from the L2" (Lyster et al., 1999).Additionally, the present study aims to provide information about the connection between corrective feedback and learner uptake.It can be claimed that some corrective feedback types give students the opportunity to find correct answers on their own, while some types implicitly correct an error and do not provide additional information, nor give opportunities for learner uptake.Which corrective feedback types, then, actually lead to learner uptake?Krashen (1985) having proposed Input Hypothesis which holds that acquisition occurs when learners understand input containing structures beyond the students' current level of competence, Swain (1985) suggested Output Hypothesis being one of the basic theoretical claims on which the notions of corrective feedback and uptake have been developed.It stated that comprehensible input (Krashen, 1985) alone does not improve learners' language acquisition in terms of syntax, and the production of output in response to input is necessary for further language development.With regard to learner production, Swain emphasized the importance of the role of modified output, arguing that it is necessary for second language mastery.Swain further suggested that modified output could be the result of ample opportunities for output and the provision of useful and consistent feedback from teachers and peers.Later, she proposed that modified output is the representation of "the leading edge of a learner's interlanguage" (Swain, 1995, p. 131).

Theoretical background
In the context of teacher-learner interaction, modified output can be manifested in the form of learner uptake, or learner reaction to the teachers' corrective feedback given to learners' error.Corrective feedback, for example, is a pedagogical means of offering modified input to students, which could consequently lead to modified output by the students.Based on Long's Interaction Hypothesis (1983), corrective feedback and learner uptake could be considered effective processes for language development, especially when corrective feedback and learner uptake serve as a source of interactional modification.
Another theoretical basis for discussing the effectiveness of corrective feedback can be found in the argument that language learning may require negative evidence, or information about what is ungrammatical.Regarding the learnability argument that comprehensible input may not be sufficient for acquisition, researchers such as White (1987White ( , 1989) ) have argued for a need for negative evidence if second language learners' aim is to attain native-like proficiency.Negative evidence is considered effective when learner hypotheses based on first language (L1) structure lead to second language (L2) overgeneralizations that are impossible to overcome on the basis of positive evidence alone (Long, 1996).As negative evidence is given in response to the erroneous forms that learners produce, it can take the form of corrective feedback in the context of classroom interaction.
Learner uptake is also examined in relation to the notions of attention (Logan, 1988) and noticing (Schmidt, 1990(Schmidt, , 1995)).Logan stated that, in the course of language learning, attention is necessary and sufficient for extracting items (i.e., linguistic input) from a stimulus array.When this claim is extended to the discussion on how to determine which items students have attended to, it may be pointed out that uptake is one way of showing which items learners have attended to in the preceding corrective feedback.Similarly, Schmidt (1995) proposed the notion of noticing as a subjective manifestation of attention, and also asserted that noticing is a necessary and sufficient condition for converting input to intake.Although there has been no research that explicitly demonstrates that uptake is an oral manifestation of noticing, it may well be speculated that there are some noticed linguistic features involved in learners' uptake.In relation to Schmidt's (1995) idea that noticing facilitates the acquisition of input, Long (1996) examined the context in which noticing could occur, and argued that selective attention can be paid most effectively during negotiation for meaning.Moreover, he maintained that feedback obtained during negotiation work might be facilitative of the L2 development in vocabulary, morphology, and syntax.That is, negotiating communication difficulties creates comprehensible input, which contains salient linguistic features that learners could notice more easily than other features.This claim might also support the notion of the effectiveness of the feedback-uptake sequence, especially when feedback focuses on linguistic errors and gives learners an opportunity to negotiate what was incorrect in their original utterances with their interlocutors.

Key Terms
For the purpose of the present study, the three main terms that appear in the field of corrective feedback are discussed below.They include error, corrective feedback, and uptake.
One of the earliest definitions describes an error as "an utterance, form, or structure that a particular language teacher deems unacceptable because of its inappropriate use or its absence in real-life discourse" (Hendrickson, 1978, p. 387).This definition seems to convey both the correct form of an utterance by native-speaker standards as well as teachers' judgment calls.Furthermore, some researchers have simply declared that "an error is a form unwanted by the teacher" (Allwright & Bailey, 1991, p. 85).In this study an error is seen as: (1) an objective evaluation of linguistic or content errors according to linguistic norms or evident misconstrual of facts, and (2) any additional linguistic or other behaviors that the teachers reacted to negatively or with an indication that improvement of the response was expected (Chaudron, 1986, p. 67).
(2) This definition covers the term error in full, since it not only includes the linguistic aspect and grammatical correctness, but also concentrates on the teachers' views of what is acceptable and what is not.
Negative evidence or negative feedback "refers to the input that tells learners what is not possible or grammatical in the target language" (Sheen, 2004, p. 296).Corrective feedback is a type of negative evidence and it can be defined as "any indication to the learners that their use of the target language is incorrect" (Lightbown & Spada, 1999, p. 171) and since it does not always provide the correct form, it will force the learners to make use of their own language knowledge.Furthermore, corrective feedback can be either implicit or explicit.Explicit feedback types are overt and offer clear information for learners about their errors.These feedback types can be, for example, grammatical explanations (e.g., metalinguistic feedback, and elicitation) that are sometimes grouped under the term negotiation of form, which refers to "the provision of corrective feedback that encourages self-repairs involving accuracy and precision and not merely comprehensibility" (Lyster & Ranta, 1997, p. 42).Implicit feedback types are less obvious and do not provide any additional information to students on the correct formulation of their utterances.If a teacher gives implicit feedback to a student, it usually does not interrupt the flow of the conversation, but simply corrects the error (Long, 1996).
Generally speaking, uptake occurs, when a teacher's feedback move is followed by a student utterance.Lyster and Ranta (1997, p. 49) describe uptake as "a student's utterance that immediately follows the teacher's feedback and constitutes a reaction in some way to the teacher's intention to draw attention to some aspect of the student's initial utterance (this overall intention is clear to the student although the teacher's specific linguistic focus may not be)".Additionally, learner uptake can be divided into two types: "(a) uptake that results in 'repair' of the error on which the feedback focused and (b) uptake that results in an utterance that still needs repair (coded as 'needs-repair')" (ibid.).Moreover, there might not always be an uptake move and "if there is no uptake, then there is topic continuation" (ibid.).In other words, the teacher (or possibly a student) continues the lesson with the same or a different topic.

Method
The data for the present study included the transcribed material from 4 audio-taped lessons in elementary and high intermediate levels.The participants were 35 elementary (with 20-30 age-range) and 25 high intermediates (with 16-25 age -range) female EFL students of Iran Language institute (ILI), whose levels were determined by the same institute in the outset; they were also the students of the same communicatively oriented.
The two female teachers were asked to record two sessions of their elementary and intermediate classes, i.e., two elementary and two intermediate sessions by the same teacher using a Sony digital music player.The teachers were given the necessary instructions concerning the recording and voice adjustments and were asked to put it in their pocket or somewhere unnoticed by the students.The instructions, indeed, were given with the purpose of providing a clear corpus for the analysis with the least amount of deletions and ignorance in the transcription of the data.The lessons selected for the analysis, excluded formal grammar or spelling lessons because the primary purpose of this study was providing a description of how teachers and students engage in error treatment during communicative interaction and the selected lessons represented a more communicative orientation.
The database for the present study included about 12 hours of recorded interaction between the teachers and their students (about six hours in each level) transcribed by the researcher.The teachers were selected on the basis of their willingness to have their lessons tape-recorded.Although the teachers knew the interest in recording classroom interaction, they were unaware of the research focus related to corrective feedback.There was no control over the way the teachers conducted their classes and they were not instructed to use any particular kinds of feedback or to focus on any particular type of error.Due to this study's focus being on the analysis of the teacher behavior, the focus was exclusively based on teacher-student interaction.Even when the students were involved in group work, the interactions were captured as the teacher went from group to group interacting with the students.
After collection of a rich corpus of class interaction, the data was transcribed and coded according to the model adapted from Lyster and Ranta (1997).The turns consisting of utterances with at least one error were codified pertaining to error type (grammatical, lexical, phonological, or use of L1), corrective feedback move (recast, explicit correction, clarification request, metalinguistic feedback, elicitation, or repetition) and the following uptake by the learner (student-generated repair, repetition, or needs repair).Afterwards, the data was imported into SPSS and Chi-square was run in search of significant differences between the levels in each of the areas under study and comprehensive answers to the research questions.

Data Analysis
The analytic model used to code the interactional data, adapted from Lyster (2001).The sequence begins with a learner's utterance containing at least one error.The non-target 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 continuation).
If there is uptake, then the student's non-target utterance is either repaired or continues to need repair in some way.If the utterance needs repair, then the teacher may again provide corrective feedback; if no further feedback is provided, then there is topic continuation.When there is repair, then it is followed either by topic continuation or by some repair-related sign of approval from the teacher.Following approval there is topic continuation.
The coding definitions for the present study were adapted from those developed by Lyster and Ranta (1997).
They described the error treatment sequence as learner error, teacher feedback, and learner uptake, and the present study follows their description thereof.Error types are categorized as grammatical, lexical, phonological, and use of L1; corrective feedback types are labeled as recasts, explicit correction, clarification requests, metalinguistic feedback, elicitation, and repetition of error; and finally, uptake is divided into repair and needs-repair.

Type of Errors
In this study, ill-formed utterances were classified as having either one or more than one error.Although there were only a few instances of combinations of error types in a learner turn in this data set, the category of 'multiple error' included two types of combinations of which one was coded and the other disregarded.
The first type was a combination of two similar error types, i.e., grammatical, lexical or phonological where the teacher's feedback targeted both of the errors.Here is one such instance: Example: (Elementary) S: I have a accident.(Error -grammatical) T: You had an accident.(Feedback recast) In the student's first utterance, she made two grammatical errors in the use of have (wrong tense), and the determiner a.The teacher provided corrective feedback in the form of recast responding to both of the grammatical errors.In this case, the student's first utterance was coded as grammatical error and the feedback coded as recast.
The second type was composed of two different errors.Although the feedback targeted both of the errors, it was not possible to codify the utterance based on one type of error.Since one of the purposes of this study is to see what kind of error leads to which type of corrective feedback, the current study focused on actual sequences of learner error and teacher feedback.Learner multiple (different) errors that were responded to with one corrective feedback were not included in the analysis of error treatment sequence in the current study.One instance is provided below: Accordingly, in the present study, 4 types of error were analyzed: grammatical errors, lexical errors, phonological errors, and unsolicited uses of L1.
Grammatical errors were non-target like use of determiners, prepositions, pronouns, number agreement, tense, verb morphology, and auxiliaries.Additionally, errors in pluralization, negation, question formation, and word order were considered as grammatical errors.The following example suggests one of these cases with an error in the use of correct preposition on: Phonological errors were inaccurate pronunciation of words that often led to difficulty of comprehension of the target words.Even in the cases where the mispronounced words were comprehensible to the teacher, the words were still considered to have phonological errors when they were given corrective feedback.Consider the following example in this regard: Occasionally a language learner resorted to her first language (unsolicited use of L1), even when the use of L1 was not allowed.Even if the student was capable of producing a certain utterance in English, the teacher might consider the use of L1 an error.Lyster and Ranta (1997, p. 45) explained that "such uses of the L1 are not errors perse", but they were still interested in examining the instances where L1 was present.These types of errors were also included in the present study, since it is interesting to see how a teacher reacts to such instances.The example below presents a case of an unsolicited use of Persian during an English classroom event: Example 7: (Elementary) S: My mother is studying [farsi].(Error -L1 use) T: Persian Literature (Feedback -recast)

Types of Corrective Feedback
The six different types of feedback that Lyster and Ranta (1997) identified were used to categorize teachers' feedback in the current study.Definitions of the six types of feedback in this study precisely followed the ones in that study.The following explains each feedback type, along with examples from the actual data collected in the present study.

Explicit correction
This refers to the explicit provision of the correct form.As the teacher provides the correct form, she clearly indicates that what the student had said was incorrect (No, what you said was wrong, or Don'tsay…).On occasion, the wrong form is identified along with providing a correct form in the teacher's turn.
Example 8: (Elementary) S: Which area is a desert?(Error -lexical) T: Say how not which.(Feedback -explicit correction)

Recasts
These involve the teacher's reformulation of all or part of a student's utterance, minus the error.They are generally implicit in that they are not introduced by phrases such as You mean, and You should say.That is, the teacher would not indicate nor point out that the student made an error, but merely gives the correct form.
Example 9: (High Intermediate) S: She kept it as her pets.(Error grammatical) T: She kept it as her pet.(Feedback -recast)

Clarification requests
These are either in the form of question such as Pardon?and Sorry? or attempts to reveal the intended form of the error with the rising tone.This type of corrective feedback is used when there are linguistic problems in the learner's turn, and also when the learner's utterance is not comprehensible.Unlike explicit correction or recasts, clarification requests can refer to problems in comprehensibility.
Example 10: (Elementary) S: What kind of water source are there in the desert?(Error -grammatical) T: Sorry?! (Feedback -clarification request)

Metalinguistic feedback
This contains either comments, information, or questions related to the well-formedness of the student's utterance, without explicitly providing the correct form.It points to the nature of the error but attempts to elicit the information from the student.This kind of corrective feedback makes the learner analyze her utterance linguistically, not quite in a meaning-oriented manner.
Example 11: (High Intermediate) S: Both of them were firefighter.(Error -grammatical) T: Do we say both of them were firefighter?(Feedback -metalinguistic)

Elicitation
This refers to techniques that teachers use to directly elicit the correct form from the student.One technique is that teachers elicit completion of their own utterance by strategically pausing to allow the students to fill in the blank as it were.The other technique is that teachers use questions to elicit correct forms.Either way, teachers do not provide correct forms in their turn.

Repetition
This refers to the teacher's repetition, in isolation, of the student's erroneous utterance.In most cases, teachers adjust their intonation so as to highlight the error.
Example 13: (High Intermediate) S: Both of them were gas station partner.(Error -lexical) T: Gas station partner?(Feedback -repetition) In addition to the preceding six feedback types, there was a seventh category included in this study called multiple feedback, which referred to combinations of more than on type of feedback in one teacher turn.
Repetition clearly occurred with all other feedback types with the exception of recasts: in clarification requests ("What do you mean by X?"), in metalinguistic feedback ("No, not X.We don't say X in English."), in elicitation ("How do we say X in English?"), and in explicit correction ("We don't say X in English; we say Y.") (Lyster & Ranta, 1997, p. 48).Since repetition was common to these combined feedback moves, it was the clarification request, metalinguistic feedback, elicitation, and explicit correction that distinguished them, not the repetition.These thus were coded as instances of clarification request, metalinguistic feedback, elicitation, and explicit correction, respectively.Feedback coded as repetition, then, involves the teacher's repetition, in isolation, of the student's error.
Another combination that occurred was recast and metalinguistic feedback.It soon became evident, however, that such a combination was not "multiple" and necessitated instead the creation of the category "explicit correction."That is, as soon as the teacher's provision of the correct form is somehow framed metalinguistically, then the characteristics of a recast, along with its condition of implicitness, no longer apply.

Example 14: (High Intermediate)
S: There we can find narrow snakes.(Error -lexical) T: Thin.We don't say narrow.(Feedback -explicit correction) Finally, there were a few instances of elicitation occurring with metalinguistic feedback, as in the following example.This was coded as "elicitation" in order to consistently follow the coding instances where correct forms were elicited not clearly provided: Example 15: (Elementary) S: They didn't see each other after a long time.(Error -grammatical) T: …after?They didn't see each other… (Feedback -elicitation)

Types of Uptake
In this study, uptake is 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 utterance" (Lyster & Ranta, 1997, p. 49).Lyster and Ranta categorized uptake moves into repair when the uptake move resulted in repair of an error, and needs-repair when an error was not repaired in the uptake move.
In this study repair refers to the correct reformulation of an error as uttered in a single student turn and not to the sequence of turns resulting in the correct reformulation; nor does it refer to self-initiated repair.Only those repairs occurring after prompting, other-initiated repair rather than unprompted self-initiated corrections were analyzed.Lyster and Ranta distinguished four kinds of repair in their study as repetition, self-repair, peer-repair, and incorporation.Some examples of the first three kinds of repair follow.

Repetition
A student repeats the correct form given in the teacher's feedback when the feedback includes the correct form.

Self-repair
This refers to a self-correction, produced by the student who made the initial error, in response to the teacher's feedback when the latter does not already provide the correct form.

Peer-repair
This refers to peer-correction provided by a student or the whole class, other than the one who made the initial error, in response to the teacher's feedback.The nature of this uptake type is the same as self-repair.In addition to the above three kinds of repair, there was another kind of repair called incorporation, which referred to "a student's repetition of the correct form provided by the teacher, which is then incorporated into a longer utterance produced by the student" (ibid., p. 50).However, the instance of incorporation was not observed in the data for the current study, so the category is omitted here.
The other type of uptake is needs-repair, which refers to a situation where the learner responds to the corrective feedback but the learner's utterance does not result in repairing the original erroneous utterance.In Lyster and Ranta (1997), there were six types of needs-repair identified in their data: acknowledgement, same error, different error, off-target, hesitation, and partial repair.The current study adopts their categorization of needs-repair, except for acknowledgement, off target, and hesitation that were not observed in the data.
Acknowledgement referred to the cases where the positively recognizes teacher's feedback, generally saying yes or yeah, as if to say, Yes, that is what I meant to say.In off target, the learner responds to teacher feedback, but not to the targeted form in the feedback, and hesitation referred to the student's hesitation in response to the teacher's feedback.

Same error
The learner gives uptake upon receiving feedback, but repeats the same error in her turn.

Results
Before describing the results, it needs to be clarified that the total number of turns made by students or teachers was not counted.It was considered that the total number turns had little relevance to the target of investigation in this study, for its overall purpose was, as the research questions stated, to observe the kinds of errors, corrective feedback and uptake and to display the distribution of occurrence thereof.
The analysis yielded a total of 310 error sequences, each initiated by a student turn containing at least one error coded as grammatical, lexical, phonological or L1 use.Of these erroneous turns, 279 (90%) were followed by a teacher turn containing the corrective feedback coded as explicit correction, recast, clarification request, metalinguistic feedback, elicitation or repetition.The remaining 31 (10%) student turns with errors were immediately followed by topic-continuation moves.Table1 shows the distribution of error types found in the entire database.
<Insert Table 1 here> Although the difference in the commission of errors was not significant between the two levels (p>0.05), the patterns of error occurrences seem worthy to be analyzed in this part.There were 193 erroneous student utterances among the elementary, 117 among the high-intermediates and 310 in total.Of the 310 instances, 50.5% (N=156) were grammatical errors, 26% (N=81) phonological errors, 22% (N=69) lexical errors, and only 1.5% (N=4) unsolicited uses of L1.For elementary students, grammatical (53.5%) and lexical (25.5%) errors were the most frequent types, whereas grammatical (45.5%) and phonological (36%) were the case with high intermediate learners.The majority of the errors, then, were grammatical errors that included ill-formed sentences, phrases, or words which did not follow or apply the native speaker norm.
Table 2 Displays the distribution of different corrective feedback types in the database.
<Insert Table 2 here> The most frequently used feedback type (in the entire data) was recast with 50.5% (N=141) of the total 279 feedback turns, followed by elicitation with 16.5% (N=46), metalinguistic feedback 10% (N=28), clarification request 9% (N=25), repetition 8% (N=23), and explicit correction 6% (N=16).There was a significant difference (p<0.05) in the use of different feedback moves between the selected levels.In response to the errors committed by the elementary students, of 178 feedback moves, recast with 54%, elicitation with 17%, and clarification request with 10.5% were the most preferred types of reaction.This is while for the high intermediate learners, of 101 feedback turns recast with 44.5%, repetition with 17%, elicitation with 15.5%, and metalinguistic feedback with 11% were the most frequently used feedback types.It is notable that the negotiation of form feedback types that help the learner to self-correct (elicitation, metalinguistic feedback, clarification request, and repetition) amounted to 43.5% on the whole, 40.5% for the elementary and 49.5% for the high intermediate learners; this is while the feedback types that do not give opportunities for learner repair (recast and explicit correction) added up to 56.5%, 59.5%, and 50.5%, respectively.
Table 3 reveals the rate at which each error type received corrective feedback: it comprises of 94% of lexical errors, 93.5% of grammatical errors, and only 75.5% of phonological errors.
<Insert Table 3 here> A comparison of the distribution of feedback types across different error types appears in Table 4.
<Insert Table 4 here> As portrayed in the figure, the majority of feedback moves following all the error types were recasts (grammatical 41%, lexical 38.5%, phonological 85.5%, and L1 use 75%).Yet, phonological, L1 use (recall the small number of unsolicited uses of L1 receiving feedback), and grammatical errors seemed to provoke recasts more than lexical errors.Moreover, for grammatical errors elicitation (23.5%) and for lexical errors metalinguistic feedback (18.5%) were the next frequently used feedback types.
Table 5 displays the distribution of uptake across levels.Here a question may trigger the mind, that is, whether all repairs are equally effective indicators that students have noticed the feedback.It seems likely that the goal of the teachers should be for the learners to self-correct or to have another student (or students) correct the error.These types of repair are important in language learning since they indicate active engagement in the learning process on the part of the students; this active engagement occurs when the students have to think about and respond to the teachers' feedback based on the cues provided by them to help how to reformulate their incorrect language.A repair in which the student simply repeats what the teacher has said does not necessarily imply that the feedback has been understood as such (Lyster & Ranta, 1997, p. 54).Hence, in a new breakdown, peer-and self-repair were separated from repetition.Following Lyster and Ranta (1997), henceforth student-generated repair -referring to peer-and self-repair, repetition, and needs repair constitute the main categories of Uptake.
<Insert Table 5 here> As it is clear, high intermediate learners experienced higher rate of student-generated repair (38.5%), whereas the elementary students tried to repeat the teachers' utterances more (59.5%).Table 6 shows an account of the distribution of uptake (N=279) across feedback moves.
<Insert Table 6 here> Of all the feedback moves (N=279), only 31.5% (N=88) led to student-generated repair, while 56.5% (N=157) of them resulted in mere repetition, and 12% (N=34) to needs repair.The least productive corrective feedback moves to generate student-generated repair were recasts and explicit corrections.Of the total of 141 recast moves and 16 explicit correction moves, all of them led to repetition because generally they provide the students with the correct forms.
Moreover, the most effective corrective feedback moves to contribute to learner uptake were elicitation, metalinguistic feedback, repetition, and clarification request.Of 23 repetition moves found in the data, 96% (N=22) resulted in student-generated repair, and 4% (N=1) led to needs repair.Similarly, metalinguistic feedback received significant signs of learner repair with 86% (N=24) of all the 28 instances, and 14% (N=4) resulting in needs repair.Elicitation with 67.5% (31cases of 46) and clarification request with 44% (11cases of 25) were the next feedback types with rather high student-generated repair rates.

Discussion
The purpose of this study was to investigate the occurrences of different error types across two different proficiency levels and various ways of providing corrective feedback on these errors in Iranian EFL classrooms.Additionally, the focus was also on learner uptake and its connection to the corrective feedback moves.The findings of this study permit the following responses to the research questions under investigation: 1) What are the different types of error and corrective Feedback in communicatively oriented classrooms and how does this pattern vary from one level to the other?The findings revealed that four types of error occurred in these classes: grammatical, lexical, phonological, and unsolicited uses of L1.Grammatical errors formed the majority of the errors either in the entire database (50.5%) or per each level (53.5% for elementary and 45.5% for high intermediate students).For the high intermediate learners, phonological errors (36%) and for the elementary ones lexical errors (25.5%) were the next frequent categories.
2) What type of learner errors lead to what types of corrective feedback?L1 use, grammatical, and lexical errors received feedback more frequently than phonological errors.All types of errors tended to highly invite recasts; yet, lexical errors were the least likely category in this regard.The other frequently used feedback moves were elicitation (23.5%) with grammatical and metalinguistic feedback (18.5%) with lexical erroneous utterances.
3) What types of corrective feedback moves lead to learner repair?Of all the feedback moves, 56.5% of them resulted in mere repetition (recast and explicit correction), 31.5% to student-generated repair, and 12% to needs repair.Repetition (96%), metalinguistic feedback (86%), elicitation (67.5%), and clarification request (44%) were the most effective corrective feedback moves to contribute to learner uptake.
In the present analysis, teachers provided feedback on 90% of the student turns with errors (this varied from 86% to 92%for the high intermediate and elementary levels); these figures suggest a high ratio of correction versus non-correction in Iranian EFL interactive contexts.The results suggest that when the teachers provide feedback, they might want to consider the whole range of techniques they have at their disposal rather than relying so extensively on recasts, which comprised over 50% of all feedback moves.In this way, the teachers would provide more opportunities for student-generated repair: the results showed that recasts did not lead to any student-generated repair.
The data indicated that the teachers showed high preference in the use of recast (50.5%) in the entire database.Indeed, they use recast so often, so as not to harm interaction.This feedback type least risks interaction because it does not ask anything of the student, but rather, it provides the students with more input.In the case of recasts, classroom learners, whether they are given the opportunity to repeat or not, are less actively engaged insofar as there is little evidence that they can actually notice the gap between their initial use of non-target forms and the teacher's reformulation.As the results showed, recast did not lead to any sort of student-generated repair; rather, it ended in mere repetition (56% of time).Student-generated repairs may be important in L2 learning for at least two reasons.First, they allow opportunities for the retrieval of target language knowledge that already exists in some form.Second, when students self-repair themselves, they "draw on their own resources and thus actively confront errors in ways that may lead to revisions of their hypotheses about the target language (Lyster & Ranta, 1997, p. 57).It is likely the case that teachers are reluctant to encourage student-generated repair more consistently so as not to jeopardize the flow of communication.However, the data analysis revealed that none of the feedback types stopped the flow of classroom interaction and uptake clearly does not break the communicative flow; on the contrary, uptake means that the student has the floor again.Hence, there is a need for the teachers to take an account of the different types of feedback moves and try to select the most appropriate ones in each situation rather than sticking to one type.In this case, the data indicated that when there is negotiation of form, that is, when the correct form is not provided to the students -as it is in recasts and explicit correction-and when signals are provided to the learners that assist in the reformulation of the erroneous utterances, the feedback-uptake sequence engages students more actively and results to maximum studentgenerated repair.In relation to the total number of feedback moves in this database, such combinations (repetition, metalinguistic feedback, elicitation, and clarification request) occurred about 44% of the time in these classrooms.
Moreover, the data indicated that there is a need for teachers to carefully take into account their students' level of L2 proficiency when making decisions about feedback.Although recast was the most frequently used type of feedback in the database, the teachers used them with the elementary students more often than their high intermediate counterparts.The learners in lower levels are not that much experienced in the use of language and this may result in their incapability to self-correct.This might lead to a greater use of recasts.Interestingly, the negotiation of form feedback types (except repetition) was also more frequently used among the elementary learners.This finding suggests that despite the elementary students' lack of skills, the teachers tried to encourage them to self-repair.The findings in this regard were representative of the teachers' inappropriate selection; the use of negotiation of form feedback types did not raise the rate of self-generated repair among the low level learners, rather resulted in a higher rate of needs repair category (13%).Since more advanced learners have more language knowledge and are able to self-correct, the teachers use less recasts to keep the communication flow going (emphasis on spoken communication skills).In terms of the higher use of repetition, this sort of feedback seems to be more demanding, since the students should carefully regard the intonation of the teacher or recognize the teacher's utterance either as a mere repetition that is representing the correctness or signaling the inappropriateness.Hence, due to higher language proficiency, this type of feedback was much more frequently used with more advanced learners and at the same time resulted in more success in deriving student-generated repair.The higher rate of student-generated repair among the high intermediate learners (38.5%) might be attributed to this issue.
Teachers demonstrated a low tolerance for L1 uses, grammatical, and lexical errors, and yet a preference to react to L1, grammatical and phonological errors by using recasts.They may have done so because recasts of lexical errors, risk being perceived by learners as alternative yet equally correct forms.Similar findings have been noted in L1 contexts.For example, Marcus (1993) argued that recasts do not indicate whether the corrective reformulation is simply a stylistic variant or synonym, or whether the learner's utterance is unacceptable.Because L2 teachers frequently provide learners with synonyms as they strive to provide rich and varied input, recasts of some lexical errors risk being ambiguously perceived by students as acceptable alternatives (Lyster, 2001, p. 24).Due to this potential ambiguity, the teachers may have tended to provide signals to prompt students rather than provide them with correct forms.Through metalinguistic feedback, they tried to help students correct themselves.They used this kind of feedback when the error was recurring, to help the student remember and understand the correct form, when the student was emotionally and linguistically ready to be more demanded upon, and when the teacher wanted the student to think and reflect about the error and the language being used.
In contrast, phonological errors were repaired at a lower rate than the other error types and tended to recur within the same lesson in spite of corrective feedback.This may be due to the complex differences between the phonological systems of the two languages (especially in the case of clusters).This may in turn explain why the teachers demonstrated a clear preference for recasting these errors.Notwithstanding the teachers' preference for recasting phonological errors, none led to student-generated repair.
The classrooms analyzed here were not considered to be traditional classrooms and their discourse was structured in ways that allowed teachers to intervene regularly; they were able to do so by interacting with students without causing frustration because students appeared to expect such interventions.Thus, it appears that corrective feedback and learner uptake constitute an adjacency pair that is clearly anticipated in classroom discourse without stopping the flow of communication.Our data indicated that the feedback-uptake sequence is significantly more effective when signals are provided to the learner, which assist in the reformulation of the erroneous utterance, as it is the case of negotiation of form feedback types.
As was mentioned before, teachers do not make full use of the different corrective feedback techniques available to them (Allwright & Bailey, 1991).Furthermore, since the feedback situation is often over in a matter of seconds or minutes, and students might be unaware of the feedback they are receiving, the teachers should discuss the issues related to corrective feedback and the learning process with their students (Ancker, 2000).
As for teacher training programs, implications can be drawn from the present data.Teachers should be guided towards self-awareness from an early stage onwards.They should be taught to examine their own teaching from time to time.It is always beneficial to take a look at one's own actions and perhaps adjust the teaching techniques to benefit the learning process of the students.Moreover, the teachers should notice that one size does not fit all.Individual learners may well differ in terms of the particular error correction technique most appropriate for their unique language development needs.The teachers, hence, should try to learn and use different types of feedback that seem to produce student-generated repairs.The teachers should consider the point that their students may be more capable than they think.If they allow time and provide appropriate cues for them to self-repair, more often than not the students will come through.The least effective technique for correcting a student's incorrect language use is to simply give them the answer and the simplest way is not always the best.
In spite of the attempts made in providing a comprehensive account of error types, corrective feedback moves and uptake across communicatively oriented EFL classes, this study could have been wider and more extensive dealing with different proficiency levels.Moreover, the actual sequences were transcribed and interpreted by only one author.Hence, conducting a more extensive study and examining the usefulness of recasts in greater detail (to see if teachers should be trained to avoid them, or continue them as before) will be subject to further research.White, L. (1989).Universal Grammar and second language acquisition.Amsterdam:Benjamins.

Example2
each ring symbol?(Error -?) T: What does each ring symbolize.(Feedback -recast) Moreover, the content errors such as the following were not included in the study: Example 3: (High Intermediate) S: The twins lived in similar environments.(Error -content)T: They lived in different environments.(Feedback-recast) Example 4: (High Intermediate) S: I have an exam in Sunday.(Error-grammatical) T: On Sunday.(Feedback -recast) Lexical errors included inaccurate use of nouns, verbs, adverbs, and adjectives -in the sense of open classes.The following instance represents a lexical error in the selection of an adjective to show the length of time.Example 5: (Elementary) S: They found each other after a small time.(Error -lexical) T: After a … (Feedback -elicitation) Example 6: (High Intermediate) S: He was born before [ChrIst].(Error-phonological) T: [ChraIst] (Feedback -recast) Example 16: (High Intermediate) S: Then they can separated it.(Error -grammatical) T: … separate it.(Feedback -recast) S: … separate it.(Repair -repetition) Example 18: (High Intermediate) S: They found each other after 26 years old.(Error -grammatical) T: Do we say …after 26 years old?(Feedback -metalinguistic) Ss: No, after 26 years.(Repair -peer repair) not correct nor repeat the error after the feedback, and makes a different error.Example 20: (Elementary) S: There are drifts of sands.(Error -grammatical) T: Sorry.Repeat.(Feedback -clarification request) S: There are drifts of sand.(Needs repair -different error) 5.3.6 Partial repair This refers to uptake that includes a correction of only part of the initial error.Example 21: (Elementary) S: He agree with me to go to train.(Error -grammatical) T: He … (Feedback -elicitation) S: He agrees with me to go to train.(Needs repair -partial repair)

Table 1 .
Number and percentage of Errors by Error Type (N= 310)

Table 2 .
Number and percentage of Corrective Feedback (CF) types (N= 279)

Table 3 .
Rate of Feedback of Error type

Table 4 .
Total distribution of Corrective Feedback (CF) types across Error types

Table 5 .
Distribution of Uptake across levels

Table 6 .
Distribution of Uptake across Corrective Feedback (CF) types