1. Introduction
In the field of instructed second language acquisition (ISLA), peer interaction has been investigated as one of the two types of interaction—the other being teacher-student interaction—that drive second language (L2) learning forward in the classroom (Philp, Adams, & Iwashita, Reference Philp, Adams and Iwashita2014; Adams & Oliver, Reference Adams and Oliver2019). Drawing on cognitively- and socially-oriented theories, this research has shown that when L2 learners interact with each other, they may engage in behaviors that are conducive to L2 learning (Sato & Ballinger, Reference Sato, Ballinger, Sato and Ballinger2016; Loewen & Sato, Reference Loewen and Sato2018). In particular, research has focused on the ways in which L2 learners temporarily shift their attention to linguistic issues during communicative interaction (Sato, Reference Sato, Nassaji and Kartchava2017b). From a pedagogical perspective, teachers spend a significant amount of time on activities in which L2 learners complete a task together. However, survey research has shown that teachers are often unsure of the effectiveness of peer interaction activities (Carless, Reference Carless2003; McDonough & Chaikitmongkol, Reference McDonough and Chaikitmongkol2007; Butler, Reference Butler2011; East, Reference East2017; Sato & Oyanedel, Reference Sato and Oyanedel2019). Teachers have reported that their students are often (a) disengaged with tasks, (b) unable to form collaborative relationships or to provide feedback to each other, and (c) reliant on their first language to complete the task. Taken together, research suggests that peer interaction can be effective, yet its effectiveness in the real-world setting can be affected by learner attitudes toward peer interaction tasks and/or their conversation partners (Sato, Reference Sato2017a).
In order to further our understanding of the way in which peer interaction contributes to L2 learning and to provide practitioners with evidence-based pedagogy with more certainty, I will call in this article for replications of two studies: Adams et al. (Reference Adams, Nuevo and Egi2011) and Philp et al. (Reference Philp, Walter and Basturkmen2010). Ten years after their publication, replications of the studies, both investigating attention to form during peer interaction tasks, may provide an important understanding of how peer interaction supports L2 learning.
Adams et al. (Reference Adams, Nuevo and Egi2011) was an experimental study in which the participants were given a set of tasks. The researchers tallied the frequency of feedback and examined its relationship with learning outcomes, with a correlational design. Despite this simple yet powerful design, Adams et al. (Reference Adams, Nuevo and Egi2011) remains to date one of the few studies investigating the direct relationship between peer feedback and L2 learning. Unlike other peer interaction studies, Adams et al.'s findings suggested that peer feedback had a negative impact on L2 learning. A replication study would be able to show whether peer feedback can sometimes be detrimental to L2 learning or the original result was an “isolated incidence” (Porte, Reference Porte2012, p. 7). Philp et al. (Reference Philp, Walter and Basturkmen2010) was a classroom observation study, in which L2 learners’ interactional behaviors were observed. The researchers conducted retrospective interviews to explore the reasons why they decided to (or not to) deliberate about language. In spite of the theoretical and pedagogical relevance, Philp et al.'s (Reference Philp, Walter and Basturkmen2010) study has not generated follow-up/replication studies. I argue that Adams et al. (Reference Adams, Nuevo and Egi2011) and Philp et al. (Reference Philp, Walter and Basturkmen2010) complement each other, in that learner attitudes may determine whether peer feedback ultimately results in L2 learning. Replications of these studies, therefore, help us answer the following key issues: (a) why some learners provide feedback or engage in LREs and some do not; (b) why some peer feedback has a positive impact and some does not; and (c) whether provision of feedback has any impact on L2 learning.
At the outset, I acknowledge that teaching/learning contexts are a variable that inevitably affects the results of studies conducted in similar experimental conditions. This is because, despite some psycholinguistic accounts that do not consider contextual variables in the equation of ISLA, teaching/learning contexts entail a myriad factors influencing learning outcomes (Van Bavel, Mende-Siedlecki, Brady, & Reinero, Reference Van Bavel, Mende-Siedlecki, Brady and Reinero2016; Morgan-Short et al., Reference Morgan-Short, Marsden, Heil, Issa, Leow, Mikhaylova and Szudarski2018; Sato & Storch, Reference Sato and Storch2020). Those factors include, among others, the socio-linguistic and socio-educational status of the target language (e.g., English as a second language (ESL) vs. English as a foreign language (EFL)), educational culture (e.g., test-driven vs. competency-driven), or the participants’ learning backgrounds (e.g., communicative vs. grammar-focused instruction). Changing any of these factors would lead to results that are different from the original study. As Porte and Richards (Reference Porte and Richards2012) argued, the aim of replication is “not to extend the methodology to another context and report new outcomes in a ‘What happened when we did the same thing in…’ situation” (p. 287). Hence, in this paper, I will not include teaching/learning contexts as an approach to replication of the focused studies. Another replication approach that I will not take into account is an increase in sample sizes. Apparently, small sample sizes are an issue when producing generalizable findings, and this needs to be considered when replicating a study (Plonsky, Reference Plonsky2015; Marsden, Morgan-Short, Trofimovich, & Ellis, Reference Marsden, Morgan-Short, Trofimovich and Ellis2018; Hedges & Schauer, Reference Hedges and Schauer2019).
2. Theoretical frameworks
In order to contextualize my suggestions for replication approaches to Adams et al. (Reference Adams, Nuevo and Egi2011) and Philp et al. (Reference Philp, Walter and Basturkmen2010), I first review two theoretical frameworks typically used to account for the impact of peer interaction on L2 learning—interactionist approaches and sociocultural theory. Then, I will discuss a new theoretical approach integrating learner psychology into understanding the ways in which learners engage with a task and/or conversation partner.
The interactionist approaches are based on an assumption that L2 development occurs due to input, feedback, and output (Mackey, Reference Mackey2012; Loewen & Sato, Reference Loewen and Sato2018). When L2 learners interact with each other, they necessarily encounter a communication breakdown, during which they may receive interactional feedback indicating the communication issue—negotiation for meaning (Long, Reference Long1983). Learners may point out each other's L2 errors when there was no communication breakdown as well—negotiation for form (Lyster, Reference Lyster1998). In either case, feedback triggers noticing of the gap between the learner's current interlanguage and what is linguistically acceptable in the target language (Schmidt, Reference Schmidt1990). The degree of noticing can be mediated by multiple factors, such as learners’ primary attention during a task (Bryfonski & Sanz, Reference Bryfonski and Sanz2018; Peace, Reference Peace2019; Sato & McDonough, Reference Sato and McDonough2020), proficiency levels (Williams, Reference Williams2001; Leeser, Reference Leeser2004), and cognitive individual differences (Mackey, Philp, Egi, Fujii, & Tatsumi, Reference Mackey, Philp, Egi, Fujii, Tatsumi and Robinson2002; Kim, Payant, & Pearson, Reference Kim, Payant and Pearson2015). After noticing the gap, the learner may modify their original erroneous utterance and produce the correct version of the error. While producing modified output, the learner may engage in cognitive processes in which they consolidate or restructure their current interlanguage system (Swain, Reference Swain, Gass and Madden1985). In addition, L2 output relates to the practice effect because the learner engages in repeated practice of the target structure (Lyster & Sato, Reference Lyster, Sato, García Mayo, Gutierrez-Mangado and Martínez Adrián2013; DeKeyser, Reference DeKeyser, Loewen and Sato2017; Sato & McDonough, Reference Sato and McDonough2019). Overall, the cognitive-interactionist approach holds that learners’ temporary attentional shift to linguistic forms is key, if communicative interaction is to be beneficial for L2 learning.
Another theoretical framework that has been used to account for peer interaction effectiveness is sociocultural theory. The premise of the theory is that cognition and knowledge are inherently social and are dialogically constructed (Lantolf, Reference Lantolf, Gass and Mackey2012). In the context of peer interaction, when L2 learners collaboratively engage in a task by contributing ideas, generating alternatives, and assessing the alternatives, the co-constructed knowledge results in internalization of this knowledge in individual learners (Donato, Reference Donato, Lantolf and Appel1994; Storch, Reference Storch2002). In order to capture the moment at which knowledge is co-constructed, researchers have used LREs as an analytical unit, defined by Swain and Lapkin (Reference Swain and Lapkin1998) as “any part of dialogue where the students talk about the language they are producing, question their language use, or correct themselves or others” (p. 326). The theoretical understanding of peer feedback in this framework differs from that of the interactionist approach—it is an indication of deliberation about the language. Nonetheless, both cognitively- and socially-oriented researchers focus on the same utterances in a transcript. Overall, research points to the importance of social relationships that learners form during a task, relative to the degree of impact that peer interaction may have on L2 learning (Foster & Ohta, Reference Foster and Ohta2005; Watanabe & Swain, Reference Watanabe and Swain2007; Storch & Aldosari, Reference Storch and Aldosari2013; Fernández Dobao, Reference Fernández Dobao2014; Ballinger, Reference Ballinger, Cenoz and Gorter2015; Storch & Sato, Reference Storch and Sato2019). When learners work collaboratively, interaction is more likely to lead to L2 learning.
The third theoretical framework relates to learner psychology and its mediating effects on interactional behaviors. Several psychological constructs have been examined in relation to interactional behaviors. For instance, Dörnyei and Kormos's (Reference Dörnyei and Kormos2000) study revealed a significant correlation between task motivation and L2 output during a task. The researchers concluded that “all the cognitive and linguistic processes discussed in the L2 task literature depend, to some extent, on this initial condition [task attitudes]” (p. 281). A number of studies have shown that willingness to communicate (WTC) also predicts L2 learners’ task-based performance (MacIntyre, Baker, Clément, & Conrod, Reference MacIntyre, Baker, Clément and Conrod2001; Cao & Philp, Reference Cao and Philp2006). Learner beliefs is another attitudinal variable which has been found to explain learners’ engagement patterns (Tomita & Spada, Reference Tomita and Spada2013; Yoshida, Reference Yoshida2013; Philp & Duchesne, Reference Philp and Duchesne2016). While this research focused on general L2 output (e.g., the number of words and the frequency of initiations of communication), Sato (Reference Sato2017a) showed that interaction mindset, defined as “a disposition toward the task and/or an interlocutor prior to and/or during the interaction” (p. 250), predicted the occurrence of feedback (see also Sato & McDonough, Reference Sato and McDonough2020). In short, learner attitudes may affect how learners interact with each other, including how they deal with language issues together (i.e., attention to form).
The cursory review of peer interaction research indicates that (a) attention to form is beneficial for L2 learning, (b) collaborative relationships are necessary for learners to benefit from peer interaction tasks, and (c) learner attitudes may mediate the regularity and means with which learners address linguistic issues together. With those theoretical and methodological issues in mind, I will discuss approaches to replication of the work of Adams et al. (Reference Adams, Nuevo and Egi2011) and Philp et al. (Reference Philp, Walter and Basturkmen2010) in the following section.
3. The original studies and suggested approaches to replication
3.1 Adams et al. (Reference Adams, Nuevo and Egi2011)
Adams et al. (Reference Adams, Nuevo and Egi2011) tested the impact of peer feedback on L2 learning. The study involved correlation analyses on the frequency of peer feedback and the gain scores in a pre-, post-, and delayed-post-test design. The focus was on two grammatical structures (English simple past tense and locatives) and comprised two developmental tests (oral production and untimed grammaticality judgement test: GJT). Although the study also examined modified output, I will focus on the results related to peer feedback.
The participants in the study were 39 ESL learners in the United States with intermediate-high proficiency.Footnote 1 The learners were paired and given a set of three tasks over three sessions. The first was a spot-the-difference task, used as a warm-up. The second was a picture description task designed to elicit the use of past tense. In the third task, designed to elicit the use of locatives, the paired learners decided a seating arrangement for an event. The learners changed their partners across the three tasks. They were not instructed to provide feedback to each other. The interactions were audio-recorded. Between the two dimensionalities of feedback types—input-output and implicit-explicit—the study examined the implicit-explicit dimension. Two types of input-providing types of feedback were coded in the transcripts—recasts and explicit corrections. Recasts were operationalized as implicit input-providing feedback, and explicit corrections as explicit input-providing feedback. Recasts included the correct version of the error produced by a learner. In addition to providing the correct version, explicit corrections included a clear indication that an error was made (e.g., “you don't say [the error]”). Furthermore, the accuracy of feedback was coded (target-like vs. non-target-like). Before and after the tasks, and also one week after the post-test, the participants took developmental tests. In the oral production tests, learners were given prompts eliciting each target structure (e.g., “the first day in the United States” for past tense) and recorded their answers individually. In the untimed GJTs, the learners were asked to identify ungrammatical sentences. The three testing times led to gain scores from two intervals: pre-post (acquisition gain) and pre-delayed post (retention gain).
In the dataset, 395 instances of feedback were identified (162 for past tense and 233 for locatives). The learners provided slightly more recasts (55%) than explicit corrections (45%). Among those, 67% comprised target-like feedback. With two feedback types, four linguistic features (regular past tense, irregular past tense, past tense in total, and locatives), accuracy of feedback (target-like and non-target-like), and two gain scores, 48 correlations were examined.Footnote 2 Of those correlations, only two reached statistical significances. For irregular past tense, the reception of explicit corrections was negatively correlated with the learning gain (the retention gain scores of the oral production test: −.43). For locatives, the retention gain scores of the oral production test were positively correlated with the reception of recasts (.46). Based on these findings, the authors argued that “feedback may not play as important a role in learner-learner interaction as it plays in NS-learner interactions” (p. 56). When it does, recasts are more conducive than explicit corrections to L2 development. In addition, albeit without data on learner attitudes, Adams et al. claimed that “positive evaluations of learner-learner interactions as a feedback giving context may not translate directly into learning benefits” (p. 57). They also explained the limited impact of peer feedback in light of learner attitudes, arguing: “learners are less likely to accept learner feedback than NS [native speaker] feedback, because they are less likely to assume that feedback from their peer is actually correct” (p. 57).
3.1.1 Approaches to replication of Adams et al. (Reference Adams, Nuevo and Egi2011)
With relatively clear methodological information and the potential mediating variables suggested by the authors of the original study, I call for approximate replications of the work of Adams et al. (Reference Adams, Nuevo and Egi2011). According to Porte and McManus (Reference Porte and McManus2019), approximate replications involve “modifications in two or more elements, crucial or otherwise, to the original study” (p. 78; emphasis in original). I propose three methodological considerations for replication: intervention materials, transcript analysis, and statistical design.
My first suggestion concerns intervention materials. Adams et al. developed experimental tasks in order to elicit specific linguistic structures during peer interaction. The tasks were not related to the curriculum of the class or the participants’ regular classroom activities. It is possible that the irrelevancy of the tasks decreased the participants’ engagement level, resulting in limited incorporation of feedback provided by their classmates. Research has found that authenticity of tasks has an impact on learners’ engagement levels (e.g., Zohoorian, Reference Zohoorian2015). In a replication study, therefore, researchers can use a task embedded in the learners’ regular classroom activities. This is not to say that different types of tasks should be used, as task type is a different variable from authenticity of task (see Pica, Kanagy, & Falodun, Reference Pica, Kanagy, Falodun, Crookes and Gass1993; McDonough & Mackey, Reference McDonough and Mackey2000). Rather, a replication study can use the same task types as Adams et al.'s, but with contents and procedures that are adjusted to the intact class to which participants belong. This replication modification comes with an additional benefit. If a researcher hopes to discover L2 learners’ behaviors in a real-world setting and provide practitioners with evidence-based pedagogical recommendations, an intervention material that increases ecological validity would be appropriate (see DeKeyser & Prieto-Botana, Reference DeKeyser and Prieto-Botana2019; Sato & Loewen, Reference Sato, Loewen, DeKeyser and Prieto Botana2019).
My second suggestion relates to transcript analysis of feedback.Footnote 3 A replication study needs to code any types of feedback before narrowing the focus to specific types of feedback. In Adams et al., two types of feedback were coded and their impact on L2 development was reported. Focusing on specific types of feedback is problematic simply because we cannot tell whether the developmental outcomes were affected by other types of feedback. It is possible that correlations of developmental test scores and other types of feedback would have yielded positive correlations. In a similar vein, a replication study can focus on other types of feedback. Teacher (or NS) feedback research has shown that whether or not learners notice the feedback explains its impact on learning. It is likely that the participants noticed explicit corrections more than recasts in Adams et al.'s study. However, the explicitness of didactic recasts (recasts with prosodic stresses) can be as marked as for explicit corrections (for categorization of feedback, see Lyster, Saito, & Sato, Reference Lyster, Saito and Sato2013). Adams et al. did not consider this and operationalized a recast as a “reformulation of an interlocutor's utterance that involved the non-target-like use of one of the target forms” (p. 51). If a researcher decides to examine the implicit-explicit dimension of feedback types, as Adams et al. did, implicitness needs to be carefully controlled (note that controlling explicitness is not difficult). To this end, comparing didactic recasts and explicit corrections may reveal whether the added negative evaluations (for example, “No, you are wrong.”) had any impact on learners’ incorporation of peer feedback in Adams et al.'s study.
My third suggestion is related to statistical analysis. L2 learners can, and likely are, feedback providers and receivers during peer interaction tasks. And, provision and reception of feedback may trigger different cognitive processes. Recent research shows that provision of feedback only may affect feedback providers’ L2 development (Sato, Reference Sato, Nassaji and Kartchava2017b; Sato & McDonough, Reference Sato and McDonough2020; Martin & Sippel, Reference Martin and Sippel2021). Hence, coding feedback receptions only while ignoring feedback provisions is problematic in understanding the relationship between peer feedback and L2 learning. Adams et al. acknowledged this aspect but did not consider it when analyzing the data or interpreting the results: “Future research could consider whether providing feedback to an interlocutor plays a role in learning through interaction” (p. 60). A replication study, therefore, must tally the frequencies of provision and reception of feedback separately. In terms of statistical models, group-based designs (e.g., t-tests and ANOVAs) may not tease apart the effect of provision and reception of feedback. For instance, by comparing a peer feedback group vs. a control group, we would not know whether the result was due to reception of feedback, provision of feedback, or both (e.g., Sato & Lyster, Reference Sato and Lyster2012; Sippel & Jackson, Reference Sippel and Jackson2015). Associational designs (e.g., correlations, regressions, and structural equation modellings) seem more appropriate.
3.2 Philp et al. (Reference Philp, Walter and Basturkmen2010)
The approach of Philp et al. (Reference Philp, Walter and Basturkmen2010) was novel because, in addition to investigating how much attention L2 learners pay to language form during peer interaction, the researchers explored attitudinal factors explaining the occurrence of attentional shifts. Such a “heuristic or hypothesis-generating type” (Philp et al., Reference Philp, Walter and Basturkmen2010, p. 274) of exploration was included in the third research question: “What factors account for differences in the incidence and extensiveness of LREs?” I will focus on this question because of its relevance to my suggestions for conceptual replications.
Philp et al. conducted a classroom observation study with retrospective stimulated recalls.Footnote 4 From an intact class of 30 students, seven learners volunteered to participate. Their proficiency levels ranged from beginner to advanced. The class was an undergraduate-level L2-French class in New Zealand. The teacher used a theme-based textbook and various pedagogical techniques including semi-scripted dialogues, explicit grammar teaching, and group work. Over a course of 12 sessions in 3 weeks, peer interaction was audio-recorded. The peer interaction tasks were role-plays and discussion tasks. First, the researchers identified LREs. These were then coded for initiation (self-, other-, and teacher) and linguistic focus (phonological, lexical, and grammatical). In the third week, stimulated recall interviews were conducted individually. In the interviews, the researchers played audios where LREs had been identified. The recall prompts included non-directive questions such as “So what's going on there?”, as well as specific questions (for example, “Why did you say it in English first?”). The interview data were submitted to a theme-based analysis.
In the interaction data, 33 LREs were identified from 26 interactions across 8 tasks. The majority of LREs (80%) related to lexical issues, followed by grammatical (f = 7) and phonological (f = 1). Of the LREs, 16 were self-initiated, 15 peer-initiated, and one teacher initiated.Footnote 5 The themes emerging from interviews related to provision of feedback, including: reasons for participants’ provision of feedback, attitude to error and error correction, attitude to interlocutor, and orientation to task. The comments related to various affective and social factors. For instance, a learner thought his proficiency was insufficient to correct his classmates’ errors, and believed providing feedback would “mark him as arrogant and different from other students” (p. 271). Other factors that prevented the learners from providing feedback were consideration about interruption to the conversation, and concerns about embarrassing their classmates. The researchers argued that the majority of participants were reluctant to correct their classmates’ L2 errors, but had different reasons for hesitation. Based on the strong influence of learner attitudes on LREs, the researchers concluded that “in the complex setting of the classroom, task features alone may not predict incidence or quality of a focus on form: what the students themselves bring to the task is important, both individually and collectively” (p. 275).
3.2.1 Approaches to replication of Philp et al. (Reference Philp, Walter and Basturkmen2010)
Philp et al.'s (Reference Philp, Walter and Basturkmen2010) study was exploratory in examining factors that foster attention to form. In fact, the third research question was added after the discovery that affective and social factors significantly impacted on ways in which the participants engaged in language-related issues. The issue surrounding learner attitudes and their impact on attention to form has attracted much research in the last few years. Yet, there has not been much follow-up of Philp et al.'s study since its publication ten years ago. With this background, I call for conceptual replication of Philp et al.'s (Reference Philp, Walter and Basturkmen2010) study that “extends beyond confirmation of the original findings toward theory and model building” (Porte & McManus, Reference Porte and McManus2019, p. 85). In discussing future conceptual replication studies, I propose the following considerations: internal validity control, data elicitation materials and analysis, and additional learner attitude data.
First, a replication study can improve the internal validity of Philp et al.'s work. For one, the proficiency levels of the learners in the original study varied widely. Given the potential impact of proficiency differences on the way in which learners engage with each other (Williams, Reference Williams2001; Leeser, Reference Leeser2004), a replication study could focus on a class in which learners are presumed to have comparable proficiency (e.g., measured by the institution). Alternatively, and ideally, researchers could collect information on participants’ proficiency with an independent measure, and the proficiency scores could be used in a covariate analysis. A replication study would also control for participants’ first language (L1); in Philp et al.'s study, the participants’ L1s were either English or Mandarin, and they switched their partners during the tasks. Given that learners could resort to their L1 when encountering a linguistic issue, the results of LREs as well as provision of feedback in Philp et al. might have been affected by the pairing patterns of the same/different L1s. A replication study can either focus on a group of learners with the same L1, or design pairing patterns so that the L1-sharing effect is controlled. Controlling those variables would allow us to understand why learners ignored the vast majority of errors produced by their peers and missed out the learning opportunity surrounding interactional feedback in the original study (see also Pica, Lincoln-Porter, Paninos, & Linnell, Reference Pica, Lincoln-Porter, Paninos and Linnell1996; Toth, Reference Toth2008; Xu, Fan, & Xu, Reference Xu, Fan and Xu2019).
Second, the analysis of qualitative data can be improved in a replication study. A key requirement for replicating qualitative methods is the clarity of data analysis (Porte & Richards, Reference Porte and Richards2012). Unfortunately, Philp et al. did not provide extensive explanations of the coding procedure for the interview data. A replication study must provide more detailed accounts of this procedure. Thus, a replication study could be repeated even with a qualitative method. If the qualitative data was sufficiently robust, a researcher could quantify the emerged themes and test them against the interaction data. For instance, with an associative design, the frequency of negative comments by a learner related to feedback provision can be examined in relation to the frequency of feedback provision by this learner. The same can be done for comments related to reception of feedback and the frequency of incorporation of feedback in the form of modified output. Alternatively, a researcher can devise a questionnaire focusing on learner attitudes to peer interaction (see, for example, the interaction mindset questionnaire in Sato & McDonough, Reference Sato and McDonough2020). In this way, Philp et al.'s study can be replicated in a quantitative manner in which provision and/or incorporation of feedback are predicted.
Finally, in a conceptual replication study, a researcher can examine different learner psychology and explore its relation to peer feedback provision and incorporation. This is an interesting area of study; it is only recently that researchers have started to explore the relationship between learner psychology and interaction behavior, especially in the classroom context, where different emotions and social relationships arise. The list of potential psychological constructs is long. It includes motivation, WTC, interaction mindset, language mindsets, foreign language anxiety, and metacognition, to name just a few. Whichever the construct, however, the study pursues the question in the original study: what fosters a focus on form? After collecting learner psychology and interaction data, a researcher can test a regression model which explores the learner psychology's predictability of feedback provision and incorporation. Having recommended different psychological constructs, however, I caution that a replication study needs to theoretically justify the choice of a construct in examining attention to form during peer interaction.
4. Conclusion
In this paper, I have listed methodological suggestions for replicating the studies by Adams et al. (Reference Adams, Nuevo and Egi2011) and Philp et al. (Reference Philp, Walter and Basturkmen2010). Such replications would help fill the current theoretical and methodological gaps in the topic of peer interaction, which persist ten years after the publications of the original studies. I argued that replications would help us understand how learner attitudes might have affected the results of the original studies. Learner attitudes in this context entailed factors which may be both affective (e.g., individual learners’ beliefs about peer interaction) and social (e.g., the degree of collaboration that they construct together).
To conclude this paper, I highlight the pedagogical implications of replication studies. Whether or not a replication study supports the original findings, the results will be useful for designing peer feedback training materials (see Sato & Lyster, Reference Sato and Lyster2012; Fujii, Ziegler, & Mackey, Reference Fujii, Ziegler, Mackey, Sato and Ballinger2016; Sippel, Reference Sippel2019; Dao, Reference Dao2020; Sato, Reference Sato, Lambert and Oliver2020). There are several possibilities. First, if a replication study identifies patterns in relation to the types of error which are given feedback, and the types of feedback that are incorporated, training material can include explicit instruction about the error types that tend to be ignored/unnoticed and facilitate the use of feedback types that are more likely to be incorporated. Second, if a replication study reveals specific learner attitudes that promote (or impede) provision and incorporation of feedback, teachers can devise pedagogical materials tapping into those attitudinal factors. The material can focus on mentally preparing students for correcting their classmates’ errors and incorporating feedback into their own learning. The material can also include aspects that mitigate attitudinal factors found to negatively impact on provision and incorporation of feedback. With successful training, learners may start to trust their classmates as a potential learning resource, which would help reduce unnecessary rejections of feedback. In turn, learners would perceive fewer obstacles in correcting their classmates’ errors. Such an interactional pattern resembles collaborative learning that researchers support, regardless of their theoretical understanding of L2 learning.
Acknowledgment
This work was partially supported by the Fondo Nacional de Desarrollo Científico y Tecnólogico from the Ministry of Education of Chile (FONDECYT: 1181533) as well as PIA (CIE160009) from the Chilean National Commission of Science and Technology (CONICYT).
Masatoshi Sato (Ph.D.: McGill University) is a Professor at Universidad Andrés Bello, Chile. His research agenda is to conduct theoretical and applied research in order to facilitate the dialogue between practitioners and researchers. He has taught English and Japanese in Japan, USA, Chile, and Canada. His research interests include peer interaction, corrective feedback, L2 motivation, metacognition, and the research-pedagogy link. On those topics, he has published in various international journals. In addition, he co-edited volumes from John Benjamins (2016, with Susan Ballinger: Peer interaction and second language learning), Routledge (2017, with Shawn Loewen: The Routledge handbook of instructed second language acquisition; 2019, Evidence-based second language pedagogy), and for the journal Language Teaching Research (2021, with Kata Csizér: Learner psychology and instructed second language acquisition). His co-authored textbook with Shawn Loewen (A practical guide to second language acquisition) from Cambridge University Press will appear in 2022. He is the Editor of Language Awareness. He is the recipient of the 2014 ACTFL/MLJ Paul Pimsleur Award.
