EFL Students ’ Preferences for Oral Corrective Feedback in Speaking Instruction

This study investigates Indonesian EFL learners’ corrective feedback preferences including the timing, types of error, strategies of corrective feedback, and providers of error correction; and the relationship between foreign language anxiety and preferences for corrective feedback among students. Two hundred fifty seven EFL English department undergraduate students from two different course grades participated in the survey. The data were collected through questionnaire as the main data and interview as the supplementary data. The students’ were assigned to either a low anxiety group or a high anxiety group. The results showed that both sophomore students and freshman students agreed that student errors should be treated; freshman students and sophomore students had significantly similar opinions about perception, types, strategies, and providers of error correction.

The studies of error correction in language learning process interest researchers since error analysis became favoured area for studying second language and foreign language learning.At present time the contrastive analysis is shifted to the error analysis which means the focus of the study changes from the relationship between the native speaker and the target language to inspecting the actual language learner (Gass & Selinker, 2008).As the result, this learner language or interlanguage, which the learner builds from environmental data, has its own rules and conventions.Errors the learners make in their interlanguage provide researchers and teachers new evidence of their knowledge of the target language.In second or foreign language learning classrooms, teacher usually provides feedback to correct students' errors, this kind of feedback is known as corrective feedback.According to Ellis (2009), corrective feedback refers to teacher's response to students' utterances containing linguistic error.Many studies have investigated the relative effects of implementing various feedback types and strategies and have suggested that providing learners with a variety of corrective feedback could help them acquire correct forms.Apart from the positive effect offered by corrective feedback, there are several factors influencing the use of corrective feedback in the classroom activities.One of them is language anxiety.Foreign language anxiety might affect language acquisition because anxiety can obstruct learners' ability to process input and to form the target language.After all, the connection between instructor-learner interactions and language anxiety had been acknowledged by many researchers (Oxford, 1999, p. 65).For example, Young (1991, p. 427) states that classroom pro-cedures and instructor-learner interactions have been identified as potential sources of language anxiety, which is why teachers should be conscious of the possible implications of their teaching methods and strategies.Indeed, learners cannot concentrate on the learning task at hand if they feel stressed and insecure (Ellis, 1994, p. 479).
In order to gain insight about the interrelationship between errors, corrective feedback and anxiety, the research problems in this study are formulated as follows.(1) What are the preferences of the students with different anxiety levels on the appropriate time of correction?(2) What are the preferences of the students with different anxiety levels on the type of error that need to be corrected?(3) What are the preferences of the students with different anxiety levels on the choice of corrective feedback strategies?(4) What are the preferences of the students with different anxiety levels on the provider of error correction?METHOD Survey research using cross-sectional design is considered to be an appropriate approach for this study because the researcher intends to examine current attitudes, beliefs, opinions, or practices.It is used to quantitatively describe and compare the students' preferences for corrective feedback in speaking instruction across students' foreign language anxiety levels.This study was conducted in English Department of State University of Malang.It was conducted in speaking courses at undergraduate level.There were four speaking course grades offered in English Department, namely Basic Speaking (Integrated during Intensive Course) offered to first semester students, Speaking for Informal Interactions (Speaking I) offered to second semester students, Speaking for Formal Interactions (Speaking II) offered to third semester students, and Speaking for Academic Purposes (Speaking III) offered to fourth semester students.Since this study compared the students' preferences across different levels of foreign anxiety, two speaking course grades were involved (Intensive Course and Speaking II) because these were the courses that were offered during the time of data collection of this study.The subjects to be surveyed, or the research sample, should be selected from the population of interest.Sample size in this study was determined based on Slovin (1960( , as cited in Seville et al., 2007: 182) : 182) using the following formulae as follows.
n is the sample size, N is the population size, e is the significance level (.05 or 5% in the current study).Drawing from this procedure, 130 subjects from the total 193 freshman students (Intensive Course) and 127 subjects from the total 187 sophomore students (Speaking II) were decided as the appropriate sample size for the present study.The researcher distributed two kinds of questionnaire; foreign language anxiety questionnaire and corrective feedback questionnaire as the main instrument in this study.A semi-structured interview was also used in the present study as a follow up to the questionnaire result.
The corrective feedback questionnaires was adapted from Fukuda (cited in Park, 2010b), Agudo (2013), and Katayama (2013) to gather information about their preferences for corrective feedback in speaking instruction.The language anxiety questionnaire was used to elicit language learners' self-reports of anxiety over oral English performance in EFL classroom.The questionnaire was adapted from Horwitz et al. (1986).FLCAS by Horwitz et al has been widely used all over the world to measure the level of foreign language anxiety.The interview guide is used to follow up the questionnaire in order to get deeper understanding of the phenomenon under the study and to explore further specific data, and to confirm or disconfirm the temporary findings revealed from the survey.

English Department EFL Learner's Levels of Foreign Language Anxiety in Speaking Course
The researcher first began to obtain quantitative data by administering foreign language anxiety scale (FLCAS) questionnaires to 130 first semester students of English Department and 128 third semester students of English Department The mean of foreign language anxiety scores was calculated in order to investigate the relationship between foreign language anxiety and preferences for error correction.Based on the students' responses to the language anxiety questionnaire, they were classified as having low or high anxiety by using the total mean score for the whole sample.In order to even out the distribution of the participants, learners who scored between 83-132 points in FLCAS were classified as "high anxiety," and learners who scored between 33-82 points in FLCAS were classified as "low anxiety."As a result for the freshman, 77 first semester students were assigned to a high anxiety group, and 53 first semester students were assigned to a low anxiety group.The whole samples (N) for first semester students (freshman) were 130 students.Moreover, the result for sophomore, 39 third semester students were assigned to a high anxiety group, and 89 third semester students were assigned to a low anxiety group.The whole samples (N) for third semester students (sophomore) were 128.Table 1 presents the distribution of the students according to their levels of anxiety.
The students' willingness to receive corrective feedback might be attributable to the teacher's teaching practice.Data obtained from the interview affirmed that all interviewees are used to receive corrective feedback when they produced incorrect utterances.As the result, they have been accustomed with the feedback.In addition, Allwight and Bailey (1991, p. 102) state that "non-native speaking teachers of the target language are consistently more severe in their reaction to le arners' errors than their native-speaking counterparts".In line with this view, Chaudron (1988, p. 137) reported that the instructional setting might influence the amount of corrective feedback.He summarized studies which were conducted in EFL and ESL classrooms in which the percentage of errors ignored in EFL classroom was between 10% and 15%, whereas in ESL classroom was between 42% and 49%.In other words, corrective feedback as a response to errors tends to occur more frequently in EFL classrooms.
In spite of the favourable attitude to receive corrective feedback, the results of inferential statistics shows that there is a significant difference in the mean rank of the two groups (comparison between freshman and sophomore students with low anxiety level and comparison between freshman and sophomore students with high anxiety level) in which the mean rank of freshman students is higher than the sophomore students.It suggests that the freshman students were more welcome to receive correction.This phenomenon might be attributed by students' different level of exposure.Sophomore students in this case have a better level of exposure in which they had gone through several Speaking classes (i.e.Speaking I class, Speaking II class).It could possibly deduce that as level goes up, the students will be more concerned to the accuracy of the utterances.As the result sophomore students will have more screening to the incoming feedback although they did want to receive feedback.
Dealing with the potential affective damage caused by corrective feedback, Truscott (1999) argues that corrective feedback has negative effects on learning because it could make the students feel embarrassed, annoyed, and inferior.The data revealed that several freshman students and sophomore students agreed to the statement that they feel embarrassed when they are being orally corrected.It suggests that the students also feel such kind of feeling.Such feeling possibly arises because they are corrected.However, feeling embarrassed does not directly mean that the students do not want to be corrected.To sum up the results of foreign language anxiety scale (FLCAS) questionnaire, it can be concluded that both freshman and sophomore EFL students are dominated by the students who have low anxiety level in their speaking course, although the number of students who experience high anxiety are quite big too for both groups.

Perception of Corrective Feedback in Speaking Instruction
The discussion about perception of corrective feedback is divided into three categories, namely student willingness to receive corrective feedback particularly feedback delivered orally, the role of corrective feedback, and the students' feeling of being orally corrected.The trend of data in this study suggests that the students' willingness to receive corrective feedback and the benefit of the feedback in learning process appears to be undeniable.The majority of the freshman students and sophomore students from both who has high level of anxiety and low level of anxiety have favourable attitude toward corrective feedback, particularly one which is delivered orally.None students did not want to receive oral corrective feedback.It is consistent with the findings found by Katayama (2007) and Park (2010b).
In response to the facilitate role of corrective feedback, the majority of the students surveyed in the present study, both freshman students and sophomore students regardless their level of anxieties, agreed that they have learnt a lot by being orally corrected and the correction is beneficial for them.The most cited reason during the interview is that the students need corrective feedback to help them to notice their errors.It indicates that the provision of corrective feedback could boost the students' awareness toward the errors.It is related to the idea of noticing a gap in which corrective feedback could assist the students to notice the discrepancy between their interlanguage and the target language (Kim, 2004, Li, 2010, Sheen and Ellis, 2011).Furthermore the interviewees asserted that when there is no corrective feedback, they assume that they have produced correct utterance.This phenomenon might lead to fossilization (i.e.cessation in developing the students' interlanguage system).
On the contrary, Edge (cited in Tomkova, 2013, p. 66) argues that the provision of corrective feedback could assist the students to be more accurate in the L2.All of the students in the interview also agreed that corrective feedback helps improving the accuracy of their utterances.Therefore, it is expected that the provision of corrective feedback could modify the students' interlanguage system.
The facilitative role of corrective feedback in learning is confirmed by Chu (2011) and Golshan (2013) in their experimental study comparing students who were given feedback and no feedback.The results consistently showed that the students who received corrective feedback outperformed those who did not receive corrective feedback.
Thus, it could be implied that corrective feedback is required by the students to develop their interlanguage due to the benefits offered by corrective feedback.

Types of Errors to be Corrected in Speaking Instruction
The respondents of this study were also required to decide which errors should be corrected.The errors are divided into two categories.The first one refers to errors from the point of communication, namely global error and local error.The second one refers to errors in the domain of grammar, phonology, and lexis.
Many researchers oppose comprehensive feedback.Martinez (2006, p. 3) states that excessive corrective feedback could reduce the students' motivation to learn and discourage them from participating in the classroom because students will not say anything unless they are sure that they have correct utterance to produce.Therefore, teacher as the most common source of feedback is suggested to be selective in choosing which error to be corrected.It is also not feasible to correct every error that the students made because of considering the time allocation.Tomkova (2013, p. 78) argues that teacher could base type of errors to correct on the seriousness of errors from the point of communication.In this case, if the errors affect the comprehensibility of the utterance or what is called as global errors by Ellis (1997, p. 20), it is considered as serious and should be corrected, whereas errors that do not affect the listener's understanding (i.e.local errors) do not need much treatment.
However, the result of the present study revealed that students, regardless of their different speaking course grades and their level of anxieties, preferred teacher correct all errors although the errors do not affect meaning of the message.Nearly over half of the freshman students and sophomore students from both level of anxieties disapproved the statement that teacher should correct only the errors that affect meaning of the utterance.It signifies that students always want to be corrected.This finding is in line with a study conducted by Tomczyk ( 2013), but in contrast with studies conducted by Katayama (2007) and Abukhadrah (2012).In their study, they found that students prefer focuses only on errors that influence the listener's understanding toward the message conveyed.
The preference of correcting all errors might be related to the essence of corrective feedback which could boost the students' language awareness.As stated by Kim, 2004;Li, 2010;Sheen and Ellis, 2011, corrective feedback assists the students to notice the discrepancy between their interlanguage system and the target language.Thus, by correcting all errors, the students could be more aware of the errors in their utterances.
Even though students wish for a comprehensive feedback, certain domains are expected to receive feedback more frequently than the others do.Grammar and phonology are identified as type of linguistic errors which should always be corrected.Grammatical errors were chosen by 53.02% of the freshman students with low anxiety level and 49.01% of the freshman students with high anxiety level, surveyed as errors which should always be corrected.For the sophomore students, there was a clear tendency (66.02%) to choose phonological errors to always be corrected.
Then, it becomes obvious that these types of error are considered more essential.The preference for always correcting grammar and pronunciation might have to do with the fact that they are more problematic.As asserted by Lyster (2001, p. 289) "the cognitive process involved in accessing and applying the system driven rules of grammar are more complex than those involved in the retrieval of lexical items".Besides, it might also be attributable to the instructional practices.Swain (cited in Abukhadrah, 2012, p. 154) asserts that most EFL classes focus their attention on grammar.As for the preference for always correcting phonological errors might be attributable by the fact that in phonology, the students require model to imitate.Lexical errors were chosen by both the freshman students and sophomore students regardless their level of anxieties as the third priority to always to be corrected.This is possibly because the students could learn it by themselves through dictionary or other sources.
In the nutshell, the data imply that whether the errors affect comprehensibility of the utterance or not, they should always be corrected, especially when dealing with grammatical and phonological errors.In other words the students prefer constant correction to a selective one.

Timing of Correction in Speaking Instruction
In the present study, the timing of correction is divided into immediately feedback which is provided as soon as the errors are made although it interrupts the flow of communication, and delayed feedback which is provided after the students finishes with the message they want to convey or before the teacher ends the class.
The finding shows that there is a clear tendency for freshman students and sophomore students to prefer delayed feedback which is given after they finish speaking.Thus, the findings of the present study corroborate with the findings found by Park (2010b) and Tomczyk (2013).
It is known that immediate feedback interrupts the flow of a communication, as the result the students might perceived it as unfavourable one.Allwight and Bailey (1991, p. 103) point out that "the problem with immediate error treatment, many teachers feel, is that it often involves interrupting the learner in mid-sentence-a-practice which can certainly be disruptive and could eventually inhibit the learners' willingness to speak in class at all".Majority of the students in the interview also thought that immediate feedback could bother their concentration resulted they forgot what to say.However, those who belong to interactionist believe that corrective feedback works best when it is assigned in context at the time the students makes the error (Ellis, 2009, p. 5).In so doing, the feedback is more salient.
Delayed feedback on the contrary might be favoured by the students because it does not disturb the flow of communication.It provides chance for the students to finish what they want to convey.
Surprisingly, based on the interview result, some interviewees who agreed for being interrupted state that correcting errors as soon as possible is considered helpful when it is in form of informal conversation, not like in middle of presentation, recitation, speech, etc.It can be concluded that whether to provide immediate feedback or delayed feedback might be influenced by the purpose of the activities.In line with this view, Ellis (2009, p. 11) argues that "There is general agreement that in accuracy-oriented activities correcting should be provided immediately".It is possible because accuracy aims at learning.In fluency-oriented activities where the emphasis is for communication, it is difficult to provide immediate correction because it is disruptive.Edge (cited in Tomkova, 2013, p. 30) states that "In order to bring about fluency students need to experience uninterrupted, meaningful communication if they are to learn to use the target language".
Since the majority of the students preferred feedback which is given after they finish speaking regardless their level of anxieties, it implies that the students do not like to be interrupted with corrective feedback when they are still speaking.

Provider of Correction in Speaking Instruction
Teacher usually regarded as the main source of feedback in the classroom interaction.As matter of fact, teacher could let the students do self-correction.If the self-correction does not work, another alternative is teacher asks other students to do correction so called peer-correction.
The data obtained in the present study reveal that most preferred corrector perceived by the freshman students and sophomore students in one which is delivered by the teacher.There were 77.9% of the low anxiety freshman students, 64.2% of the high anxiety freshman students, 58.4% of the low anxiety sopho-more students, and 64.1% of the high anxiety sophomore students who strongly agreed with feedback from teacher.Zacharias (2007) who has researched Indone-sian students' attitude toward teacher correction found that there were several reasons why Indonesian stu-dents' tend to rely more on the teacher correction.Firstly, the students' assumption that teacher is more competent which makes his/her feedback is more valid and reliable.Secondly, the cultural belief that teacher is the source of knowledge which makes him/her is always right.In a nutshell, what causes Indonesian students' preference for teacher correction is the dif-ferent level between the teacher and the students.Teacher is regarded as more proficient in terms of the language and knowledge.
Moreover, the students' preference for teacher correction might also be the result of teaching practices tradition in which the classroom is frequently more teacher-centered because teacher is regarded as an authoritative figure.
In spite of the clear tendency to prefer teacher correction, the students in the present study actually valued all sources of feedback.It is indicated by a high percentage of agreement for those three agents of correction.In addition, the interview data revealed a surprising finding in which although the students preferred to be corrected by teacher, they mostly disagree if the teacher is said to be the only source of feedback.
The findings of the present study also show that the preference for teacher-correction is followed by self-correction and peer-correction.It is similar with that found by Park (2010b).Mendez and Cruz (2012, p. 68) state that "self-correction seems to be preferred correction provided by others because it is face-saving".In self-correction, students show their effort to repair their errors.Nonetheless, the students' ability should be considered in deciding to engage the students in self-correction.Ellis (2009, p. 7) highlights that selfcorrection could work well if the students possess knowledge about the linguistic system.Hence, it could be successful as long as the students realizes that he/ she has made the errors, identify the errors, and know how to repair the errors.Self-correction also allows the students to engage in deeper mental processing which in turn, they might have a better result on learning (Loewen, 2007).In line with this view, de Bot (cited in Lyster and Panova, 2005, p. 592) claims that "language learners are likely to benefit more from being pushed to retrieve target language forms that from merely hearing the forms from the input, because retrieval and subsequent production stimulate the development of connections in memory".In short, providing chance for the students to do self-correction could promote their L2 learning.
Interview result showed that some of the interviewees in the present study claim that it is fine to be corrected by peers as long they are more superior.It point out that the students' preference for peer-correction could be influenced by who provide the correction.Philip, Walter, and Basturkmen (cited in Lyster et al, 2013, p. 28) point out that student may be doubtful of his/her friend's ability that makes them deliberately disregards their friend's feedback.Overall, data from the present study imply that there is a tendency for students to prefer to be corrected by someone who are more proficient than themselves.

Types of Corrective Feedback in Speaking Instruction
There are many types of corrective feedback which could be used by teacher to respond errors made by the students.However, teacher should be very careful in choosing those feedbacks.Young (1991, p. 429) states that "a harsh manner of correcting student errors is often cited as provoking foreign language anxiety".It means that that the most significant factor affecting the students' emotional state toward corrective feedback is the way the correction is provided.Lyster and Ranta (2007, p. 152) classify corrective feedback into two categories, namely reformulations and prompts.Reformulations supply the students with the correct form of the language covering explicit feedback and recast.Reformulations, therefore, could not lead to the students' repair.Meanwhile, prompts consist of elicitation, metalinguistic cue, clarification request, and repetition.They provide the students with cue that lead them to repair the errors.As the result, they often called as negotiation of form (Lyster, 2001, p. 273) which is found effective to lead to higher rate to repair (Lyster and Ranta, (2001).Repair is a situation where the student could correct their errors.Sheen and Ellis (2011, p. 594) add new type of feedback which is called as paralinguistic signal which is also belong to prompts.The types of corrective feedback that were used in the present study combined feedbacks proposed by those experts.
According to Lyster and Saito (2010a, p. 288) the types of error could significantly affect the choice of feedback.Therefore, in this present study, the pref-erences for the types of feedback were linked with the types of error.In addition, Lee (2008, p. 157) proposes that teacher's classroom practice is one of the contributing factors affecting the students' preference.Besides, it might be influenced by how the students perceive the effectiveness of the feedback (Abukhadrah, 2012, p. 155) in which the effectiveness if influenced by the quality if the feedback covering consistency, accuracy, and comprehensibility (Lee, 2008, p. 157).
The finding of this study reveals that the two groups of the students prefer to be corrected using explicit feedback of all types of error.This finding indicates that the students want overt correction of their error no matter what type of errors that they have made.
Explicit feedback as the most preferred type of corrective feedback involves obvious statement from the teacher that the student has produced incorrect utterance followed by the correct form of the utterance.Thus, the error and the correct form are overtly contrasted which makes it more comprehensible.The comprehensibility of explicit feedback might be plausible reason why the students' prefer this type of feedback.Hendrickson (cited in Lyster, 1998, p. 54) states that "the procedure whereby teachers provide students with correct forms is ineffective when helping students learn from their mistake".In spite of the comprehensibility offered by explicit feedback, it does not provide chance for students to repair their error because it already provides the correct form.
Another appealing finding from this study is that both of the groups surveyed rank paralinguistic feedback as the least favoured type of feedback to respond to all types of error.It might be attributed by the vagueness of the nonverbal language as gesture or facial expression.Even though paralinguistic signal provides chance for the students to correct their errors, it may only signal the error by using gesture or facial expression.Besides, this type of feedback also does not provide cue to help the students to self-correct.

Preferences for Corrective Feedback in Grammatical Errors
The low anxiety freshman students' preferences for corrective feedback to treat grammatical errors reveal the following order; explicit correction, metalinguistic, clarification request, elicitation, repetition, recast and paralinguistic signal; as for the high anxiety freshman students: explicit correction, metalinguistic, clarification request, repetition, elicitation, paralinguistic sig-nal, and recast.Whereas for the low anxiety sophomore students: explicit feedback is also followed by metalinguistic, elicitation, clarification request, repetition, recast, and paralinguistic signal; as for the high anxiety sophomore students: explicit feedback, metalinguistic, elicitation, repetition, clarification request, recast, and paralinguistic feedback.
The present study result is somehow similar with Katayama's study.In Katayama (2007), it is found that to correct grammatical errors, students prefer the teacher to provide metalinguistic that lead them to notice the errors and self-correct.Despite the fact that explicit feedback is most favourable feedback type by all students in the present study, metalinguistic which ranked as second most favourable type of feedback by all of the students from both groups is actually have slightly lower percentage from explicit feedback.So in treating their grammatical errors despite the comprehensibility offered by explicit feedback, the students ranked metalinguistic as the second favourable corrective feedback to notice and self-repair.
Based on inferential statistics calculation, a significant difference is found in the way the freshman students and sophomore students rate the effectiveness of repetition, elicitation and recast.The mean rank of repetition, elicitation, and recast from the sophomore students is higher than that found in freshman students In particular, the mean rank of repetition for freshman students is 80.09 for freshman students and 86.45 for sophomore students.For elicitation, the mean rank of freshman is 73.18 and the mean rank for sophomore is 92.43.Furthermore, freshman students obtained 75.64 of the mean rank and sophomore students obtained 90.3 for recast.It can be inferred that sophomore students more appreciate repetition, elicitation, and recast.
When providing feedback using repetition, the corrector asks the students to clarify their utterance by spelling the incorrect utterance.Repetition is more suitable and often be used for asking for explanation rather than as a corrective feedback.The finding indicates that the sophomore believed that repetition can allow them to think about their utterances once more, so that they can notice an error they made in their speaking.However, freshman students did not regard repetition as an effective feedback type to help them find the target-like forms to the same degree as sophomore.Repetition in correcting grammar can be confusing because it might not always be clear whether the teacher is repeating student's utterance to indicate the problem or to acknowledge the content.
In elicitation, teachers elicit completion of their own utterance by strategically pausing to allow students to "fill in the blank.Thus, these types of feedback allow the students to self-correct.They also overtly tell the students that there is something wrong with their utterance, yet, the corrector neither locates the error nor provides the correct utterance that ultimately could make the students have difficulty to self-correct.However, elicitation can help the students to develop selfeditorial skill by providing them with time to think about the target form (Lyster, 2001).
Another type of feedback attracting more attention because of its controversy is recast.On one side, Lyster (2001) and Mackey, Gass, and Mc Donough (2000) found that teachers tend to use recast to correct grammatical errors because grammar involves a series of complex rules which is difficult to be retrievable.Thus, it is better to directly supply the students with the correct grammatical form.However, on the other sides, Lyster and Saito (2010a, p. 289) argues that recast tends to be ambiguous for grammatical errors because it might be perceived as another way to express the same thing in order to confirm the comprehensibility of the utterance.The ambiguity perhaps caused by whether it concerns with form of meaning which consequently might not be perceived as feedback.It requires the students' high awareness to distinguish the mismatches between the incorrect and correct form of the target language.In fact, Lyster (2001) found that negotiation of form (i.e.metalinguistic cue, repetition, clarification request, and elicitation) could lead higher rate of grammatical repair.

Preferences for Corrective Feedback in Phonological Errors
To treat phonological errors, the low anxiety freshman students preferred teacher to use explicit feedback, followed by metalinguistic, repetition, clarification request, recast, elicitation, and paralinguistic signal.Whereas, the high anxiety freshman students preferred teacher to use: explicit feedback, followed by metalinguistic, repetition, clarification request, elicitation, paralinguistic signal, and recast.For the low anxiety sophomore students' explicit feedback was also their first choice, followed by metalinguistic, repetition, elicitation, clarification request, recast, and paralinguistic signal.While, the high anxiety sophomore students preferred teacher to use: explicit feedback, metalinguistic, repetition, elicitation, clarification request, recast, and paralinguistic signal.
It is interesting to note that repetition is mostly chosen by both groups of students as one of the effective types of corrective feedback to respond to phonological errors.Repetition refers to teacher's repetition, in isolation, of the student's erroneous utterance.In most cases, teachers adjust their intonation so as to highlight the error.However, the result of previous study shows that recast is the most frequently used to respond to phonological errors (Lyster, 2001).Since the correction of phonological errors does not require word change, it makes recast more salient.It is true that recast could provide a model for the students to imitate which is necessary in learning phonology but repetition allow them to think about their utterances once more, so that they can notice an error they made.
Furthermore, the inferential statistics calculation a significant difference is found in repetition and elicitation because the sig.values for the three types of corrective feedback are below .05,precisely, .013for repetition and .001for elicitation.The mean rank of repetition, and elicitation from the sophomore students is higher than that found in freshman students In particular, the mean rank of repetition for freshman students is 74.09 for freshman students and 91.64 for sophomore students.For elicitation, the mean rank of freshman is 71.17 and the mean rank for sophomore is 94.17.It can be inferred that sophomore students with low anxiety level more appreciate repetition and elicitation as the corrections for their phonological errors.

Preferences for Corrective Feedback in Lexical Errors
As regard to lexical errors, there is no difference order of preference the data suggest that the most preferred correction perceived by the freshman students with low and high anxiety level is explicit feedback, followed by metalinguistic, recast, elicitation, clarification request, repetition, and paralinguistic signals.While, for the most preferred correction perceived by the sophomore students with low and high anxiety level is explicit feedback, followed by metalinguistic, repetition, elicitation, clarification request, recast, and paralinguistic signals.
Explicit feedback and metalinguistic cue are rated very effective by both groups of students.This might be attributed by the explicit sense of these feedback types.Explicit feedback directly points out the errors and tells the correct forms, whereas metalinguistic cue provides the students with the information about how the word supposed to be used, for instance through the provision of word definition.It provides the students a deeper understanding about the usage of the word.Metalinguistic, therefore, assist the student to self-correct.
Overall, it could be implied that regardless their anxiety levels and course grades, the clearer the teacher points out the error, the more students like it, no matter whether it is used to respond grammatical, phonological, or lexical errors.

CONCLUSIONS
The purposes of the present study were to investigate: (1) the students' perception of corrective feedback in speaking instruction, including the timing of correction, types of error, strategies of corrective feedback, and providers of error correction; and (2) the relationship between anxiety and preferences for error correction.
The findings show that both the freshman students and sophomore students agreed that student's errors should be treated, particularly one which is delivered orally.In the matter of errors from the point of communication, both the freshman students and sophomore students agreed that teacher should correct all errors that students made in speaking.Thus, they also wanted error treatment even on infrequent and individual errors comprehensively.And for the error domain, both freshman students and sophomore students identified grammar and pronunciation as type of linguistic errors which should always be corrected.Respectively, grammar and pronunciation error are considered more essential for the students.The preference for always correcting grammar and pronunciation might have to do with the fact that they are more problematic rather than lexical error.A correspondence was also found in the timing of correction, the finding shows that there is a clear tendency for freshman students and sophomore students to prefer delayed feedback which is given after they finish speaking.Accordingly, the students regarded immediate error correction that can interrupt the flow of conversation as ineffective.From the providers of correction, teachers were the most popular source of feedback among self-correction and peer-correction, as the students have a tendency to prefer what they are used to.There are two possible explanations why did this happen: (1) the students' assumption that teacher feedback is more competent which makes his/her feedback is more valid and reliable, (2) the cultural belief that teacher is the source of knowledge which makes him/her is always right.Explicit feedback, followed by metalinguistic, and repetition were the most popular types of feedback among both freshman students and sophomore students.On the contrary, paralinguistic signal was the least popular type of feedback among both freshman students and sophomore students.The findings show that the freshman students and sophomore students had significantly similar opinions about perception, types, strategies, and providers of error correction.In contrast, a significant difference between the high and low anxiety groups between the freshman and sophomore students found in the orders of corrective feedback preferences for grammatical, phonological, and lexical errors.As far as the relationship between anxiety and corrective feedback is concerned, of all the corrective feedback strategies investigated in the present study, explicit correction and metalinguistic feedback create the least amount of anxiety in the learners, as they were the most popular strategies among the anxious students of the study.As argued by Lyster and Ranta (1997) explicit correction may not lead to learner uptake as often as some of the others strategies, which may encourage teachers to avoid using it.What should be kept in mind, however, is that the student is not likely to benefit from the teacher's corrective feedback, if it causes anxiety in him/her.Consequently, when choosing the suitable error correction strategies, teachers should always consider the student's individual affections and perceptions on corrective feedback.
Based on the discoveries of the study, there are some implications offered.First, future studies need to go beyond the simple identification of relationships between freshman and sophomore students or between the pairs of students' characteristics regarding preferences for error correction in order to overcome limitations of the present study and obtain more reliable results.Second, due to teachers' immediate error correction could decrease students' motivation to speak.Spoken errors should be treated after students finish speaking.Also, teachers should use various types of feedback to facilitate the effects of error correction and promote language learning.Third, although students want to receive error treatment as much as possible, constant corrective feedback from the teacher can discourage students from participating in activities in class and increase anxiety.Teachers, therefore, need to understand their students' various needs, concerns, and expectations toward error correction by using a variety of tools, such as questionnaires, inter-views, and observations to determine the students' needs.In so doing, teachers can promote students' learning.What should be kept in mind, however, language learning is a complex and gradual process and that everyone should not expect that a reaction in response to learner errors in the course lead to immediate effects on their learning.Improvement takes place over time and, for feedback to be maximally effective it should be provided regularly and consistently over period of time.In addition, learners must have opportunity not only for feedback, but also for processing and using the target forms in various form and meaning so that the forms can become part of their interlanguage system.
On the basis of the findings of the study, some recommendations are put forward.The recommendation will firstly be given to the learners, the most important element in their learning process, and then the teacher or the lecturer who can become the facilitator who will facilitate and guide the learners, and finally the recommendation is also given to the future researchers who intend to do research in the same field.

For The EFL Learners
The effectiveness of the feedback given by the teachers depends on the social and instructional context of the feedback itself.Learners have shown different respond to different types of feedback depending on the context in occurs.Thus, contextual differences could affect the language learners, as the result teacher has to be aware of this and adjust his/her feedback strategies in ways that suit best for the context.Learners are advised to perceive the corrective feedback given by their teachers although, sometimes, it is hard to always notice the corrective nature of teachers' feedback.Furthermore, there is also a perceived gap between learners' response and teacher's perceptions of those responses, resulted in a mismatch between both of them.One implication, it is very important for the learners to value teachers' feedback despite teachers' effort doing his/her best to be clear on what works for the student's and how.

For The Teacher or Lecturer
The next recommendation is directed to the teacher or lecturer of English.In helping learners to attain success through an effective way of learning, there are some tasks that can be accomplished by the teacher.Firstly, oral corrective feedback can be used in a classroom with one enabling condition, which is the appropriateness of corrective feedback with the students' ability.The present study found that each type of corrective feedback has different effect on the students.The corrective feedback which explicitly points out the errors and provides the correct form, such as recast and explicit correction, seems to work better.Meanwhile, the corrective feedback which prompts the students to do self-repair, such as elicitation, metalinguistic feedback, seems to be less favourable but it actually could lead them to the learning.One thing that should bear in mind is that teacher should lower the frequency of traditional spoon-fed and moves forward to challenging and creative teaching.Secondly, teacher tends to give feedback in a spontaneous manner, in other words giving the correction tends to be indistinctive.Actually, teacher should take into account learners' developmental readiness and provide feedback in such way that matches learners' level.With all due respect, it is not always easy for teacher to determine developmental readiness of the individual learners.One possibility is to be flexible and make use of the different types of feedback on different occasions.When teacher use a variety of strategies, they can address a wider group of the students with more varied linguistic abilities, feedback needs, and preferences.Another possibility is to use what Nassaji (2000) & Swain (2011a) have called negotiated and scaffolded feedback (feedback which occurs through negotiation and consist of feedback exchanges that involve multiple feedback moves).

For The Future Researcher
The research of oral corrective feedback for students' of different anxiety levels is suggested to be conducted in wider various contexts and settings.Thus, the similarity and dissimilarity of the result can be found.The research which investigates the whole aspect of oral corrective feedback in a longer period of time is also recommended to be conducted since it can give a holistic picture of the phenomenon of oral corrective feedback for students' with different backgrounds.In addition, a longitudinal study also can be conducted to discover the effect of oral corrective feedback for students' second language acquisition.