An Investigation of Iranian EFL Teachers and Learners’ Preferences in the Selection of Different Types of Direct and Indirect Oral Corrective Feedback

Introduction

In the last few decades, error correction in language classroom settings has gained considerable attention in second language acquisition (SLA) research. According to Chaudron (1977, p. 31), “any reaction of the teacher which clearly transforms, disapprovingly refers to, or demands improvement of the learner utterance” provides feedback in a corrective way. Lightbown and Spada (1999) defined corrective feedback as “any indication to the learners that their use of the target language is incorrect” (p.171). To their views, such an indication includes various responses that the learners receive from their teacher. Li (2010) has recognized CF in second language classroom as the responses to a learner’s non-target-like L2 production. Moreover, Leeman’s (2007) defined feedback as a reaction to the utterance of the learner.

There have been several studies concerning the role of corrective feedback (CF) in SLA studies (Bitchener, Young, & Cameron, 2005; Burnett, 2002; Ellis, 2009; Ellis, Loewen, & Erlam, 2006; Han, 2002; Leeman, 2007; Li, 2010; Li & Li, 2012; Lightbown & Spada, 1999; Lyster, 1998a, 1998b; Lyster, 2004, Lyster, & Ranta, 1997; Mackey, 2012; Philp, 2003; Rassaei, 2013 a& b , 2015 a & b; Russell & Spada, 2006; Schachter, 1991; Suzuki, 2012, Zhang & Rahimi, 2014). Discussing the increased attention to CF studies, Ellis (2005) stated that the theoretical motivation for this interest lies in the claim that L2 learning (unlike L1 learning) requires negative evidence as well as positive evidence (i.e., learners need to be shown what is NOT correct as well as provided with examples of what IS correct). (pp. 19-20)

According to Diab (2006), CF will be more promising when both teachers and students share common ideas about the use of CF strategies. Moreover, Lyster, Saito, and Sato (2013) called upon the significance of doing research on language learners’ CF preferences. They opined that the knowledge about the learners’ preferences can lead to a more effective error correction on the part of language teachers. All the same, most studies have so far focused on the learners’ preferences for CF on specific language tasks
(e.g., grammatical structures) in classroom settings to the neglect of teachers’ preferences for certain strategies for error correction (Amrhein & Nassaji, 2010; Diab, 2006; Halimi, 2008; Nassaji, 2012; Rassaei & Moinzadeh, 2014; Zhang & Rahimi, 2014). Hence, this study aimed to describe and compare the Iranian EFL teachers and learners’ preferences for the use of four types of CF strategy.

Review of the Related Literature

In the past three decades, error correction has been extensively investigated (Burnett, 2002; Ellis, 2009; Lyster & Ranta, 1997; Philp, 2003). As Li (2010) stated, the growing bulk of research in error correction demands studies into the variables influencing its effectiveness.  Some recent meta-analyses (Li, 2010; Lyster & Saito, 2010; Mackey & Goo, 2007; Russell & Spada, 2006) showed the positive role of CF in language teaching and learning. Beigi Rizi and Ketabi (2015) had a close look at six decades of corrective feedback wherein they examined changes in the way CF was viewed and practiced. They also reviewed its evolution through history from 1950s to 2000s in English language teaching. They tried to make researchers and practitioners get acquainted with different views about CF and change their methodological perspectives on CF.

Some studies have been done regarding the perception of different types of CF (e.g., Carpenter, Jeon, Mac Gregor, & Mackey, 2006; Egi, 2010; Kim & Han, 2007; Mackey, 2006; Rassaei, 2013a, 2013b). Following Schmidt’s noticing hypothesis, most of these studies put forward noticeability and perception of corrective feedback as an effective factor, suggesting that the explicit types of CF outperform the implicit ones. Some of these studies (e.g., Gass & Lewis, 2007; Mackey, Gass & McDonough, 2000; Mackey, Al-Khalil, Atanassova, Hama, Logan-Terry, & Nakatsukasa, 2007) found that learners generally better understand their teachers’ intentions when feedback is explicit. Moreover, several empirical studies demonstrated that more explicit types of corrective feedback (e.g., metalinguistic clues) are more effective than implicit CF types (e.g., recast) in developing a second language (e.g., Carroll & Swain, 1993; Ellis, 2007; Ellis, Loewen, & Erlam, 2006; Lyster, 2004; Sheen, 2007).

However, in comparison to the studies on the role of CF in SLA, and the perception of different types of CF, there has been a few studies on preferences for different CF types with mixed findings (Kaivanpanah, Alavi, & Sepehrinia, 2012; Lee, 2013; Oladejo, 1993; Zhu, 2010). Oladejo (1993) realized that the learners preferred those CFs enjoying the feature of presenting comments, guidance, or clues (e.g., metalinguistic clues) which consequently provided them with the chance of self-correction, or those CF types which pointed them the error especially if  the CF provided them with the correct answer. Kaivanpanah et al. (2012) found that learners had strong preferences for metalinguistic clues and recasts. They presumed that such a preference is largely due to the Iranian teachers’ authority in their classrooms. In another study by Lee (2013), metalinguistic clues gained a weak preference since learners considered it beyond their proficiency level. The participants of this study confessed that disagreements about CF may bring about discouragement and embarrassment in language classroom which might affect their conversational practices in future. Although Lee (2013) found strong preferences for explicit or direct feedback, other studies (e.g., Yoshida, 2008; Zhu, 2010) showed the learners’ preferences for CF types with self-correction opportunities over the mere provision of correct form of erroneous utterances.

Moreover, recast proved to be the most frequently used CF in language classrooms (Lyster & Ranta, 1997; Lee, 2013; Yoshida, 2010). Some researchers found that recasts were not preferred for such reasons as low degree of noticeability or comprehensibility (Lyster & Ranta, 1997; Mackey et al., 2000). In other words, the preferences were pertinent to learners’ level of perception about CF at the time of feedback provision. However, as Yoshida (2010) stated, recasts were preferred strongly by the EFL teachers over other types since they were introduced as easy, short, and implicit, and thus less unapproachable, causing no negative emotions, anxiety, nor embarrassment. The mixed research findings on preferences for CF types and the role of CF in language learning have pushed the researchers to discuss the issue more deeply.

This study aimed to investigate EFL teachers and learners’ preferences for different types of CF. Accordingly, for the purpose of this study, a typology for oral corrective feedback (OCF) was needed to recognize different CF types and assess the levels of preferences. Chaudron (1977) presented an inclusive model on the basis of the data collected from his immersion classrooms. Chaudron’s (1977) model was an important endeavor in identifying a number of corrective techniques and also a significant step toward exploring the relationship between error types, feedback types, and repairs of the learners. He found that the teachers’ most commonly used feedback type was the one in which the learner’s utterances are reformulated, and this reformulation is accompanied by several features (e.g., emphasis, reduction, and negation, expansion and unaltered repetition.

Moreover, in a typology presented by Lyster and Ranta (1997), a six-type taxonomy was provided that can be claimed as one of the most influential typologies of CF. Lyster and Ranta (1997) identified and categorized different types of CF as follows: a) explicit correction (viz., The teacher provides the student with the correct form and noticeably shows that what the student had said was incorrect); b) recast (viz., Without directly indicating that the student's utterance was incorrect, the teacher implicitly reformulates all or part of the student's error, or provides the correction); c) elicitation (viz., Teacher directly elicits a reformulation from student (making the student produce correct form) by asking questions or by pausing to allow the student to complete teacher’s utterance, or by asking him/her to reformulate his/her utterance); d) metalinguistic clues (viz., Teacher provides a meta-linguistic comment, that is providing comments, information, or questions related to the well-formedness of the student’s utterance, without explicitly providing the correct form); e) clarification request (viz., By taking advantages of using some phrases like “Pardon?”, "Excuse me?" or "I don't understand," the teacher indicates that the message has not been understood or that the student's utterance has been ill-formed and so a repetition or a reformulation is required); f) repetition (viz., the teacher repeats the student’s erroneous utterance, adjusting intonation to highlight the error). For the purpose of this study, Lyster and Ranta’s (1997) CF typology was utilized because it has frequently been used in the literature (i.e., Ellis, 2009; Smith, 2004), and it has remained the most credited to the present. Hence, the study was an attempt to answer the following questions:

Are there any statistically significant differences among the low-, moderately-, and high-experienced EFL teachers’ preferences for indirect corrective feedback (i.e., recast and repetition) and direct corrective feedback (i.e., metalinguistic clues and explicit correction)?

Are there any differences among Iranian teachers and learners’ preferences for indirect corrective feedback (i.e., recast and repetition) and direct corrective feedback (i.e., metalinguistic clues and explicit correction)?

Method

For the purpose of this mixed-method study, a sequential design (Hashemi & Babaii, 2013) was used. Data was collected both quantitatively and qualitatively to triangulate the data with a confirmation purpose. For the quantitative data collection phase, CF preference questionnaire, and bio-data questionnaire was used. The interview was used to collect the qualitative data.

Participants

Multilevel sampling design (Hashemi & Babaii, 2013) was used to select the participants of the study. Hence, two samples from different populations (i.e., EFL teachers and learners in private institutes in the capital city of an Iranian province) were selected through purposive sampling. However, because it was impossible to balance the two groups (n_teachers=39, n_learners=84), it was decided to match the groups in such way that the variances of the groups were equal (Dörnyei, 2007). Moreover, in order to control the extraneous variable of gender, it was decided to select both subsamples from one gender (i.e., male). The age of EFL teachers ranged from 22 to 25, and the age range of EFL learners was from 16 to 18. In terms of academic degree, all of the participating EFL teachers held a BA in English, and the participants of the second subsample (i.e., EFL learners) were high school students, who were identified as intermediate level learners as measured by the Quick Oxford Placement Test. Furthermore, Persian was the first language of the participants of both subsamples. However, the EFL teachers were heterogamous in terms of the years of experience in teaching English as a foreign language. Following Tajeddin and Khodaverdi (2011), they were therefore assigned into three groups in terms of years of teaching experience, namely, low (3˂ years), moderate (3-5 years), and high (5 ˃ years).

Instruments

CF Preference Questionnaire

For the purpose of collecting data on the teachers and the learners’ overall preferences for the type of CF to give and receive respectively, a 10-point Likert-scale scenario-based questionnaire was constructed based on the CF typology by Lyster and Ranta (1997) (Appendices A and B). Out of the six types of CF, recast, repetition, metalinguistic clues, and explicit correctionwere selected.Recast as an input-providing CF type and repetition as an output-prompting CF type represented indirect oral corrective feedback (IOCF), while metalinguistic clue as an output-prompting CF type and explicit correction as an input-providing CF type were representative of direct oral corrective feedback (DOCF). Hence, elicitation and clarification request were excluded on a random decision because they characterized the features of metalinguistic clues and explicit correction as output-prompting and input-providing CF types, respectively (Ellis, 2009).

Moreover, it should be noted that a larger Likert scale (i.e., 10-point) will offer more variance than a smaller Likert scale (e.g., 7-point or 5-point Likert scales), higher degree of measurement precision, and better opportunity in detecting changes (Hassan & Ghazali, 2012). The feedback scenarios were extracted from the literature on CF. The scenarios in the questionnaire were similar to situations wherein the EFL teachers and learners might have already faced in their EFL classes. To ensure the content validity of the questionnaire, it was expert-judged by four university professors who had experience in conducting research in CF. The instrument was then revised based on the comments given by the experts. Yet, out of the six items only four items (recast, repetition, metalinguistic clues, and explicit correction) were taken into consideration for later data analyses, and elicitation and clarification request were excluded because they necessitated the EFL learners’ immediate reaction to their teachers’ feedback.  The participants of both subsamples were required to rate the scenario-based items of the questionnaire based on their preferences, ranging from 0 (weak preference) to 9 (strong preference). The questionnaire showed acceptable internal consistency reliability (α=.83). Moreover, four demographic items were included in the instrument to control gender, age, first language and academic degree of the participant of the study as the intervening variables.

Semi-Structured Interview

In order to confirm the quantitative data, semi-structured interviews were also conducted and audio-taped in order to triangulate the data gathered through the questionnaire. Each interview was conducted in three sessions of fifteen minutes. For the purpose of the qualitative data collection through interview, an interview protocol, containing 4 questions and 4 probes, was developed.10 out of 84 EFL learners, and 9 out of 39 EFL teachers voluntarily took part in interviews.

Procedures

Initially, a subsample of both EFL teachers and learners were selected through purposive sampling. Then, the teachers were divided into three groups based on their years of teaching experience, labeled as low-experienced (LEG), moderately-experienced (MEG), and highly-experienced (HEG). Afterward, a researcher-made questionnaire was constructed based on the literature on CF. The instrument was distributed to the participating teachers in a one-to-one fashion by one of the researchers since they were not available in one single institute. However, the questionnaire was given to EFL learners collectively in their classes by the same researcher. Finally, the quantitative data collected through questionnaire were subjected to descriptive and inferential statistics, and then the interview data were analyzed and compared with the quantitative data to triangulate and confirm the results.

Results

Results for the First Research Question

Table 1 shows the teachers’ overall preferences for different types of OCF, regardless of their levels of experience from the best preferred to the least preferred.

Table 1. Descriptive Statistics for Teachers’ OCF Preferences

As shown in Table 1, the overall preferences of the Iranian EFL teachers on the types of feedback from the most preferred one to the least were repetition, recast, metalinguistic and explicit feedback, respectively (M_Rep= 6.94> M_Rec= 6.23> M_Met= 4.76> M_Exp= 2.74). Moreover, according to the results, teachers were more inclined toward repetition and recast than metalinguistic clues and explicit correction. In other words, the teachers preferred IOCF over DOCF. According to Table 2, the results of ANOVA demonstrated that there was a significant difference among three groups of teachers’ preference for repetition as an IOCF, F(2,36)=5.09, p< .05). Moreover, the results showed that there was a significant difference among the three groups of teachers’ preference for metalinguistic clues as a DOCF, F(2,36)=5.63, p< .05). However, the results of ANOVA showed no significantly statistical differences among the three groups of teachers’ preference for recast and explicit correction (Table 2).

Table 2. One-way ANOVA among three groups of teachers’ preferences for different types of OCFs

The meanplot for each CF group displays the association of the CF type’s preferences and the teachers’ levels of experiences.

Figure 1. Meanplot for repetition

As Figure1 shows, the low-experienced teachers showed the least average score in preferences for repetition, the high-experienced teachers showed the highest average score. The moderately-experienced ones took a rather moderate view of repetition preferences as their average scores suggested that. This suggests that the more experienced a teacher is, the more s/he is interested in repetition as an IOCF.

Figure 2. Meanplot for recast

Moreover, as shown in Figure 2, the high experienced teachers preferred recasts more than the moderately experienced and low experienced teachers.

Figure 3. Meanplot for metalinguistic clues

As Figure 3 suggests, the average preference score for metalinguistic clues was the highest among the low experienced teachers, and the lowest among the high experienced teachers. Meanwhile, the moderately experienced teachers also showed a tendency toward explicit correction, which was notably higher than that of the high experienced teachers but lower than that of the low experienced teachers (Figure 3).

Figure 4. Meanplot for Explicit correction

As shown in Figure 4, high experienced teachers had the lowest preference for explicit correction, whereas the low experienced ones showed the highest preference for it. This means that as the level of EFL teaching experience increases among the teachers, the EFL teachers’ preference for explicit correction decreases.

Furthermore, the results of qualitative data analysis via interviews with the teachers confirmed the results of the teachers’ preferences. The analysis of the three teachers’ interviews extracts (Appendix C) showed that all three high experienced teachers preferred or used IOCF while only one of them preferred or used DOCF, too. Even, they preferred IOCF in the first place, and then preferred or used IOCF as well. Therefore, the interview data demonstrated that high experienced teachers preferred IOCF over DOCF, and such a result was in line with the teachers’ overall preferences for different types of OCF as shown by the questionnaire data, which showed low preference for DOCF. Moreover, the findings of the interview corresponded with the mean differences of HEG teachers’ OCF preferences
(Table 1). This demonstrated that the results of both the questionnaire and interview data indicated that IOCF was more preferred.

The moderately experienced teachers’ interview data indicated that although they preferred all types of CF regardless of its directness or indirectness, they emphasized IOCF more than DOCF. However, the preference for IOCF was not as strong as the one existing among highly-experienced teachers. The interview data from low-experienced teachers displayed that they preferred the DOCF more, and even almost no preference for IOCF. Such findings were in complete harmony with the mean differences of LEG teachers’ preferences (Table 1). This demonstrates that the results of both the questionnaire and interview indicates that DOCF was more preferred by the teachers from LEG.

Moreover, it was found that all teachers knew about the differences between students’ errors and their mistakes; as a result, they had never corrected a student’s erroneous utterance for the first time since it could have been a mistake due to stress or fatigue. Thus, they recognized an erroneous utterance as an error it was repeated for the second time.

Results for the Second Research Question

To answer the second research question, the data from students’ preferences for different types of OCF were subjected to descriptive statistics (Table 3).

Table 3. Descriptive Statistics for Student’s OCF Preferences

As shown in Table 3, students preferred recast the most, and explicit correction the least (M_Rec= 23.26> M_Rep= 21.82> M_Met= 20.86> M_Exp= 11.78). Compared to the results from the mean scores of teachers’ OCF preferences (Table 1), the same results were true for the students. That is, they preferred IOCF over DOCF. The only difference between the teachers’ preference for CF types and that of the students was the order of the OCF types. In other word, the students preferred recast over repetition. Table 4 summarizes the results of the students’ interviews obtained from the extracts from the students’ interview (Appendix D).

Table 4. Summary for the 10 students’ interviews

As displayed in Table 4, IOCF was more frequently preferred by the interviewees (n=15/22), than DOCF (n=7/22). Moreover, according to the results of the interview data, the most preferred was recast, and explicit correction was the least preferred (n_recast(IOCF)=9 > n_rep(IOCF)= 6 > n_met(DOCF)=5 > n_exp(DOCF)= 2).

As it can be seen, the analysis of the students’ interview showed that IOCF was more preferred than DOCF by the students, and this was in line with the teachers’ preferences in the interviews. Moreover, the mean differences of student’s overall OCF preferences showed the same results as compared to the teachers’ overall OCF preferences, that is, IOCF showed to be more preferred than DOCF.

Discussion

The present study explored Iranian EFL teachers and learners’ preferences for four types of CF (i.e., recasts and repetition as IOCF, and metalinguistic clues and explicit correction as DOCF). The participating EFL teachers and learners’ preference for CF types were tapped on using a research-made questionnaire based on Lyster and Ranta’s (1997) CF typology. Moreover, semi-structured interview was used to triangulate the data with a confirmation purpose (Hashemi & Babaii, 2013).

Concerning the first research question, the results of ANOVA showed that teaching experience may play a key role in determining the choice of CF type which EFL teachers prefer. Accordingly, the choice of CF strategies based on directness/indirectness dichotomy (Ellis, 2009) is influenced by EFL teachers’ teaching experience level. Based on the means plots of EFL teachers’ preferences, it can be argued that the more experience EFL teachers gain, the less they prefer explicit or direct CF types. This finding contradicts with those studies which called for direct CF types (e.g., Carroll & Swain, 1993; Ellis, 2007; Ellis, Loewen, & Erlam, 2006; Lyster, 2004; Sheen, 2007). The results of the quantitative and qualitative data analyses suggest that implicit types of CF are preferred over the explicit ones, a finding which was true about both EFL teachers and learners.

As to the second research question, the results revealed that Iranian EFL teachers and learners preferred IOCF (i.e., repetition and recasts) over DOCF (i.e., metalinguistic clues and explicit correction). This result is in contrast with the studies reviewed earlier, which argued for the higher efficacy of the explicit types of feedback in language instruction (e.g., Carroll & Swain, 1993; Ellis, 2007; Ellis, Loewen, & Erlam, 2006; Lyster, 2004; Sheen, 2007). This finding may be due to the nonthreatening nature of IOCF, which is psychologically counted as a passive error correction strategy (Yoshida, 2010). Hence, the experienced teachers preferred this CF type more as compared to the less experienced teachers. This finding can also justify EFL learners’ higher preferences for IOCF, who may have felt more secure and unstressed when receiving CF in its indirect form. Although this finding is in line with Diab’s (2006) opinion that both teachers and students should share common ideas regarding CF strategies if more promising results expected, this preference for IOCF might negatively affect L2 development in Iranian EFL learners because there are numerous studies emphasizing the explicit OCF types (e.g., Gass & Lewis, 2007; Mackey, Gass & McDonough, 2000; Mackey, Al-Khalil, Atanassova, Hama, Logan-Terry, & Nakatsukasa, 2007).

At the same time, while previous studies (e.g., Lyster & Ranta, 1997; Lee, 2013; Yoshida, 2010) have shown that recasts are the most frequently used type of IOCF, the results of this study disclosed that there was a mismatch between EFL teachers and learners’ preference the two types of IOCF as the former group preferred repetition over recast and the latter group preferred recasts over repetition. This finding may be justified on the ground that the EFL teachers, especially the high-experienced ones, think of repetition as output-prompting (Ellis, 2009) because it provides learners with more chances to self-correct themselves. Moreover, the teachers preferred not to provide their learners with correct forms since repetition is a non-provisional (Ellis, 2009) type of feedback. Moreover, the EFL learners’ preference for input-providing IOCF may be due to both psychological reasons such as security (Yoshida, 2010), if not linguistic reasons.

Future studies can be conducted with a larger sample of EFL learners at other levels of English proficiency or from high school or university settings. Moreover, further studies can be done to investigate the EFL teachers and learners’ preferences for elicitation and clarification request as other output-prompting and input-providing CF types. Moreover, future studies should investigate the effect of other factors that may influence EFL teachers and learners’ CF preferences, such as psychological and linguistics factors. Furthermore, replication of this study will help the generalizability of the findings. It is also suggested that future studies investigate EFL teachers and learners’ preferences in the context of experimental studies.

Conclusions and Implications

This study aimed at exploring Iranian EFL teachers and learners’ preferences for OCF and the differences among OCF preferences of EFL teachers with differing levels of teaching experience. Based on the quantitative and qualitative results, it can be concluded that IOCF types are preferred over DOCF by both Iranian EFL teachers and learners. Based on the results, this conclusion can be drawn that the EFL teachers’ preference for IOCF results from the level of their teaching experience and the non-threatening nature of IOCF. According to the results, it can also be concluded that EFL teachers and learners’ preferences for OCF types are more bound to indirectness of the OCF.  Based on the results of both quantitative and qualitative data, a final conclusion may be drawn that there is a match between the EFL teachers and learners’ preferences for IOCF.

The findings of this study may have implications for various groups. In the light of insights from the results of the recent studies, OCF has turned to an important constituent of foreign language instruction. Moreover, students errors are not viewed failures but as mediums that contribute to second language acquisition. The results of this study may provide EFL teachers with the evidence that the more experienced the EFL teachers are, the less mismatch exists between their preferences for CF and those of their learners. Hence, language teachers must take their learners’ expectations and preferences into consideration and take advantages of them in their classes. This study offers insights to assist students to clarify their misconceptions about their teacher’s CF strategies, and to decrease the mismatch between their teachers’ CF strategies and their expectations. In the long run, this will lead to more efficient language development on the part of EFL learners. This study may also persuade material developers to include OCF instructions in their teacher manuals. Regarding the potentials of the OCF discussed in this study, textbook writers are recommended to prepare the instructional guides in accordance with the preferences and the teaching experience level of EFL teachers.

Acknowledgments

The authors of this paper would like to express their gratitude for the helpful comments by the respected reviewers.