Preferences for Oral Corrective Feedback: Are Language Proficiency, First Language, Foreign Language Classroom Anxiety, and Enjoyment Involved?

Background. The effectiveness of oral corrective feedback (OCF) in language learning is influenced by learners’ comprehension and response to various OCF techniques. Therefore, it is essential for teachers to consider learners’ preferences for OCF strategies. Purpose. This quantitative study aimed to investigate the preferences of Thai as a foreign language (TFL) learners for ten commonly discussed types of OCF. Specifically, it examined whether these preferences are influenced by four learner variables: proficiency level, first language (L1), foreign language classroom anxiety (FLCA), and foreign language enjoyment (FLE). Method. The study involved 288 university students from Chinese, Japanese, and Korean TFL settings, and the data from questionnaires were analysed using appropriate statistical methods. Results. The findings indicate that, regardless of proficiency level, L1, FLCA, or FLE level, learners prefer more explicit OCF techniques, such as metalinguistics feedback and explicit correction. However, Korean undergraduates scored lower in the majority of OCF strategies (i.e., ignoring, elicitation, recast, explanation, and public feedback) compared to the other participants. The MANOVA analysis revealed significant differences in ignore, peer correction, recast, and private feedback based on proficiency level and L1 background. Although the differences between the FLE and FLCA approaches were not statistically significant, high FLE and FLCA groups tended to prefer more OCF strategies than the low groups. Conclusion. This study has significant implications for instructional practices in TFL settings and for L2 lecturers in the classroom. By understanding learners’ preferences for OCF, educators can tailor their instructional approaches to meet the specific needs of their students.


INTRODUCTION
Oral corrective feedback (OCF) refers to remarks made by teachers or interlocutors on errors made by second language (L2) learners during speech production.The use of OCF in the language class-room is a common technique for increasing students' awareness of the mistakes (Yang, 2016).Empirical and observational studies have proved the effectiveness of OCF in accelerating language learning (Fidan, 2015;Geckin, 2020;Gómez Argüelles et al., 2019;Gooch et al., 2016Ha & Nguyen, 2021;Lee, 2017, Li, 2018, 2021).

| Research Papers
Previous research has highlighted that the impact of OCF on language learning is linked to learners' understanding and reaction to different OCF techniques.It has been suggested that discrepancies between teachers' and learners' interpretations of OCF could have negative impacts on language learning (Kartchava & Ammar, 2014;Lyster et al., 2013).In particular, Kartchava and Ammar (2014) found that learners' preferences for specific OCF strategies were correlated with their comprehension of teachers' error repair intentions.Therefore, it is crucial for teachers to consider their learners' preferences for OCF strategies.
However, while several studies (Deptolla, 2019;Lee, 2013;Muslem et al., 2021;Sakiroglu, 2020;Wiboolyasarin et al., 2020;Zhu & Wang, 2019) have focused on learners' general preferences or preferences for obtaining OCF for specific errors, research exploring the influence individual differences in L2 situations other than EFL on learners' preferences for OCF techniques is limited.Hence, it is necessary to further investigate the preferences of language learners for OCF techniques in various L2 contexts.

OCF Techniques and Their Significance in L2 Classrooms
The role of OCF techniques in language learning has been a subject of interest among scholars who have examined the various ways in which these techniques can facilitate L2 acquisition.Lyster and Ranta (1997) provide a succinct framework for classifying feedback, based on their observations in French immersion classrooms.Their classification includes six main types of feedback: recast, explicit correction, elicitation, repetition, clarification requests, and metalinguistic clues.More recent research has resulted in hierarchical taxonomies of OCF methods, based on theoretical understanding of how CF functions for acquisition.Two essential distinctions of OCF techniques are input-providing vs. output-prompting, and explicit vs. implicit CF, as shown in Table 1 (Ellis, 2009;Li, 2018).
There are differing opinions regarding the classification of OCF techniques.For instance, Li (2021) classifies recast and explicit correction in one category, while the other four are put in another category because the former provides the correct form, and the latter (referred to as 'prompts') mo-tivates students to self-correct.OCF may take place both explicitly and implicitly.Explicit feedback or explicit correction comprises a blatant admission of the student's error.It is offered to provide immediate rectification or clarification of grammar.Conversely, implicit feedback happens when the source (typically the instructor) does not make it clear that the student committed an error (Tasdemir & Yalcin Arslan, 2018).The selective OCF strategies, along with their explanations and examples, are presented in Table 2.
The source of OCF has also been classified into three categories: teacher feedback, peer feedback, and self-feedback (Carless, 2006;Ha & Nguyen, 2021;Rollinson, 2005).Teacher feedback has been discussed in the preceding section, whereas peer feedback is a valuable feedback source that involves students commenting on each other's work.Peer feedback can assist L2 learners in becoming critical readers and acquiring the ability to reflect on their feedback experiences, especially when reading the work of others (Rollinson, 2005).
In the L2 classroom, OCF may be perceived as useful for letting students know when they have performed correctly, but it may also be viewed as potentially harmful because it can impair students' openness to learning (Ellis & Shintani, 2014).The affective and cognitive dimensions of OCF are widely recognised, and practitioners are cautions about offering prescriptive guidance on OCF.While OCF can have cognitive benefits, potential emotional harm also be considered.Lyster and Saito (2010) found that prompts had more significant effects than recast, indicating that prompts are likely more effective than recast in the classroom since they are more prominent.Gooch et al. (2016), in Korean EFL classrooms, found that recast only improved the comprehensibility of the sound in controlled tasks, whereas prompts were helpful for both controlled and free output.However, learners' expectations of OCF may be one of the most important factors that influence L2 acquisition, as highlighted by Kartchava and Ammar (2014).Inconsistencies between students' preferences for OCF strategies and instructors' practices when rectifying mistakes can contribute to inefficient teaching (Lyster et al., 2013).Therefore, it is essential to investigate students' perspectives on OCF, particularly their preferences for OCF categories, based on the material discussed above.

Learners' Preferences for OCF, FLCA, and FLE
The preferences for various types of feedback have been the subject of numerous studies in the past, but more recent research has started to take into account fresh factors in an effort to ascertain whether preferences for OCF may be influenced by learner characteristics such as gender, proficiency, motivation, learning style, educational programme, or educational context (Fidan, 2015;Deptolla, 2019;Gómez Argüelles et al., 2019;Papi et al., 2021;Sepehrinia et al., 2020;Tasdemir & Yalcin Arslan, 2018;Wiboolyasarin et al., 2020Wiboolyasarin et al., , 2022;;Yakisik, 2021).These investigations have been motivated by the possibility that learner traits could affect feedback efficacy (Nassaji, 2016).Additionally, the concept of emotion has been included in the description of learner characteristics.For instance, foreign language learners have reported anxiety due to various factors such as fear of failing tests or public speaking and receiving negative feedback from classmates and teachers (Horwitz et al., 1986).Foreign language classroom anxiety (FLCA) is characterised by the state anxiety that language learners experience during language study and/or use (Horwitz, 2017).Several studies have investigated FLCA to determine its sources and relationship to student performance and accomplishments (e.g., Dewaele, 2017;Horwitz, 2010).
Studies on directed learning situations have examined the link between FLCA and OCF preferences, as well as other variables, among learners.For example, Geckin (2020) discovered that Turkish pre-intermediate EFL female learners differed from male participants in their higher levels of anx-iety, preferences for delayed feedback, and preferences for repetition as the primary error correction approach.Both males and females considered feedback as a critical aspect of language learning and assessed the teacher's input more favourably, especially concerning major and personal errors.Rassaei (2015) demonstrated that low-anxiety students benefited from both metalinguistic feedback and recast, whereas metalinguistic input had a stronger impact on their progress.In contrast, learners with high anxiety benefited substantially more from recast than from metalinguistic feedback.
Positive emotions, conversely, enhance a student's ability to notice classroom details and increase their awareness of language input (MacIntyre & Gregersen, 2012).Students in a positive emotional state, or with foreign language enjoyment (FLE), have a better comprehension of the foreign language and can reduce the adverse effects of negative emotions.However, individuals overwhelmed by negative feelings are more likely to have a restricted focus, thereby limiting their language input intake (Dewaele & Dewaele, 2020).Positive emotions increase learners' sense of security while studying a foreign language.Although Dewaele and MacIntyre (2014) conducted a study that simultaneously analysed variation in FLCA and FLE, a few studies have examined the influence of both FLCA and FLE on learner-internal factors (such as preference) and learner-external variables (e.g., educational context or, in this case, TFL).This study combines FLCA and FLE in the same research design based on Dewaele and MacIntyre's (2014) questionnaire items to determine whether they influence OCF approaches.Additionally, research (e.g., Yang, 2016) shows that the proficiency level and first language (L1)

Elicitation
The instructor elicits the correct form through the use of a technique in which a part of the student's speech is repeated, omitting the incorrect portion and prompting the student to complete it independently.
S: I went there tomorrow.
I: I …… there tomorrow.

Explicit correction
The instructor accurately identifies the error made by the student and subsequently delivers a correction, providing clear indication that the error has been made.
S: I went there tomorrow.
I: No, 'went' is incorrect.You should say 'I will go there tomorrow.'

Metalinguistic feedback
The instructor addresses questions or remarks and explains using grammatical or other linguistic concepts in order to elicit information from the students.
S: I went there in Monday.
I: Not in Monday, We use a preposition 'on' for days and dates.We say, 'I went there on Monday.'

Recast
The instructor provides only the correct form of the student's erroneous speech without any additional commentary.
S: I went there tomorrow.
I: I will go there tomorrow.

Repetition
The instructor utilises a questioning intonation to highlight the error in the student's statement, thus indicating its incorrectness.
S: I went there tomorrow.
I: I WENT there tomorrow.

Clarification request
The instructor indicates that the student's statement is unclear or incomprehensible, and requests clarification to ensure clear communication.
S: I went there tomorrow.
I: What?Can you please explain it to me more precisely?
| Research Papers context of learners are the main determinants shaping their preferences for OCF.Consequently, it is vital to thoroughly analyse learners' preferences for OCF techniques in connection with their L1, proficiency level, FLCA, and FLE.This study aims to address the following research questions to enrich the understanding of the existing literature: (1) What are the preferences of TFL students for OCF techniques?
(2) Do the proficiency level and L1 background of TFL learners influence their preferences for OCF techniques?
(3) Do the FLE and FLCA levels of TFL learners influence their preferences for OCF techniques?

METHOD Context
This study examines Thai as Foreign Language (TFL) programmes offered at Chinese, Japanese, and Korean universities where Thai is used as the medium of instruction.
The language departments of these universities offer Thai elective courses to local undergraduate students, aimed at enhancing their academic Thai language skills and cultural knowledge.

Participants
A non-probabilistic convenience sampling was utilised to select undergraduate students from TFL classrooms for this study.To ensure an adequate sample size, G*Power 3.1.9.7 was used to calculate a medium sample size of 0.25, power (1-β) = 0.95, and α = 0.05.The estimation suggested that 252 people would be sufficient.A total of 288 students participated in the study and completed an online questionnaire administrated through Microsoft Forms (see the following section for more information).

Questionnaire
An online questionnaire consisting of four sections were used to collect data.Part 1 began with a section on demographics, from which the preceding information was extracted.Following this, participants were asked to rate their preferences for ten OCF techniques (OCFT) (see Appendix), modified from previously discussed literature and Wiboolyasarin et al.'s (2020Wiboolyasarin et al.'s ( , 2022) ) instruments, using a 5-point Likert scale to assess how learners feel when they make an oral error in TFL lessons, ranging from 'very good' (5 points) to 'very poor' (1 point).In Part 3, the FLE scale is a 21-item instrument proposed by Dewaele and MacIntyre (2014) to which participants replied using a 5-point Likert scale ranging from 'strongly agree' (5 points) to 'strongly disagree' (1 point).For measurement of FLCA in Part 4, the FLCA scale constructed by ibid.( 2014) contained a total of 8 items.The alpha reliabilities for the OCFT, FLE, and FLCA in the present investigation were 0.81, 0.94, and 0.93, respectively, showing acceptable internal consistency for the instruments (Cohen et al., 2018).

Procedure and Data Analysis
Prior to the commencement of the study, the research design and questionnaire were subject to approval by the Ethics Committee of the first author's institution.Prior consent was sought from all participants before administering the survey.The online questionnaire was made accessible via Microsoft Forms for a period of one month.Initially, 288 students completed the questionnaire online.The participants were then segregated into two groups, based on their scores on the FLE questionnaire: high-enjoyment learners (n = 134) and low-enjoyment learners (n = 154).High-enjoyment learners were classified as those who scored higher than the mean, while low-enjoyment learners were classified as those who scored lower.Similarly, participants were divided into two groups based on their FLCA questionnaire scores: those with high anxiety (n = 154) and those with low anxiety (n = 134).The mean score on the FLCA questionnaire was 3.55, with a standard deviation of 1.05.Participants who scored above the mean were classified as high-anxiety learners, while those who scored below were classified as low-anxiety learners.

RQ1: What are the Preferences of TFL Students for OCF Techniques?
On the 10 items of Part 2 in the questionnaire, all respondents were required to rate their feelings when making oral blunders in TFL classes.The frequencies and average rating scores on a 5-point Likert scale are presented in The results show that TFL students preferred metalinguistic comment (M = 4.39) and explicit correction (M = 4.28), respectively.Private feedback (M = 3.96), peer correction (M = 3.66), and recast (M = 3.66) also received positive responses, with overall 'good' ratings.These findings suggest that participants preferred OCF techniques that were more explicit.
In contrast, ignoring errors was the least desirable OCF type, with mean scores of 2.65.The average values for elicitation, repetition, explanation request, and public feedback ranged from 2.80 to 3.28, indicating that participants tended to consider them unfavourable TFL classroom strategies.Descrip-tive statistics of the participants' characteristics classified by L1 background, proficiency level, FLE, and FLCA can be found in Table 4.
The skewness and kurtosis values for each OCF strategy were also analysed, and the results showed that the values between -1.21 and 0.75 were distributed normally, which falls within the range of ±2 for skewness.The kurtosis scores ranged from -1.05 to 0.64, indicating a typical distribution, as they fall within the scale of ±2 (George & Mallery, 2016).The normality assumption was assessed using the Shapiro-Wilk test, which tested the univariate normality of ten OCF approaches.This test was chosen due to the sample size being less than 2,000 (Adigun, 2021).A multivariate analysis of variance (MANOVA) was then conducted, with OCF techniques serving as the between-subjects variable, and proficiency level and L1 background as dependent variables.The MANOVA results indicated significant differences in proficiency level and L1 background across OCF strategies (Pillai's trace = 0.18, F = 1.30, p = 0.031), suggesting that the interaction of these two factors can explain the variance in the preferences of L2 learners for OCF.To ensure the validity of the MANOVA, it was confirmed that the covariance matrices among groups were equal, as determined by a Box's M test (p < 0.001) (Huberty & Petoskey, 2000).
Based on the aforementioned results, a between-subject analysis or a univariate test was conducted for each dependent variable, as presented in Table 5.
Table 5 displays  and public feedback (F (4, 283) = 0.95, p = 0.44) indicated a small variation between the average of the groups.As is evident, there was a strong interaction of L1 by proficiency on disregard, peer correction, recast, and private feedback.
A simple slope was plotted to gain a clearer understanding of the overall pattern of interactions and to identify the factor that predicted whether the participants' L1 background (Chinese, Japanese, or Korean) and their proficiencies were significantly related to these four dependent variables.The slopes of the graphs in Figures 1, 2, 3, and 4 are represented graphically, showing that four OCF techniques for learners had significant interactions with respect to their L1 background and proficiency level.

RQ3: Do the FLE and FLCA Levels of TFL Learners Influence Their Preferences for OCF Techniques?
A MANOVA was conducted with OCF techniques as the between-subjects variable, and the FLE and FLCA levels as the dependent variables.The overall MANOVA, which examined the differences in FLE and FLCA levels across OCF strategies, was found to be non-significant.To further explore the relationship between FLE, FLCA, and L2 learners' OCF preferences, Pillai's trace was employed.The analysis revealed no significant interaction between FLE and FLCA (Pillai's trace = 0.38, F = 1.10, p = 0.58), indicating that only 38% of the variability in L2 learners' OCF preferences can be attributed to the interaction between FLE and FLCA.These results suggest that other factors may have a more significant impact on L2 learners' preferences for OCF, and further investigation is warranted.
In light of these results, the between-subject effects or univariate tests for each dependent variable was conducted, which are depicted in Table 6.

DISCUSSION
It is widely observed that L2 learners tend to prefer receiving OCF as it acknowledges errors as a natural occurrence in the process of acquiring an L2.However, traditional TFL instruction places emphasis on language form, leading to Chinese, Japanese, and Korean TFL students feeling less confident in their ability to speak the target language when teachers disregard their errors.Our study shows that the majority of students preferred the use of explicit OCF techniques, specifically metalinguistic remarks and explicit correction.Therefore, it is recommended that foreign language courses adopt strategies that teach the correct version of errors and linguistics terminology, allowing students to enhance their language skills without the burden of self-correction (Lee, 2017;Ur, 2012).
In L2 programmes, OCF from instructors is regarded as crucial, as students often lack sufficient exposure to the target language in their environment, as Muslem et al. (2021) have emphasised.It is acknowledged that Thai was not employed as a lingua franca in the previous classes of other students, depriving them of enough opportunities to practise their oral Thai.Our findings indicate that learners appreciate being informed directly of their errors and the correct forms, and they do not feel any shame when corrected by their teachers or classmates.This finding reinforces prior studies, suggesting that OCF may not lead to antagonising or demoralising East Asian students (Wiboolyasarin et al., 2020) or other EFL learners (Wiboolyasarin et al., 2022;Yang, 2016;Yu, 2019).
Although it is widely recognised that providing OCF in a public setting can be beneficial for the entire class, this study's results indicate that the instructor should provide such feedback in private.Some students consider OCF to be restrictive and humiliating when it is frequently used in the classroom (Martínez Agudo, 2013).In particular, students who receive excessive corrective feedback in front of their peers may become bashful and experience an increase in anxiety.This finding is consistent with previous research (Oflaz, 2019), which found a positive correlation between shyness and FLCA, demonstrating conclusively that shyness increases as speaking anxiety grows.
Based on the findings, participants tended to perceive elicitation, repetition, and explanation requests as unfavourable TFL classroom tactics.Several reasons can explain this phenomenon.For example, when the teacher echoes the incorrect speech with a rising intonation and a doubtful expression, trying to imply that there was an error, students may not recognise it as a correction or may not realise that the instructor's pronunciation was different from their own, interpreting it merely as an echo or confirmation.Similarly, active OCF strategies that encourage students to rethink what they have said and construct the correct form independently did not yield positive results for East Asian students who are accustomed to being teacher-fed information.
Li's (2021) research revealed that various factors related to the learner and the context can constrain the effectiveness of feedback.Among these factors, a student's emotional engagement with an instructor's OCF can shape the student's perception and response to the feedback received.It would be intriguing to investigate if two learner variables, namely proficiency level and L1 background, could influence learners' preferences for OCF approaches.The literature has shown that learners' preferences for OCF can differ based on their proficiency and L1 background.Our study confirmed that Korean TFL learners preferred fewer OCF techniques, such as disregard, peer correction, recast, and private feedback, than their Chinese counterparts.In addi-tion, advanced-level students favoured more OCF approaches than intermediate-and beginner-level students.These results may be due to the fact that less proficient students may feel embarrassed when receiving OCF on their erroneous speech, especially when the feedback includes wrong information.These findings suggest that learners from diverse linguistic backgrounds and proficiency levels are aware of their errors and expect instructors to provide OCF feedback appropriately.
Although instructors use various OCF techniques to provide feedback to students, it is crucial to adapt their strategies to suit students' preferences (Tasdemir & Yalcin Arslan, 2018).
As Bada and Okan (2000) stated, instructors should consider their students' expectations and provide each student with the opportunity to express their opinions.To our knowledge, previous studies have not explored the relationship between FLCA/FLE levels and OCF preferences extensively.Therefore, our study provides a basis for further research.Although there were no statistically significant differences between FLE and FLCA for OCF techniques, we observed that low and high FLE/FLCA groups had distinct preferences.Specifically, high FLE groups favoured more OCF techniques than low FLE groups, while high FLCA groups selected more OCF techniques than low FLCA groups, except for elicitation, recasting, and public feedback.This finding has significant pedagogical implications, as it suggests that instructors need to use OCF techniques that enable L2 students with different psychological profiles to feel confident that their errors are being detected and corrected appropriately.Furthermore, instructors should focus on frequent and major errors and refrain from providing OCF strategies for some errors.Nonetheless, if instructors believe that feedback on specific errors is essential, they can use metalinguistic feedback and explicit corrective approaches, which are more effective than recast (Lyster, 2004;Lyster & Saito, 2010), and

CONCLUSION
The current study offers a novel contribution to the field by examining the characteristics of L2 undergraduate TFL learners in East Asian universities, and exploring the potential impact of proficiency level, L1 background, FLE, and FLCA on preferences for OCF approaches.Such information could prove useful for educators working with this cohort of students across different regions.Despite the complexity of the findings, the results suggest that TFL learners generally prefer metalinguistic explanation and explicit correction, although there is a significant relationship between L1 and proficiency with regards to disregard, peer correction, recast, and private feedback.Furthermore, FLCA and FLE were found to have a notable impact on individuals' perceptions of the value of more prompts and explicit correction.Thus, TFL teachers may wish to provide learners with more explicit OCF on oral errors, particularly those with high FLE and FLCA, as this could promote greater engagement with feedback.This study offers insights into the factors that may shape OCF preferences and highlights the potential significance of these features for OCF preference formation.
It is important to note several caveats in interpreting the results of this research.Firstly, the participants were drawn from a population of students at universities in three countries where Thai is the language of instruction, and therefore, the findings may have limited generalisability to other institutional contexts.Additionally, as this study only examined four characteristics of student preferences for OCF, its findings may not fully capture their impact on learning outcomes.Moreover, the sample size was limited to university-level students, which may not be representative of other age groups.Further research is therefore needed to investigate the preferences for oral feedback across a wider range of students.

ETHICAL STATEMENT
The investigation involving human participants was approved by the Ethics Committee of the first author's university.Informed consent was obtained from all participants, who were informed that participation in the online questionnaire survey was voluntary, and they could withdraw at any time.The online questionnaire was designed to automatically obtain the participants' informed consent upon submission.

Figure 1 SimpleFigure 2 SimpleFigure 3 SimpleFigure 4 Simple
Figure 1 Simple Plot of Interaction Effects Between Proficiency and L1 Towards Disregard

Table 3
Frequencies of Answers and DescriptiveStatistics of 10 Items (n = 288)

Table 5
Test of Between L1 and Proficiency Effects on OCF Techniques

Table 6
Test of Between FLE and FLCA Effects on OCF Techniques