Teachers’ Perceptions of Oral Corrective Feedback in Form-focused Language Classrooms: Why do they Correct the Way they do?

This study focuses on Saudi teachers’ motivations regarding their choice of oral corrective feedback (OCF) forms, such as recasts, elicitations, and metalinguistic feedback in foreign language (FL) contexts. Many previous studies of teachers’ choices of OCF forms, and motivations for these choices, have been conducted in more communicative contexts where recasts are most commonly used, with the aim of keeping the communication going. The current study, in contrast, aims to explore teachers’ choices of, and motivations for, OCF forms in a more accuracy-focused context. The study uses rigorous methods to investigate 207 Saudi teachers’ perceptions of OCF, including 100 classroom observations, and 100 stimulated recall (SR) sessions with 10 teachers to further investigate their choices of, and motivations for, particular types of OCF, with reference to their learners’ uptake. The findings demonstrate that the teachers consider recasts to be the most effective method of correction for their students’ learning, especially in the case of pronunciation errors, in a context where the emphasis is placed on accuracy rather than on maintaining the flow of communication. This is in contrast to previous studies of OCF in more meaning-focused contexts, where recasts were used to maintain the flow of communication. The current study concludes by offering insights into some challenges that teachers in FL contexts might face and suggests some possible implications for teachers’ practice in these contexts.


INTRODUCTION
Much of the existing research on OCF has been conducted in contexts that are primarily communicative-based (Brown, 2016;Yüksel, Soruç, & McKinley, 2021) and findings have shown that recasts were the most common type of OCF used, although learners are usually less aware of these in comparison to other forms of oral feedback (Al-Faki & Siddiek, 2013;Brown, 2016;Ellis, Loewen & Erlam 2006;Kamyia, 2014;Lee, 2013;Lyster & Ranta, 1997;Roothooft, 2014;Safari, 2013;Yoshida, 2008).Recasts have been used to keep the conversation flowing and avoid triggering negative feelings among learners that may be associated with other forms of OCF.It appears that the recurrent use of recasts, particularly as an oral corrective technique, and instructors' motives for providing various types of OCF in more form-based contexts, has not been investigated in much depth to date.This study therefore aims to explore teachers' views on, and choices of, various OCF forms in the context of Saudi high schools, where the focus is more on grammatical forms and translation.
A combination of rigorous qualitative and quantitative tools was used to gain a deeper understanding of instructors' S: I goed to the park yesterday.T: What? (Or, Pardon?) 2. Meta-linguistic clues: Gives technical linguistic information or clues about the error without explicitly providing the correct answer to elicit the answer from students.
S: I goed to the park yesterday.T: In the past tense 'go' is an irregular verb.
3. Elicitation + prompt: Prompts the student to self-correct by pausing with intonation so the student can fill in the correct word or phrase.
S: I goed to the park yesterday.T: I...?

Elicitation + enhanced prompts:
Involves a request for the student to repeat, correct, or continue.
S: I goed to the park yesterday.T: Can you try to say that again?
5. Repetition + prompt: Repeats the student's error while highlighting the error or mistake by means of emphatic stress.
S: I goed to the park yesterday.T: I goed?

Repetition + enhanced prompt:
Repeats the student's error while highlighting the error or mistake by means of emphatic stress, in addition to verbal prompt.
S: I goed to the park yesterday.T: Are you sure, I goed?

Non-verbal hints or
'paralinguistic signals', as defined by Ellis (2009): This type of feedback includes identification of the errors by using gestures or facial expressions.
S: I goed to the park yesterday.T: (Teacher gives an unhappy questioning look.) Student: The woman found a police on the street.Teacher: Okay, the woman found a police officer.4. Embedded recast + prompt: this is a reformulation of the whole utterance in rising intonation and/or with additional emphasis to prompt students to reply to the correction, and/or paralinguistic signals.For example: Student: The woman found a police on the street.Teacher: The woman found a police officer? 5. Isolated recast + enhanced prompts: defined as a reformulation of the erroneous part of the utterance using rising intonation and/or extra emphasis in addition to oral prompts or explanation (e.g.Does that refer to …?) (p.528).For example: Student: At this time the wallet, the wallet fall, um, fall to the ground.Teacher: Do you mean it fell?6. Embedded recast + enhanced prompts: defined as a reformulation of the whole utterance using rising intonation and/or extra emphasis in addition to oral prompts or explanation (e.g.Does that refer to …?).For example: IJALEL 11(4): [58][59][60][61][62][63][64][65][66][67][68][69] Student: At this time the wallet, the wallet fall, um, fall to the ground.Teacher: Do you mean the wallet fell to the ground?

Previous Research into OCF and Differences with the Current Saudi EFL School Context
Instructors' perceptions of the efficiency of corrective feedback.Previous research into the efficiency of oral forms of correction and instructors' behaviour with regard to their students' oral errors has shown that using more direct forms of OCF is generally regarded as more effective.Nevertheless, studies have suggested that direct forms of OCF can interrupt the flow of communication and/or trigger negative feelings in the recipients (Al-Faki & Siddiek, 2013;Brown, 2016;Ellis, Loewen & Erlam 2006;Kamyia, 2014;Lee, 2013;Lyster & Ranta, 1997;Roothooft, 2014;Safari, 2013;Yoshida, 2008).As explained earlier, this may help to account for the popularity of recasts in contexts with relatively communicative aims, despite recasts being the least explicit type of corrective feedback, but the one most commonly used by instructors.Efficiency of recasting.Although recasts were found to be the most frequently used form of OCF, the justifications for its efficiency are generally unsatisfying (Goo & Mackey, 2013;Lyster & Ranta, 2013).However, the aim of this study is to attain a better understanding of instructors' choices of OCF by exploring their behaviour during classes to investigate which types of corrective feedback are used and why.
Considerable differences between the Saudi EFL context and previous studies in terms of instructional aims in classes in which OCF has been explored.The instructional aims of this research differ quite significantly from those of previous studies, which have mainly stemmed from interaction-based learning theories where maintaining the flow of communication is a key aim of the teaching.In this study, English instruction in the Saudi context predominantly "focus[es] on knowledge transmission; [and] classroom interaction is largely dominated by teachers" (Al-Seghayer, 2014a, p.20).This constitutes a very different context from those of previous studies conducted on OCF, as its main goal is to reveal knowledge explicitly, which takes priority over syntactical rules and practising the language.In addition, it has been suggested that in form-focused contexts, such as in most schools where English is taught as a foreign language (EFL) (Ahangari & Amirzadeh, 2011;Li, 1998), as is the case in this research, providing corrective feedback on students' errors is prioritized for learning improvement purposes (Ahangari & Amirzadeh, 2011;Harmer, 2003;Harmer, 2007).Therefore, instructors are likely to regard the explicit correction of learners' mistakes in these contexts as crucial.Consequently, the way in which mistakes are treated in contexts involving EFL instruction could differ from other contexts, in that correction is likely to be more explicit and direct.
Justifications for using recasts in interaction-based contexts.Previous research (for instance, Kamiya, 2016;Yoshida, 2008;and Roothooft, 2014) appears to show that, according to teachers, recasts are beneficial because they avoid disrupting the conversational interaction and trigger-ing negative feelings in the learners.Additionally, studies have reported that recasting is helpful in terms of saving time and providing corrections when learners were unable to self-correct (Yoshida, 2008).One instructor commented that prioritizing fluency over focusing on form is crucial (Roothooft, 2014, p.71).
Few studies have investigated instructors' cognition of various learner-related variables such as language level, or personality, as perceived by teachers, to gain a better understanding of instructors' justifications for their choice of various OCF forms (e.g., Yoshida, 2008).Furthermore, according to previous research on OCF, instructors indicated a preference for recasts over other forms of OCF.However, these studies were conducted in interactive contexts in which, unlike the context of the current research, the focus is primarily on learners rather than teachers, and fluency is prioritized over form (e.g., Al-Seghayer, 2014a;Alshammari, 2012).Consequently, teachers' cognitions in the context of this study and their reasons for choosing particular types of OCF in their classes, and pupils' requirements, may differ.Therefore, this research set out to conduct several observations followed by stimulated recall sessions (SRS) to explore instructors' motivations for their choice of oral corrective feedback, in relation to their students and the types of mistakes treated.Additionally, surveys were carried out, which included explanations accompanied by short videos for each type of oral correction to make sure that participants comprehended the form of OCF that they were being asked about.
The lack of student involvement in FL contexts in this study vs past research on oral correction.The limited amount of learner interaction that takes place is one of the most noticeable characteristics of some FL instructional settings which tend to be teacher-centred rather than student-centred.Therefore, students talk to a lesser extent and, consequently, are likely to make fewer oral errors.Another feature of these contexts is the focus on form, that is, on teaching syntactical rules and prioritizing accuracy over fluency, unlike previous studies that have been conducted in interactive contexts, which are more learner-centred and involve learners being prompted to engage in interactions as a way of learning the target language, an approach which stems from the communicative language teaching (CLT) method (Harmer, 2003;Harmer, 2007).This approach gives students multiple opportunities to correct oral mistakes and make amendments while conversing with each other.
Nevertheless, there are various reasons that could prevent or discourage students in some FL instructional settings from participating; for instance, a lack of confidence about talking in front of others, concerns about making mistakes and/or receiving unpleasant feedback, language deficiencies, and a lack of preparation, as argued in Hamouda's (2013) survey-based research involving approximately 160 undergraduates in Qassim.Knowledge about the quantity of interactions and the opportunities for mistakes to be corrected in these FL settings is limited.
Summary.In short, the instructional setting in this research is form-focused and involves a significant amount of practis-Teachers' Perceptions of Oral Corrective Feedback in Form-focused Language Classrooms: Why do they Correct the Way they do?61 ing to allow learners to gradually increase their proficiency, which is in line with the skill acquisition theory (SAT) (De-Keyser, 2007).This teaching method tends to be characterised by less communication and fewer mistakes, and possibly by dissimilar types of OCF and justifications for choosing these types of oral correction as well.As discussed previously, the main justification given by teachers in previous research for choosing recasting, namely, to maintain the flow of interaction, may not apply to the current research context.Therefore, recasting might not be popular in the context of this research as it involves a limited amount of communication.This research aims to explore teachers' choices of different forms of OCF, and their justifications for doing so in an instructional setting in which there is less communication to interrupt.

The Context of the Current Study
Teaching method.Despite the English curriculum having undergone many changes designed to promote the teaching of English in Saudi schools, the results have been underwhelming in terms of students' language proficiency (Al-Seghayer, 2005, 2014a, 2014b).Possible explanations for this include instructors having to abide by the curricula set out by the Saudi Ministry of Education (Allehyani, Burnapp, & Wilson, 2017), as well as insufficient teaching time, and the excessive number of students per class (Al-Seghayer, 2014a;Alshammari, 2012;Li, 1998).This has resulted in a lack of activities involving oral English skills in these classes.Furthermore, although the main purpose of the English curriculum is to improve learners' interactive skills such as listening and speaking (Allehyani, Burnapp & Wilson, 2017;Faruk, 2014), it has been found that instruction in these settings is primarily based on syntactical rules and translation (Al-Seghayer, 2014a;Alshammari, 2012;Hamouda, 2013;Harmer, 2003;Harmer, 2007;Li, 1998).
Test.The fact that English tests are mainly grammar-based could be one important reason why teaching grammar rules is prioritised over other skills in these contexts (Alharbi, 2015;Alshammari, 2012;Li, 1998).
Additionally, based on data obtained in previous studies, it has been argued that students' involvement is limited in these contexts (Al-Seghayer, 2014a;Alshammari, 2012;Hamouda, 2013).Consequently, students tend to only participate orally to respond to their teachers' prompts and often with brief answers such as providing the present continuous form of a particular verb.This approach is largely teacher-centred as teachers are considered the main source of information and learners are essentially seen as passive (Alharbi, 2015;Harmer, 2003).
To sum up, in some FL contexts such as that of the current study, it has been shown that, for various reasons, including instructors being compelled to adhere to the prescribed curricula, insufficient time, the large number of learners per classroom, and exams being mainly grammar-focused, learners have very limited opportunities to demonstrate their interactive competence (Al-Seghayer, 2014a; Alshammari, 2013;Li, 1998).

Participants
10 Saudi female teachers were the core participants in this study, they each had between 5 and 15 years English teaching experience and were aged between 26 and 40.They participated in 10 interviews and 100 observations, followed by SRSs designed to ask about their thinking with regard to their choices of OCF, taking into consideration students' immediate uptake, and individual variables such as students' perceived language proficiency.These research methods were particularly important in terms of addressing research questions (RQ) 1, 2, & 3.The online surveys, which were distributed to a larger population of 207 of English teachers in high schools throughout Saudi Arabia, were designed to answer RQ 1 & 3.

Data Analysis
The data were analysed using the following methods: A) SPSS for quantitative data: For the questionnaire, data were extracted from Qualtrics for each question (which showed how many participants had answered each question).
The quantitative data from the questionnaires and observation schedules were analysed using SPSS.B) NVivo for qualitative data: The software package, NVivo, was used to analyse the qualitative data: teachers' SR sessions and interviews.This was particularly helpful in terms of coding relevant themes from the 100 SRs.Matrices of codes were created to examine combinations of coding categories.For example, the teachers' choices of oral corrective feedback were analysed according to (as a function of) their reasons for making those choices; the teachers' choices of OCF were also examined in relation to their perceptions of the students' proficiency.

The Aim of the English Teaching Classes
A short summary of their instructional goals, as well as some of the related issues that confronted the core participants, according to their beliefs as expressed in the interviews, will be presented here.This information is crucial in order to comprehend the possible impact of the context on the study's results in relation to OCF.It appeared that what the teachers reported regarding their teaching aims generally corresponded to their actual practices.
Grammar-based context: According to the instructors' interviews, a greater emphasis is placed on teaching grammar than other skills: "Teaching grammar and vocabulary is the most important and then speaking, reading, and writing come after" (Teacher S).
In addition, instructors reported associated challenges that prevented them from achieving some of their goals, for instance undertaking exercises designed to improve their students' communicative competence, such as insufficient time, and concentrated on teaching what was essential for students to pass their exams.For instance, Teacher A1 reported that the tests are mainly grammar-focused and that IJALEL 11(4):58-69 "speaking skills require time and we don't have the time, so we skipped them and focused on the other skills: grammar and reading".
Personal matters: It was discovered that other reasons related to personal circumstances may also have an influence on the instructors' teaching practices, as Teacher A3 illustrated.She explained that, although the course aims to promote communicative skills, she only focused on reading because of a stressful situation in her private life relating to the custody of her child after a divorce.
Learners' language level: Teachers F and L reported that because learners had limited language proficiency, it was too challenging to make them interact in English.Therefore, they were mostly encouraged to take it in turns to read dialogues from their textbooks: "they are not yet ready to make conversation in the target language because of their language deficiency, and lack of vocabulary, so we allowed them to read the conversations from their students' books" (Teacher F).
There appeared to be a possible lack of comprehension of the fundamentals of the communicative approach amongst some of the participants, as they justified omitting interactive exercises from their classes due to learners' language deficiencies.

Teachers' Views Regarding OCF According to the Interview Data
This section presents the interview results regarding teachers' preferences for more explicit or less explicit types of OCF and their reasons for these choices.It then discusses the effects of various learner-related variables such as the learners' perceived language levels, and the types of mistakes they make.

Implicit vs. explicit types of OCF
More salient corrections: All the teachers reported in the interviews that the more salient the OCF is, the more beneficial its effects in terms of students' learning.This coincided with the survey results which indicated that teachers prefer more explicit forms of OCF compared to implicit ones.
Nevertheless, some justifications were given for using less salient types of OCF, as shown in Table 4.1.
Less salient corrections: 7 out of 10 of the teachers interviewed reported that they used less explicit forms of recasting.Some of the reasons they gave for doing so are shown in Table 4.1.By contrast, few of the participants claimed to use implicit OCF because it is ineffective in terms of their students' learning.

Error type
The participants reported that they believe there is a connection between their OCF patterns and the type of oral mistakes that learners make; for instance, they often treat their students' erroneous utterances with recasts, as they felt that doing so would help learners to understand what the correct pronunciation should be.However, they used correction with explanations to correct learners' syntactical mistakes.About half of the participants confirmed that there is a connection between irregular or regular errors and the type of OCF used, as the following example from Teacher S illustrates: "I use more implicit correction, such as recasts without indication, for regular errors because I have explained it previously, but I need to use more explicit correction with irregular errors".Meanwhile, instructor N explained that she used "more explicit correction with major errors and less explicit correction such as recasts without prompt for minor errors".Furthermore, they all agreed that their usage of non-verbal gestures was beneficial for treating the mistakes learners made when translating new vocabulary from English into their first language (L1).However, we will see later whether these results were in line with the observations of their actual practices.

Students' perceived language proficiency
The instructors were also asked whether there was a relationship between the types of correction they used and their learners' perceived language level, to which most of them responded that they were related.For example, instructors would tend to elicit the corrections from highly proficient students, whereas they would provide explanations to students who they perceived as having low levels of proficiency.One teacher stated that she would use recasts with highly proficient students but provide explanations for those whose language proficiency was more limited.However, only two participants believed that there was no connection between their OCF patterns and their learners' language level.

Summary
In the next section, firstly, the data produced by the current study are summarised from the interviews and the survey results regarding instructors' cognition of oral corrective feedback, taking into consideration factors such as different types of mistakes, in order to address RQ1, 'What are teachers' attitudes towards different types of oral corrective feedback?'.
Secondly, the remaining research questions are addressed.
Teachers' Perceptions of Oral Corrective Feedback in Form-focused Language Classrooms: Why do they Correct the Way they do?63 Teachers' attitudes to OCF Importance of spoken correction: The data from the surveys showed that the majority of teachers believed OCF to be crucial , which corresponds with the findings of the interviews.Additionally, many instructors indicated a prefer-ence for correcting errors as soon as they occur or once the learner has finished speaking, either at the end of the exercise or the end of the class.Most of the participants seemed to fa-vour more traditional methods of teaching, similarly, to formbased teaching in that correcting their learners' oral errors is seen as a greater priority than maintaining the flow of interaction, as was the case with previous studies set in interactive contexts.
Sources of OCF: According to the survey results, the majority of participants believed self-correction by learners to be their preferred option, which coincided with their preference for "elicitation with enhanced prompts" as their second most favoured type of OCF.
Explicit over implicit OCF: The most important finding from the surveys was that embedded/isolated recasts, and elicitations with enhanced prompts were reported to be the most efficient forms of oral correction, which suggests that the more salient the type of OCF the more beneficial it was believed to be.More implicit forms of OCF, such as embedded/isolated recasts without prompts, were the least favoured ones, a finding that coincided with the interview data which suggested that participants preferred more salient types of OCF to implicit forms.
Nevertheless, the data from the participating instructors revealed two important factors that might have an impact on their oral correction patterns which are debated in the following sections.

The role that learners' perceived language level plays in teachers' OCF choices
According to the interview data, most of the instructors believed that there is a relationship between their learners' perceived language level and their oral correction patterns.The majority of participants revealed that they use elicitation with highly proficient learners but provide explanations for those learners whose language proficiency is more limited.However, the results of the observations will help to provides further insights into whether their views correspond to their actual behaviour in the classroom.

The role that the type of errors plays in teachers' beliefs
The targeted language that was corrected had an influence on the instructors' opinions.For instance, all of the core participants confirmed that the type of error that their learners committed was related to their choice of OCF techniques.All ten of the teachers stated that they provided recasts to correct their learners' pronunciation errors.In addition, all ten core participants also reported that they preferred giving non-verbal hints with regard to their learners' errors in translation.

Teachers' actual behaviour and students' uptake with regard to OCF
This section presents the results of the 100 class observations followed by the SRSs in order to answer RQ2, 'what types of oral error correction do teachers use and why?'.This question, and its related sub-questions, explores the types of OCF chosen by the instructors and their reasons for those choices, taking into consideration students' responses to their corrections and other variables such as their students' perceived language proficiency and error type.The main findings from the observations and SRSs are summarised below and briefly discussed in relation to the previous data obtained from the interviews and questionnaires about teachers' perceptions of OCF.
• Recasting with prompts was found to be the most frequently used type of oral correction; by contrast, there was very limited use of elicitation.
• This finding broadly corresponds to the data from the teachers' interviews, which confirmed that they favour using recasts with prompts because they believe this to be the most efficient type of OCF.However, it conflicted with their second preference, which was for elicitations with enhanced prompts.• Uptake occurs after the majority of OCF provided by the teachers.
• This result also coincided with the participants' reported expectations that most of their OCF involving recasts would be noticed by their learners.• A clear connection was also found between the type of errors that the learners made and the teachers' OCF patterns in the classes.For instance, they tended to use recasting to correct their students' pronunciation errors.
• This generally concurred with the data from most of the interviews.• There was little evidence to suggest that instructors' oral corrections were related to their learners' perceived language level.
• This result diverged from the majority of the instructors' self-reports; according to the interview data, they believed that there is a connection between the learners' perceived language level and the type of OCF provided.This could be because it is not easy for instructors to ascertain a student's language proficiency level quickly and decide on the appropriate form of OCF.Another possible explanation is that most of the OCF provided were recasts, designed to treat phonological errors.

64
IJALEL 11(4):58-69 • There was limited evidence to support the existence of a connection between instructors' patterns of correction and the amount of teaching experience they had.However, generally instructors who had been teaching for longer used more elicitation to correct their students' errors, although their use of more salient recasting was also higher than those with less experience.Nevertheless, this result suggested that more research is required involving a larger number of participant observations.

Summary
This section draws on the main findings from all the data gathered from the observations, SRSs, interviews and questionnaires combined, regarding instructors' perceptions compared to their actual practices in the classroom.The key results in terms of instructors' beliefs and their actual choices of OCF in the classroom are highlighted, regarding the goal of their lessons, their oral correction patterns, and their perceptions of whether learners noticed their corrections.
Classes' instructional goals, communication, quantity of oral corrections.It was noticed that there were some differences between the instructors' views about the goal of their English lessons and their actual practices in the classroom.For instance, although the majority of participants cited speaking skills as the main goal of their lessons, which would be expected to involve learners engaging in interactive exercises such as role play, or narration, the observation data revealed that these types of exercises were rarely employed.Instead, their lessons predominantly focused on grammatical rules, translation, and reading.One teacher mentioned asking students to read dialogues from a textbook as a speaking activity, which she justified on the grounds of her students' limited language proficiency.This suggests that some teachers may have limited understanding of speaking activities.Additionally, it may account for why the majority of errors were phonological errors made by learners when reading the dialogue from the prescribed curriculum.
Salience of oral corrections.The data from this study revealed teachers' preference for explicit correction over less explicit forms of correction.In this respect, the observation results, which revealed a high usage of recasts with prompts, concurred with the data from the interviews.
During the interviews, several teachers mentioned the ineffectiveness of less explicit types of OCF such as recasts without prompts, which was in accordance with the observation findings, given that there were very few examples of implicit recasting.The justifications they gave for using less explicit OCF included: to save time, to prevent triggering negative feelings, or using them to treat frequent errors.
Although most of the teachers mentioned in their interviews that they believed there was a connection between their choice of oral correction types and their learners' level of language, the Chi-square test results obtained from the SRSs revealed that this relationship was weak.
Whether Learners' noticed teachers' OCF: the instructors reported in the SRSs that their learners were aware of their OCF most of the time.This was consistent with the observation results in that 65.8% of instructors' OCF were followed by full uptakes on the part of the learners, 3.3% were followed by partial-uptakes, and 9.2% by uptakes by their peers, while only about 21% of corrections were not followed by immediate uptakes.
The relationship between spoken errors and OCF: It was found that there was a connection between instructors' oral correction patterns and the types of errors made by learners.The Chi-square test (Cramer's V = 0.313) results indicated a strong relationship between the teachers' choice of feedback and their students' types of errors.The ma-jority of recasts with prompts and recasts without prompts were provided to treat their learners' pronunciation errors, whereas approximately half of recasts with explanation were provided to address their syntactical errors, two thirds of the metalinguistic feedback was intended to correct syntactical errors, and most elicitations were also provided to treat syntactical errors.These findings aligned with the instructors' perceptions, according to the interview results in which the 10 core participants confirmed that their choice of OCF form was associated with error types and that they mainly chose recasting to correct their learners' pronunciation errors, and explanatory feedback with regard to their syntactic errors.

Summary of Key Findings
The key results of the current study can be summarised as follows: • According to the observation results, it was found that there were few cases of spoken errors per lesson, which shows that the form-based context of the current study was influential.In other words, these lessons were teacher-centred with learners mostly play-ing the part of passive listeners, and teaching being fo-cused mainly on grammar, translation and reading.• All the interview participants mentioned certain fundamental challenges that they faced, which could help to explain some of the mismatches between what they claimed to be their preferred approach and their instructional goals, and their actual behaviour in the classes.These challenges included a lack of time, and examinations being mainly grammar-focused, a finding which has been reported by other studies conducted in similar contexts (see e.g., Li, 1998).• The results of the questionnaires revealed that embedded-recasting with enhanced prompts was the most popular oral correction method; which concurred with the results of the observations, namely that most of the teachers used recasting.This finding was in line with previous research conducted in interactive settings (see e.g., Al-Faki & Siddiek, 2013;Ellis, Loewen & Erlam 2006;Kamyia, 2014;Lee, 2013;Lyster & Ranta, 1997;Roothooft, 2014;Safari, 2013;Yoshida, 2008).• According to the survey results, teachers believed elicitation with enhanced prompts to be the second most beneficial OCF technique.However, this finding does not align with the observations as very few cases of elicitations were witnessed.• Most of the OCF were followed by learners' immediate uptakes.The teachers mentioned in the SRSs Teachers' Perceptions of Oral Corrective Feedback in Form-focused Language Classrooms: Why do they Correct the Way they do?65 that they thought their learners were aware of most of their OCF because these corrections were generally followed by uptakes.This finding was also supported by the interview data, in which instructors reported that the majority of their OCF was noticed by their learners.• The SRSs data revealed that most teachers used recasting because they believed it to be an efficient way of developing their students' learning, especially in the case of phonological errors, which differs from the findings of previous research conducted in communicative settings where recasts were provided to keep the interaction flowing and to avoid embarrassing learners (see e.g.Kamyia, 2014).• The majority of errors were pronunciation errors and these were treated by recasting.This finding conflicted with Brown's (2016) review which revealed that syntactical errors were the main type of errors produced in form-based contexts.However, it is difficult to make any generalisations given that only two form-based studies were reviewed in this research.• To conclude, it appears that some of the observation results contradicted with the findings of the interviews, while others aligned.For example, in contrast to the interview findings, the observations suggested there was no connection between teachers' choices of OCF and their perceptions of their learners' language level.This suggests that it might be difficult for instructors to make quick judgments based on perceptions of their learners' perceived language skills, for example, to employ unplanned aspects of instruction such as issuing oral corrections (Basturkmen, 2012).
To conclude, the following section sums up the key data with reference to past research.It then discusses the limitations of the current study and makes some recommendations for future studies and instructors' classroom practice.

GENERAL CONCLUSIONS
This section discusses the main results of the study in light of the literature to date and will be divided according to the research questions.
RQ1.What are teachers' attitudes towards different types of oral corrective feedback?
The current research produced crucial results regarding the matches and mismatches between the participants' perceptions and their behaviour in classes, thus building on the findings of past research (e.g., Basturkmen, 2012).One example of correspondence between their perceptions and practice was that, in the surveys and interviews, the majority of the instructors emphasised the importance of treating all their learners' errors, which was in line with the core instructors' actual practices in the classroom, and was also in accord with the findings of some past studies such as that by Feryok (2008).
RQ2.Which types of oral error correction do teachers use and why?
All the results of the current study obtained from various quantitative and qualitative sources revealed that recasting was the most frequently used form of oral correction in classes.This finding coincided with those of most previous studies conducted in different educational contexts, although recasting was regarded as the least efficient form of oral correction in terms of eliciting students' uptake and with regard to their learning (Al-Faki & Siddiek, 2013;Ahangari & Amirzadeh, 2011;Brown, 2016;Ellis, Loewen, & Erlam 2006;Kamyia, 2014;Lee, 2013;Lyster & Ranta, 1997;Roothooft, 2014;Safari, 2013;Yoshida, 2008).
Nevertheless, the justifications given by teachers who participated in the current study differed from those cited in past studies.The main justification for using recasts was because instructors consider recasting to be very efficacious in terms of students' learning.This was in contrast to past research which found that recasting was mainly used by instructors to avoid interrupting their students' interactions, or embarrassing them (e.g Yoshida, 2008).It appears that recasting might therefore be more effective in contexts that are primarily form-based than in meaning-based contexts, mainly because, in form-based contexts, recasts are more likely to be understood as an oral correction tool rather than a communicative strategy.As Kim and Han (2007) pointed out, their findings revealed that recasting for the purpose of correction in form-based contexts was more beneficial with regard to learning the recast words than using recasting for communicative purposes.
RQ2.1.How do learners respond to these different types of OCF in terms of correcting their productions?
The final key example of correspondence between perceptions and practice that is worth highlighting is that most of the recasts used were followed by learners' uptake (repeating the corrected forms), which was consistent with the instructors' views, as most of them stated that they chose to use recasts because these were very effective in terms of improving students' learning.This finding was corroborated by the SRSs results which showed that the students were aware of most of their OCF, as evidenced by their uptake, as well as the instructors' belief that their students were aware of their corrections.
However, the results relating to the efficiency of recasting, as self-reported by teachers and observed from the number of uptakes, contradicted the findings from past research which was conducted in more meaning-based contexts.Previous studies did not perceive recasts as particularly beneficial in terms of their students' learning and pointed out that students may not notice them, but that teachers used them to maintain the flow of communication.Therefore, it was suggested that the main explanation for such variations in comparison with the results of past studies could be because of the difference in context.The setting for the current study was more form-focused, which meant that recasting was more likely to be noticed and comprehended as OCF.
RQ2.2.How do the following factors influence teachers' use of oral error correction: RQ2.2.1.Teachers' perceptions of learners' proficiency?Nonetheless, some contradictions were observed between instructors' beliefs and their behaviour in classes.The most important example of this was that most instructors expressed the view that there was a relationship between their choice of oral correction type and their perceptions of their learners' language level, but that was not supported by the observations or the SRSs.This might be because it is hard for instructors to make instant decisions regarding their perceptions of learners' language levels and what form of oral correction would be best suited to addressing the errors they make using unplanned features of instructors' teaching such as OCF.
RQ2.2.2.Type of language errors?Another significant example was that the majority of instructors emphasised their preference for more salient recasts which coincided with their choice of this type of correction in class, where most of the recasts used were recasts with prompts, such as isolated recasts with prompts, which served to increase their explicitness as an OCF technique.Additionally, the core teachers emphasised that recasting was particularly beneficial for correcting their learners' pronunciation errors, which was in line with their actual behaviour in that pronunciation errors were treated by recasting.RQ2.2.3 Teachers' length of experience?Teachers with more experience tended to use less recasting and more elicitations than those who had less experi- ence.It should be noted, however, that it is difficult to make this type of generalisation from a study involving only 10 instructors, and so further research with a larger number of instructors with various different lengths of teaching experience could help to achieve a better understanding of the connection between the amount of experience instructors have and their behaviour regarding oral correction.

Limitations of the Study
As this research focused on teachers' perceptions and their actual practices with regard to OCF, this might have resulted in the limitations that are explored below.
As the purpose of this study was to gain deeper insights into instructors' beliefs regarding types of oral correction, by combining vigorous quantitative and qualitative data, learners' perceptions were not examined.
The efficiency of various OCF forms with regard to students' learning was not tested.However, students' immediate uptake was interpreted as evidence that oral corrections were beneficial, but it is difficult to make claims purely on that basis.Goo and Mackey (2013) pointed out that various forms of OCF require different responses in terms of uptake, and so rates of uptake are not a sure sign of learning.For instance, uptakes by learners are optional whereas elicitation prompts students to respond.To fully assess the impact of oral correction would require data on long-term learning to be investigated, which cannot be achieved within the scope of this study.
Furthermore, the current study did not provide an individual measurement with which to assess students' language proficiency.Originally, it had been planned to use the X-lex vocabulary test but this was abandoned as a result of time limitations and the need to focus more on instructors' perceptions of their learners' language level which was more relevant to exploring instructors' justifications for using OCF.
Another key limitation was the recording quality which sometimes made it hard to hear learners' mistakes; however, the observation checklists helped to mitigate this limitation to an extent.
Learners' salience is another crucial issue that needed further exploration but, because it was not systematically recorded, and because the current study focused mainly on OCF forms and learners' errors, it was omitted.Hence, further research is required to investigate this aspect via more constructed observations.Furthermore, as the majority of errors were phonological errors, that might have had an impact on the choice of OCF forms, most of which were recasts, it was difficult to draw generalisations regarding which form of oral correction students were most aware of.Therefore, more studies are needed to investigate a broader range of other types of error in these contexts.

Contribution to Knowledge and Recommendations for Future Research
The current research took into consideration gaps in the previous studies combined with vigorous techniques for gain-ing a deeper comprehension of instructors' perceptions and behaviour regarding OCF in the Saudi context, which could subsequently be applied to similar teaching contexts.It produced key findings regarding the influence of context on the amount and type of learners' errors and instructors' justifications for making specific choices about different types of OCF, and instructors' perceptions of whether their learners were aware of their corrections, as well as the influence of various other aspects such as what effect the length of their teaching experience had on their behaviour relating to OCF.Acquiring insights into these matters could be useful for enhancing our understanding of instructors' views and behaviour regarding OCF in the context of the current study, as well as in other similar form-focused educational settings.
This research highlighted the influence of the nature of FL contexts on teachers' choice of OCF.These type of educational settings share certain aspects, such as the teacher being the information provider, learners having limited involvement, and minimal chances of casual oral communication.These features could constitute the fundamental reason for the small number of mistakes made in classes.The lessons were mainly focused on reading and teaching syntactic skills, which could explain why most errors were pronunciation errors.The current study contradicted the findings of past studies which indicated that syntactic errors comprised the majority of errors in primarily form-based contexts (see e.g., Brown, 2016).Nevertheless, it was acknowledged by Brown that it was hard to make such claims due to the very limited volume of research carried out in predominantly form-based teaching settings.Therefore, more research, in addition to the current research, could be beneficial in order to increase our understanding of the influence of different contexts on types of errors and choices of OCF.
This research produced significant results regarding teachers' views about oral correction.Recasting was the most common form of OCF, which was in line with most past research examining more meaning-based teaching approaches (see e.g., Al-Faki & Siddiek, 2013;Ahangari & Amirzadeh, 2011;Brown, 2016;Ellis, Loewen, & Erlam 2006;Kamyia, 2014;Lee, 2013;Lyster & Ranta, 1997;Roothooft, 2014;Safari, 2013;Yoshida, 2008).However, while the results of previous studies showed recasts to be the least beneficial form of OCF in comparison to elicitation (see e.g., Ellis, 2009;Ellis, Loewen, & Erlam, 2006;Lyster, 2004;Lyster & Saito, 2010), and mainly used to maintain the flow of communication and avoiding triggering negative emotions (Kamiya, 2016;Roothooft 2014;Yoshida, 2008), instructors who participated in this research found recasting very useful for promoting their students' learning, especially in the case of pronunciation errors.It should, however, be noted that participants in this educational form-focused context used salient recasting, such as stressing the reformulated word, which is a key difference from the majority of past research in which recasting was possibly less salient.
Another key outcome suggested that uptakes were not the only sign that instructors could use to judge whether their learners were aware of OCF.For example, it was mentioned by some instructors that when they felt that their OCF was IJALEL 11(4):58-69 obvious, that constituted sufficient proof for them that their learners were aware of the OCF.This finding was in line with Mackey and Philp (1998) who found that the absence of uptakes does not mean that students failed to notice corrections.The finding that recasting is beneficial for learning did not concur with past studies' claims that recasting is more likely to be understood as repeating (i.e. as a communication tool) instead of OCF.Nevertheless, further studies on students' perceptions are recommended to confirm or deny this assertion.
Furthermore, this research revealed that instructors had a general tendency to use less recasting and more elicitations when they had more extensive teaching experience than those who had less experience.Nonetheless, it is difficult to make this type of generalisation from a study involving only 10 instructors, and so further research with a larger number of instructors with various different lengths of teaching experience could help to achieve a better understanding of the connection between the amount of experience instructors have and their behaviour regarding oral correction.

Implications for Teacher Practice
By using various vigorous research methods, namely 100 observations, followed by 100 SRSs, and over 200 surveys, the current research provided insights into instructors' thinking regarding their selection of OCF, and the factors that had an impact on these selections.This offered insights into the nature of form-based contexts of the type generally found in Saudi high schools.
The data generated by this research could assist instructors and policymakers in taking into account the potential influence of limited class time, the shortage of teacher-training workshops, deficiencies in teachers' comprehension of authentic speaking exercises (rather than reading written texts aloud), the positioning of learners mainly as passive listeners, and the grammar-based tests used to assess learning.These aspects could help to explain the lack of oral communication that takes place in classes, which resulted in a lower quantity of oral errors and oral corrections.These challenges could also be faced by other instructors in similar FL contexts, and therefore this research could have inferences that extend beyond the Saudi secondary school environment in which it was conducted.
It is possible that the quantity of spoken communication and, consequently, useful OCF that could follow it, might be developed and expanded by investigating the impacts of changes in policies and practices.These modifications could help to determine the goals and content of tests designed to assess all skills equally.They could also have an effect on the administering of teacher training workshops, in particular with regard to unplanned features of their instruction such as oral correction in FL contexts.

ACKNOWLEDGMENTS
A huge thank you to my family and friends, who all supported me during my PhD studies.Special thanks for my lovely kids, and for the University of Hail for giving me this incredible opportunity, and to all my research participants; without them, I would not have been able to carry out my research.I would especially like to thank my participants for their incredible support and patience, which has enabled me to complete my research successfully.
Special thanks must go to my research supervisor, Professor Emma Marsden, for her invaluable guidance, support, and patience.

Figure 4 . 1 .
Figure 4.1.Teachers' responses to the question "How do you rate this type of spoken error correction: "Elicitation with enhanced prompt"?(Video clip in questionnaire).[183]

Table 2 .
1. Classifications and definitions used in the current study for elicitation/prompts

Table 4 .
1. Teachers' justifications for choosing implicit forms of correction (Interview data)

Table 4 .
3. How students responded to teachers' corrections regarding the type of correction (Observation data)

Table 4 .
4.Percentage (number)of correlation between the type of correction and the type of error (Observation data)