Instructed Second Language Pragmatics for the Speech Act of Apology in an Iranian EFL Context: A Meta-Analysis

Introduction

A remarkable shift in the area of applied pragmatics and Second Language Acquisition (SLA) has occurred during the last decades. To nullify the pragmatic failure and communication breakdowns, some factors such as pragmatic competence (PC), language skills, and knowledge about norms and conventions of the target language have gained great momentum. Neither knowledge nor skills are enough to avoid misunderstanding and pragmatic failure, so achieving PC can be paramount in the course of language use. Taguchi (2009) defined PC as an “ability to use language appropriately in a social context” (p. 1), and the developments of PC have gained a surge of interest in foreign language education from curriculum designers to language teachers (Kasper & Rose, 2002; Derakhshan, Malmir, & Greenier, 2021). The developments of EFL learners’ PC (Ishihara & Cohen, 2014; McConachy & Hata, 2013) have been covered in the field of Interlanguage Pragmatics (ILP).

Kasper (1997) stated that ILP scrutinizes how L2 pragmatic is utilized and achieved by foreign learners and non-native speakers. ILP also takes into account the learners’ performance and understanding of the speech acts, which are the pivotal subcategories of pragmatics. As Bardovi-Harlig (2010) explained, both language use and structure mitigate language learning in ILP, and a learner must know how to apologize, refuse, or even complain in a real situation. No matter how advanced the learners are, they may commit some pragmatic errors because their knowledge of pragmatic is limited, and they cannot interpret the intended meanings to say the necessary act. According to Blum-Kulka and Olshtain (1984), some advanced learners still face many difficulties while making requests or performing apologies properly, and it heightens the fact that grammatical competence is not adequate for meaningful communication.

It has been postulated that intervention can be the only panacea that ameliorates the hardships of meaningful communication in learning pragmatics (Derakhshan, 2014; Rose 2005; Shakki, Naeini, Mazandarani, & Derakhshan, 2020). Considering the instruction of apology, which is one of the most influential speech acts, there have been a large number of papers corroborating the importance and significant effect of teaching pragmatics (Birjandi & Derakhshan, 2014; Derakhshan & Arabmofrad, 2018; Derakhshan & Shakki, 2021; Eslami & Eslami, 2008; Malmir, 2020; Malmir & Mazloom, 2021; Mirzaee & Esmaeili, 2013; Olshtain & Cohen, 1989), although there is no systematic review or meta-analysis on the effect of instruction for the speech act of apology around the world. Because meta-analyses show superiority over other research methods (Norris & Ortega, 2000), in recent years, several studies have been conducted on the effect of instruction on various features of pragmatics (Badjadi, 2016; Derakhshan & Shakki, 2021; Jeon & Kaya, 2006; Plonsky & Zhuang, 2019; Yousefi & Nassaji, 2019). Taking an EFL context into consideration, and knowing the prominence of the speech act of apology in teaching and learning pragmatics, the purpose of the study is to investigate the overall effectiveness of instruction on apology and the moderator variables, such as research design and treatment type which are involved in an Iranian EFL context. 

Literature Review

Theoretical Background

Noticing Hypothesis

Schmidt (2001) pinpointed that if students are provided with enough input or exposure of any kind, but they are not made aware of the pragmalinguistic and sociopragmatic features of the input, they cannot develop their ILP. Schmidt (1993, 2001) cogently propounded that according to psycholinguistic theory and research on second language learning, for input to be acquisitionally germane, it needs to be noticed or detected under attention. Schmidt (2001) elaborated that a whole awareness of second language input is not eloquent, but attention has to be allotted to specific learning objects. He further illuminated that to learn pragmatics, learners should acquire not only linguistic forms of the interlocutors but also the contextual and relevant social features with which they are associated. Schmidt (1995) also made a demarcation between understanding and noticing. Although he defined noticing as the surface level of learning in which conscious registration happens, he believed that understanding purports a deeper level of learning by focusing on some general principle, pattern, or rule. Like other theoretical constructs, noticing hypothesis, which can be one of the most frequently referenced theories in second language pragmatics research, has some implications for pragmatics development and it is divided into two levels, such as explicit and implicit through which researchers and teachers expand learners’ knowledge and is considered as one of the main areas of research in SLA.

Explicit and Implicit Teaching

According to Long’s (1991) view, on the one hand, Focus on Forms (FonFS) resembles a highly metapragmatic, decontextualized, teacher-centered instruction in which explicit awareness has happened, and practice and production tasks have been activated. On the other hand, the Focus on Form (FonF) methods instruct indirectly, and learners gain attention incidentally and receive input without any direct focus implicitly. In the present study, explicit instruction was operationalized as the direct teaching of the speech act of apology to the learners, although the non-explicit group is the one in which teaching apology has been carried out indirectly and implicitly. Motivated by Schmidt’s (1993) noticing hypothesis, the comparison between implicit and explicit teaching methods has been accentuated, and the role of consciousness and attention reveals that explicit explanation is more beneficial than an implicit condition which enhances learning through input exposure and consciousness-raising. According to Taguchi’s research (2019), which is one of the most recent studies in instruction, “effective teaching is closely related to the depth of processing” (p. 7), and she believed that explicit teaching outperforms implicit instruction. Rezvani, Eslami, and Vahid Dastjerdi (2014), for instance, have scrutinized the role of implicit and explicit teaching on Iranian EFL students’ pragmatic development. In doing so, 60 intermediate EFL students were selected from an English language institute in Iran. The researchers divided the participants into two main groups, namely) Implicit Group (IG) and Explicit Group (EG). To assess the participants’ ability to employ suggestions and requests, the researchers distributed a pretest among the participants. Then, both IG and EG were shown recordings of short dialogues, including the speech acts of suggestions and requests. The results of the posttest demonstrated that both implicit and explicit instruction greatly impacted the students’ pragmatic competence.

More recently, Ziafar (2020) explored the effect of implicit, explicit, and contrastive lexical approaches on EFL students’ pragmatic competence. In doing so, 63 EFL students were selected and the researcher divided the participants into three treatment groups. Some episodes of a TV sitcom were opted as the primary source of pragmatic instruction. The results of the posttest showed that explicit instruction notably improved students’ pragmatic competence and outperformed the implicit group. In a similar vein, Derakhshan and Shakki (2020) attempted to probe the impact of implicit and explicit metapragmatic instruction on the Iranian EFL students’ pragmatic comprehension of apology and refusal. To do this end, based on Oxford Quick Placement Test (OQPT) (2004), they selected 49 EFL students and assigned the selected participants into three groups of implicit, explicit, and ‎control. They employed a validated Discourse Completion Test (DCT) with 128 items (8 conversations for the speech act of refusal and the same number for the speech act of apology). The findings of one-way ANOVA revealed that students’ pragmatic comprehension improved, and the explicit group was better than the implicit one.

Instruction of Pragmatics

In order for successful communication to take place, individuals need to know how to get others’ intentions and how to use English appropriately in a new situation, so the need for instruction in pragmatic seems to be urgent (Sánchez-Hernández & Alcón-Soler, 2020). All learners should be instructed about the sociocultural norms to have a better negotiation and to be able to interpret the meaning in the context (McKay, 2018). Instructed second language acquisition (a subfield of second language acquisition) happens as a result of teaching a second language and occurs in the classroom, although it may include some self-studies such as using the target language in everyday life. Guiding and facilitating the process of learning by teachers and materials can be defined as L2 instruction. Not crucially, the rate of L2 acquisition can be increased by assisting learners to achieve high proficiency in the target language, and it shows that instruction is valuable. 

Based on early studies in the 1980s and 1990s, there is a consensus that pragmatics is teachable, and instructed groups often outperformed the non-instructed groups (Kasper & Rose, 1999, 2002). Specifically, in an EFL context like Iran, language learners’ pragmatic awareness must be raised since they do not have access to real situations in English. The only way through which this awareness may be heightened could be the classroom, and the teachers are the ones who can provide suitable conditions. According to Ellis (2005), input is a prerequisite for more opportunities to produce output and better language learning; therefore, the vital role of instruction as a disincentive factor in misunderstanding is undeniable. The purpose of an instructed second language which has its roots in the noticing hypothesis (Schmidt, 2001) is to improve the communicative competence and help the learners to increase their ability in using L2 sociopragmatically and pragmalinguistically appropriate.

Apology

In a face-threatening speech act of apology, which belongs to expressive acts, the speaker/writer identifies the contravention or violation caused by his or her mistake and tries to compensate the relationship with the other interlocutor. The condition may be relatively tense if the infraction is large or if the listener has a higher social status and a more dominant position on the social scale than the speaker (Ishihara & Cohen, 2010). According to Cohen and Olshtain (1981), by apologizing, the speaker is conscious of the fact that a social norm has been violated and that he or she is at least partly responsible for it. As a result, apologies, which are post-event actions, entail face loss for the speaker and assistance for the hearer, while requests, which are pre-event actions, may entail face loss for both parties.

As it was reviewed by Shakki et al., (2020), apology is the second dominant speech act that has been investigated during the last 20 years in Iran, and this is the reason for which we chose this speech act. Many studies have been conducted on the instruction of apology (Bagherkazemi, 2018; Derakshan & Shakki, 2020; Fakher & Panahifar, 2020; Tajeddin, Keshavarz, & Zand-Moghadam, 2012), and all of them believe that apology is improved through instruction, among which more recently is Pourmousavi and Zenouzagh (2020), which scrutinized the effect of individual feedback and teacher’s group on Iranian EFL learners’ learning of apology in letter writing. They have used 32 pre-intermediate learners, and after the OQPT, individual and group feedback were provided for the participants. The results revealed the significance of individual feedback, and it corroborates that teaching pragmatics is highly proposed. Similarly, Derakhshan and Arabmofrad (2018) investigated the effect of video-enhanced input on the comprehension of pragmatics specifically, speech acts of apology, refusal, and request on 69 Iranian EFL students. The participants were randomly grouped into four groups, namely form-search, interactive translation, metapragmatic, and control. They were exposed to 60 video vignettes extracted from various episodes of Friends and Seinfeld sitcoms. Based on the results of DCT, they reported that interactive translation, metapragmatic consciousness-raising, and form-search groups led to the improvement of pragmatic comprehension. Because apology has been approved to be amenable to instruction, the overall effectiveness of teaching on this speech act needs to be investigated.

Related Studies

Some scholars conducted meta-analyses and systematic reviews on pragmatic instruction (Derakhshan & Shakki, 2021; Jeon & Kaya, 2006; Plonsky & Zhuang, 2019; Taguchi, 2015; Yousefi & Nassaji, 2019) and argued that instruction is generally more effective for L2 pragmatic features. Norris and Ortega (2000) was the first meta-analysis done on the effectiveness of L2 instruction, which employed 49 samples published from 1980 to 1998 to work on focus on form and focus on forms studies. They found that the focused L2 instruction and explicit groups are more effective than the other counterpart. More specifically, the first meta-analysis on L2 pragmatics instruction was conducted about 15 years ago by Jeon and Kaya (2006), using 13 studies published prior to 2003. They coded the studies based on their settings, proficiency level, sample size, treatment type, outcome measure, and length of the instruction. The findings illustrated that direct instruction provides a significant difference over no instruction counterpart. They found that explicit teaching yielded larger effect size (d = 0.70 for experimental vs. control group & d = 1.91 for pretest, posttest) than the implicit instruction (d = 0.44 for experimental vs control group & d = 1.01 for pre-test, post-test). Furthermore, the results on the relationship between different instructional methods, length of instruction, and outcome measures, and the effectiveness of L2 pragmatics instruction were not totally convincing. Inconclusive findings and the limited number of studies that were analyzed are the drawbacks of this meta-analysis.

Taguchi (2015) also carried out a study on the development of instructed pragmatics over the past 30 years used 58 instructional intervention studies. She reported that that instruction was more effective than non-instruction. She coded the studies based on the following criteria: (1) a research with pretest, posttest design, (2) enough description of the participants, (3) clear teaching method (explicit & implicit), and (4) outcome of instruction. She found that explicit teaching is typically more beneficial than implicit one, although implicit teaching can be conducive if it involves exercises for noticing and awareness. Interestingly, in some cases, the implicit approach could produce better results than the explicit approach. For example, Li (2012) suggested that inductive and implicit teaching might support some pragmatic features. Her findings are not conclusive enough, and more studies are needed to be justified so that further investigation was proposed to take into account variations and stability in the findings.

Similarly, Badjadi (2016) utilized 24 studies to find the effects of instructional tasks in second language pragmatics related to production and comprehension outcome measures. He coded the studies based on their context, instructional tasks, proficiency level, outcome measures, and methodological nuances. The findings revealed that, in conformity with instructional tasks, production and comprehension mean effect sizes change from small to large. Moreover, meta-pragmatic discussion or provided recasts produced a larger effect size in contrast with other groups. 

Alternatively, Plonsky and Zhuang (2019) utilized a total of 50 studies to answer the following questions: a. what is the overall effectiveness of L2 pragmatics instruction? b. What is the relationship between the effectiveness of pragmatics instruction and the following types of moderating variables: treatment and target features, contextual and learner factors, research and reporting practices, and outcome measures? They gained an effect size of (d = 1.52) which demonstrates the overall effectiveness of instruction. Their findings also supported the previous meta-analyses in which the importance of explicit was accentuated over the implicit one. They found that pragmatics instruction, providing opportunities for practice, was more effective than instruction without opportunities for practice, and longer instruction is better than the other one in general. They also stated that role-plays as free outcome measures produced larger effects than multiple-choice questions which are among controlled outcome measures.

Regarding the effects of corrective feedback and instruction on L2 pragmatics, Yousefi and Nassaji’s (2019) corpus included 39 published studies from 2006 to 2016. They analyzed the studies based on the following coding process: type of instruction, mode of instruction, the durability of instructional effects, length of instruction, and language proficiency. Their results indicated that computer-assisted instruction generated larger effects in comparison with face-to-face instruction. Furthermore, their findings revealed that instruction for L2 comprehension produced a larger effect size than L2 production. It is also mentioned that longer treatments generated better effect sizes than shorter ones. Considering the level of proficiency, intermediate learners had larger effect sizes than advanced and beginners. 

By the same token, the most recent and pioneering meta-analysis, focusing on the Iranian EFL context, was conducted on the instruction of the speech act of request (Derakshan & Shakki, 2021). They coded a total of 37 studies based on the inclusion and exclusion criteria, and 17 primary studies were analyzed according to their age, gender, treatment type, proficiency level, data collection procedure, and research designs. They found that the overall effectiveness of instruction for the speech act of request is (g = 1.48), which is large and positive. Subsequently, regarding the moderator variables, treatment type, the explicit group yielded a larger effect size (g = 1.53) than the implicit one (g = 1.20). Moreover, with respect to gender as another moderator, the male group, produced a larger effect size (g = 3.09) than the female one (g = 1.10).

The abovementioned reviews and meta-analyses precisely have covered various factors in the teachability and effectiveness of pragmatics instruction. Nonetheless, scant attention was devoted to a meta-analysis on the effect of instruction for the speech act of apology, specifically in an EFL context, so the present study aims to broaden our understanding by shedding light on the effectiveness of instruction in L2 pragmatics on the speech act of apology in Iran.

This study aimed to answer the following research questions:

1. What is the overall effectiveness of instruction on apology in Iran?
2. Do research designs and treatment types moderate in teaching apology in an Iranian context?

Method

Data Collection

This meta-analysis is limited to research published in refereed journals during the last 20 years, from 2000 to 2020. Both manual and electronic searches were carried out to find all the dissertations and full-text papers on L2 pragmatics instruction of the speech act of apology in an Iranian context. Some databases in applied linguistics such as Google Scholar, Google, Microsoft Academic Search, Academic Search Complete (EBSCO), Web of Science, Project MUSE, Blackwell Reference Online, PhycINFO, Oxford Journals Digital Archive, Academic Search Premier, Springer Link, Wiley Online Library Researchgate, iSEEK Education, Academic Index, Magiran, Internet Public Library, Linguistics and Language Behaviour Abstracts (LLBA) (ProQuest), Virtual LRC, ProQuest, Oxford Handbooks Online, Noormags, ERIC - Education Resources Information Center, ScienceDirect, Sage Journals Online, and SAGE Knowledge, SID, SAGEResearch Methods Online, RefSeek, Scopus, and Iranian local journals were used to search the relevant studies. Moreover, to have a thorough search, a combination of the keywords (a) ‘pragmatics’, (b) ‘instruction’, (c) ‘teaching’, (d) ‘intervention’, (e) ‘speech act(s)’, (f) ‘apology’, (g) ‘Iran’, and (h) ‘interlanguage pragmatics’ were also utilized while gathering the data. Last but not least, the end references of the related papers were explored so as to find the missing papers on the instruction of apology in Iran. It was also found that there was no dissertation on the effect of instruction for the speech act of apology in an Iranian context, and this may be the first time that a study has been done in that field of research. The collected studies were organized and coded to check which study effects and moderators are suitable to allow for meta-analysis. 

Criteria for Study Inclusion and Exclusion

Each study had to meet the following criteria to be included in our corpus:

First, the studies both in English and Persian published from 2000 to 2020 were included in the present meta-analysis, so the papers earlier than 2000 or later than 2020 were excluded from the study. Second, the instruction of apology in an Iranian context was the main concern of the current study; thus, the papers that have been done in other contexts or on other speech acts were not included in the present study. It is also worth mentioning that only the studies, focusing on apology are included here, and the papers, which included more speech acts besides apology, were excluded, and that is why the number of papers is limited. Third, it had to report sufficient data for calculating the effect size to be used in this study; in other words, papers in which means, standard deviation, sample sizes were missing were also excluded. Finally, out of 31 papers, 12 studies were chosen to be used and analyzed in this meta-analysis.

Study Feature Coding

The coding protocol used in this study is based on the data gathered and categorized from the primary studies. The treatment type and design of the studies are the two variables that were investigated as the moderators of the present study. Considering the treatment type, the studies were coded and operationalized as explicit and non-explicit groups. By explicit group, we mean the studies in which direct instruction has been used, while non-explicit group deals with the studies that implicit instruction or indirect teaching has been conducted. Furthermore, the design of the studies was also coded as experimental and quasi-experimental. Regarding the experimental group, the participants are randomly assigned to either control or treatment groups, while in quasi-experimental, the subjects are based on non-random criteria. The coding scheme was fixed based on the previous meta-analyses which have been carried out and other experts’ and peers’ suggestions and recommendations in L2 pragmatics.

Table 1. Coding Protocol

Analysis

Analyzing the 31 conducted studies in Iran in which apology strategies have been taught, 12 studies were selected to be added in the present meta-analysis. The remaining studies were excluded due to one or more of the following reasons: Other speech acts besides apology (Derakhshan & Arabmofrad, 2018; Derakhshan & Shakki, 2020; Fakher, Vahdany, Jafarigohar, & Soleimani, 2016; Fakher & Panahifar, 2020; Mirzaei & Esmaili 2013; Tajeddin & Bagherkazemi, 2014), missing data (Farrokhi & Atashian, 2013), treatment condition (Birjandi & Pezeshki, 2013; Derakhshan & Eslami, 2015). As a result, the limited number of the papers is actually because of the inclusion and exclusion criteria of the current paper through which just the following studies were used to be checked in the analysis. The studies which have focused only on apology and have experimental or quasi-experimental designs with explicit or non-explicit treatments were chosen.

Table 2. Study Design across Studies

Publication Bias

Publication bias can be considered a threat for meta-analyses since journals tend to publish studies with significant results. Studies with large sample size which lead to small sampling error and higher precision values are placed near the mean effect size and toward the top of the graph (Plonsky & Gass, 2011), though the papers with small sample size and big sampling errors concentrated on the right side of the mean. Generally, two types of analyses have been done to check the presence of publication bias: first of all, the funnel plots provide a clear graphic tool to scrutinize the impact of publication bias on the treatment effects. If availability bias is not present, the plot demonstrates a symmetrical inverted funnel.  As can be seen in Figure 1, the funnel plot is not symmetrical, showing that there is some publication bias in the present meta-analysis. Actually, four studies are missing from the left side of the plot.

Figure 1. Funnel Plot of Precision by Effect Sizes for the Observed and Imputed Studies for L2 Pragmatics Instruction

Second, to address the issue of missing studies and publication bias, the researchers conducted the trim-and-fill analysis, which is a technique to estimate the missing studies and employ a re-computing to adjust the asymmetric funnel plot, developed by Sue Duval and Richard Tweedie (2000).

Table 3. Duval and Tweedie’s Trim and Fill Test of Publication Bias Estimation for L2 Pragmatics Instruction

In Table 3, considering the fixed effect model, the point estimate and 95% confidence interval for the combined studies is 0.99 (0.82, 1.17), while the imputed point estimate is 0.72 (0.56, 0.88) by using Trim and Fill. On the other hand, under the random-effects model, the point estimate and 95% confidence interval for the combined studies is 1.34 (0.80, 1.87) and the imputed point estimate is 0.75 (0.17, 1.33) while using Trim and Fill.  The recomputed combined effect shows a shift from medium effect to small effect due to the impact of publication bias.

Results

Overall Meta-Analysis Results

The first research question aimed at finding the overall effectiveness of instruction on the speech act of apology in L2 pragmatics. Out of 12 original studies, from 2000 to 2020, with 535 participants from various Iranian contexts, 12 effect sizes (Hedges’ g) were collected for the meta-analysis. As depicted in Table 4, the results of the average weighted Hedges’g, the 95% prediction intervals, between-study variance, the Q-test for heterogeneity, the percentage of variation between studies due to heterogeneity rather than sampling error, and the two-tailed test of null, are represented. Based on Plonsky and Oswald (2014), concerning the interpretation of effect sizes in SLA, as a field-specific benchmark, a d value of 0.60 is considered small, 1.00 as medium, and 1.40 as large. In this study, the overall effect size was found to be g = 1.00, with a standard error of 0.09, a z-value for a test of the null of 11.38, a corresponding p-value of less than 0.001 for the fixed model and a mean of 1.34, a standard error of 0.27, a z-value for a test of the null of 4.91 and a corresponding p-value of less than 0.001 for the random model.  For both models, we concluded that the mean effect size was significant.

According to Plonsky and Oswald (2014), since the d value (Hedges’g ) is less than 1.40, for both models, the mean effect size is considered medium. However, the Q statistic on the heterogeneity of effect sizes was 101.92, df = 11, and p <.001. It indicates that all the variance is unlikely to be due to sampling error, and we also conclude that the true effect size is likely to differ from study to study. Therefore, the fixed model is not appropriate and does not match the data. For this reason, we applied the random effect model (Borenstein, Hedges, Higgins, &, Rothstein, 2013).

Table 4. Results of the Univariate Random-Effects Meta-Analyses of the Instruction of Apology on Learning L2 Pragmatics

Note: N= total number of participants, k =number of effect sizes, g= mean weighted effect size in Hedges' g, SE = standard error, CI = confidence interval, Z = Z value, P = P value, Q = Cochran's heterogeneity test; df = degrees of freedom Q-test, τ² = between-study variance; I² = percentage of variation between studies that is due to heterogeneity rather than sampling error. V= variable

Figure 2. Forest Plot of Effect Size (Hedges’g) in Random-Effects Model

Moderator Analyses

The second research question intended to study the moderating factors of the instruction of apology in L2 pragmatics. A meta-regression analysis for each group of moderator variables was conducted independently. Q-Statistic was used to evaluate if a particular variable was a significant moderator. In Table 5, the findings of the meta-regression moderator analyses can be seen for the design and treatment type of the studies.

Table 5. Moderator Analysis of Design and Treatment Type, on L2 Pragmatics Instruction

Note: N = total number of participants, k = number of effect sizes, g = mean weighted effect size in Hedges' g, CI = confidence interval, P = P-value, Qb = Q-between, df = degrees of freedom, I2 = percentage of variation between studies that is due to heterogeneity rather than sampling error, τ2 = between-study variance, R2 = the proportion of the original variance explained by the covariates.

Research Designs

We calculated the moderating effect of the study’s design on the instruction of apology in L2 pragmatics in the analysis. The results, Qb = 5.25, df  = 1, pb = 0.02, τ2 = 0.77, I2 = 89.21, R2 = 0.10, were significant regarding the probable moderating effect of the design of the study on variables. The effect was small for the experimental group (g = 0.92) and the Quasi-experimental group displayed a large and positive effect (g = 2.39). It is obvious that design can be a predictor for teaching apology in an Iranian context.

Figure 3. Moderator Analysis of Designs of the study and L2 Pragmatics Instruction of Apology

Treatment Types

The treatment type was our next moderator, and we included two treatment types, explicit and non-explicit instruction of apology on learning L2 pragmatics. For the explicit group, based on Plonsky and Oswald (2014), field-specific reference for effect size interpretation, the mean effect size (g = 1.25) was found to be positive and medium, although for the non-explicit group, the average effect size (g = 1.44) was positive and large. The results heterogeneity test, Qb = 0.09, df = 1, pb = 0.76, τ2 = 0.77, I2 = 89.21, R2 = 0.00, were not significant. In other words, treatment type was not found to be a significant predicting index for the instruction of apology on learning L2 pragmatics.

Figure 4. Moderator Analysis of Treatment Types and L2 Pragmatics Instruction of Apology

Discussion

Following what Kasper and Rose (2002) claimed about the instruction of pragmatics, some researchers and scholars commenced investigating the effectiveness of instruction on L2 pragmatics (Alcón-Soler & Martı’nez-Flor, 2005; Bardovi-Harlig, 1996; Blyth & Sykes, 2020; Cohen, 2020; Culpeper, Mackey, & Taguchi, 2018; Derakhshan & Eslami, 2015, 2020; Derakhshan & Shakki, 2020; Derakhshan, Shakki, & Sarani, 2020; González-Lloret, 2008; Jeon & Kaya, 2006; Rose & Kasper, 2001; Takahashi, 2010a, 2010b; Taguchi, 2011, 2019). Kasper and Rose (2002) stated that all norms and features of pragmatics, such as speech acts, hedges, address markers, and speech functions, are amenable to instruction, and such a postulation was endorsed by many scholars (Alemi & Haeri, 2020; Bardovi‐Harlig, 2018; Bardovi-Harlig & Vellenga, 2012; Chalak & Abbasi, 2015; Derakhshan & Arabmofrad, 2018; Hassaskhah & Ebrahimi, 2015; Kaivanpanah & Langari, 2020; Khatib & Ahmadi Safa, 2011; Mirzaei & Esmaeili, 2013; Moradian, Asadi, & Azadbakht, 2019; Plonsky & Zhuang, 2019; Samavarchi & Allami, 2012; Shakki et al., 2020; Tajeddin, Keshavarz, & Zand Moghadam, 2012)

Among the speech acts, apology is recognized as the second most researched speech act in Iran (Shakki et al., 2020), and it is a face-saving speech act that plays a special role in human life and aims to maintain harmony between the interlocutors. The instruction of apology has been carried out in a number of studies, and the advantages of teaching apology are highlighted and recommended (Doan, 2019), particularly in Iran (Bagherkazemi, 2018; Derakhshan & Arabmofrad, 2018; Derakhshan & Shakki, 2020; Eslami & Mardani, 2010; Pourmousavi & Zenouzagh, 2020).

Neither in Iran nor in other parts of the world is there a meta-analysis on the effectiveness of instruction for the speech act of apology, and our findings can be justified by the results of other papers which have been done on teaching apology or the instruction of pragmatics. Our findings of the superiority of instruction over non-instruction can be consistent with the most recent following papers. For instance, in line with our study, Derakhshan and Arabmofrad (2018) also found the effectiveness of the instruction not only for the speech act of apology but also for request and refusal speech acts in an Iranian context. Moreover, Derakhshan and Shakki (2020) examined whether the instruction is beneficial for apology and refusal or not. Their findings are also in harmony with what we have found in the present meta-analysis. Similarly, in one of the most recent papers on the effect of instruction on pragmatic awareness, Ziafar (2020) claimed that instruction dramatically enhances the learners’ ability to achieve pragmatic knowledge. In their meta-analysis on the effect of instruction, Plonsky and Zhuang (2019) supported other review articles and meta-analyses, and they reported that instruction provides more opportunities for the learners, and they suggested the researchers that they analyze the research designs in future studies, which is one of our moderator variables and our concern in the present meta-analysis. 

Resonating what Taguchi (2015), Yousefi and Nassaji (2019), and Shakki et al., (2020) stated, the present meta-analysis is in line with the previous studies on the effectiveness of instruction, revealing that the instruction of the speech act of apology has had many benefits for the learners in an Iranian context. The first research question of the current study, dealing with the overall effectiveness of apology instruction, indicated that the speech act of apology lends itself moderately to instruction since the d value (Hedges’g) is less than 1.40, and it demonstrates that the effect size is medium and significant. Similarly, the only meta-analysis (Derakhshan & Shakki, 2021), which was conducted on the speech act of request in an Iranian context, accentuated the effectiveness of instruction for the speech act of request. They also found the medium and significant effect size (g = 1.48) in their analysis. Comparing effect sizes with other meta-analyses, the first meta-analysis conducted by Jeon and Kaya (2006), found a smaller effect size (d = 0.59), and Plonsky and Zhuang (2019) reported a larger effect size (d = 1.52) for this effectiveness. 

Following our second research question, and considering explicit and implicit teaching, some researchers recapitulated that the results of the explicit teaching outweigh those of implicit teaching (Brock & Nagasaka, 2005; Taguchi, 2011, 2015; Vitale, 2009), and it has been confirmed that direct teaching of the speech acts raises the learners’ pragmatic awareness (Culpeper et al., 2018). On the one hand, providing enough input and organized materials may lead to the production of suitable output by means of the language (Taguchi, 2015). On the other hand, Brown (2007) elucidated implicit instruction as a kind of learning without attention and awareness which provides necessary input for pragmatic knowledge and reinforces the learning process.

Taguchi (2015) believed that although explicit teaching outperformed implicit instruction, implicit teaching can be as efficient as explicit just by enhancing learners’ notice of the form and function of the target language. She illuminated that the difference that may exist here lies in the fact that the learners who receive direct instruction may be able to process a deeper level of input than those of indirect instruction. Surprisingly, what was found in the present meta-analysis was a new finding for the instruction of apology in an Iranian context. Treatment type is not a predictor for the effectiveness of teaching apology, and the effect size for the explicit group was medium and positive (g = 1.25), while the implicit group was large (g = 1.44), and it is in line with what Li (2012) stated on the advantages of implicit teaching over the explicit one. The next moderator variable was research design which can be regarded as a predictor for the instruction of apology. For the experimental group, the effect size was small (g = 0.92), and the quasi-experimental group displayed a large and positive effect (g = 2.39), which is in harmony with what Derakhshan and Shakki (2021) found in their study on the effectiveness of instruction for the speech act of request (both groups of studies, the experimental (g = 1.72) and quasi-experimental (g  = 1.40) displayed a large and positive effect), although in their study design was not a predictor for the effectiveness of instruction on request. 

Conclusion

The present meta-analysis wan intended to present a summative description of the empirical studies which have been conducted on the instruction of the speech act of apology in an Iranian EFL context during the last two decades. To this aim, some special methodological moves have been made: firstly, a set of inclusion and exclusion criteria have been defined and based on which the fixed and random effect sizes were calculated; and secondly, we have used Q-tests to check the group differences and find the moderator variables. The purpose of conducting this meta-analysis was that there is no meta-analysis on the effect of instruction for the speech act of apology either in Iran or around the world. Our paper explored the studies which have been carried out on the instruction of apology in an Iranian context to see whether this instruction is effective or not and, if yes, whether research designs and treatment types can make a difference. 

The findings revealed that instruction is effective for the learners who aimed to learn apology, and research design can be a predictor and moderator variable for the instruction of pragmatics. Treatment types (explicit and non-explicit) are also positive, but they are not significant and cannot be considered as a predictor based on the findings. The results of the current meta-analysis may be useful for the researchers who are interested in the instruction of pragmatics for their future studies. Considering the limitations of the present analysis, the following issues are suggested to be scrutinized in future research: (1) the presence of publication bias indicates that more research is needed to be done on the effect of teaching apology in Iran; (2) Inasmuch as the fact that not many studies have been done on the effect of such moderators as age, gender, proficiency, context, assessment types, and language background, more studies are suggested to be undertaken on these variables; (3) There is also a dearth of research on other speech acts like refusal, compliment, complaint, etc. and other meta-analyses can be conducted on those speech acts.