Introduction

Footnote With much empirical evidence of a beneficial role of interaction in second language (L2) development, researchers have become interested in investigating specific aspects of interaction (e.g., negotiation for meaning, corrective feedback (CF), modified output, noticing, etc.) that likely influence the extent to which interaction benefits L2 learning (Mackey, Reference Mackey2012; Mackey, Abbuhl, & Gass, Reference Mackey, Abbuhl, Gass, Gass and Mackey2012; Mackey & Goo, Reference Mackey, Goo and Chapelle2013; Gass & Mackey, Reference Gass, Mackey, VanPatten and Williams2015; Loewen & Sato, Reference Loewen and Sato2018). Among varied features of interaction, CF has been found to be quite effective at drawing learners' attention to L2 linguistic features during interaction, and has engendered much scholarly discussion of pivotal importance and numerous empirical studies on its potential for L2 development (see Russell & Spada, Reference Russell, Spada, Norris and Ortega2006; Mackey & Goo, Reference Mackey, Goo and Mackey2007; S. Li, 2010*; Lyster & Saito, 2010*; Lyster, Saito, & Sato, 2013*; Brown, 2016*; Nassaji, 2016* for reviews and meta-analyses). Recasts, inter alia, have been at the center of most CF research and greatly explored with a view to understanding the nature of recasts, their characteristics (in various L2 learning contexts), their relative efficacy over other CF moves, and moderator variables that may mediate the effectiveness of recasts.

It should be noted that recasts have also been researched to explain how children's emerging errors are eradicated in the first language (L1) learning process. Relevant L1 research on recasts, adult reformulations of children's erroneous utterances with corrections, has provided evidence for the existence and utility of recasts in adult-child conversation and positive potential of recasts for L1 development (e.g., Baker & Nelson, Reference Baker and Nelson1984; Bohannon & Stanowicz, Reference Bohannon and Stanowicz1988; Farrar, Reference Farrar1992; Saxton, Reference Saxton1997, Reference Saxton2005, Reference Saxton2010; Saxton, Kulcsar, Marshall, & Rupra, Reference Saxton, Kulcsar, Marshall and Rupra1998; Chouinard & Clark, Reference Chouinard and Clark2003; Saxton, Backley, & Gallaway, Reference Saxton, Backley and Gallaway2005; Clark, Reference Clark2014; Clarke, Soto, & Nelson, Reference Clarke, Soto and Nelson2017). Chouinard and Clark (Reference Chouinard and Clark2003), for instance, found that negative evidence in the form of recasts was available in adult reformulations and for different types of linguistic errors, and that the overall number of reformulations decreased with age in most linguistic areas. Their findings support the idea that ‘children monitor what adults say to them, and, in particular, attend to and respond to adult reformulations of their intended meanings’ (p. 658). Nevertheless, it is true that adult recasts and related research/data have also been viewed and interpreted from different angles, and the role of recasts in L1 development has often been undervalued (e.g., Grimshaw & Pinker, Reference Grimshaw and Pinker1989; Gordon, Reference Gordon1990; Morgan, Bonamo, & Travis, Reference Morgan, Bonamo and Travis1995). In addition, one empirical concern that likely obscures evidence for the positive potential of recasts for L1 development is, as Saxton (Reference Saxton2010) states, that ‘teasing apart the effects of one kind of adult response from all other sources of input influence is no easy matter, empirically’ (p. 98). In other words, controversy still exists concerning the role of adult recasts in L1 development. On the contrary, recasts in L2 learning contexts have been much discussed and investigated as an important contributing factor in L2 development, to which I now turn.

Long (Reference Long2007) defined a recast as ‘a reformulation of all or part of a learner's immediately preceding utterance in which one or more nontargetlike (lexical, grammatical, etc.) items is/are replaced by the corresponding target language form(s), and where, throughout the exchange, the focus of the interlocutors is on meaning, not language as object’ (p. 77; see also Chaudron, Reference Chaudron1977; Long, Reference Long, Ritchie and Bhatia1996; Nicholas, Lightbown, & Spada, 2001*; Ellis & Sheen, 2006* for more definitions and reviews of recasts). To be more specific in a pedagogical sense, a recast, as has been coded/utilized in most L2 research, is ‘the teacher's reformulation of all or part of a student's utterance, minus the error’ (Lyster & Ranta, 1997*, p. 46). In Example 1, the researcher provides a recast by reformulating the ungrammatical portion of the student's preceding utterance (‘follow’) with respect to the English past tense morpheme, which is followed by the student's modification of her original output.

Recasts, which may function as ‘crucial points at which implicit and explicit learning converge in optimal ways’ (Long, Reference Long2015, p. 55), likely create propitious conditions for L2 development because they are provided immediately following the learner's erroneous utterance, the meaning of which has already been comprehended, that is, more attentional resources can be directed to the linguistic form(s) targeted in recasts (e.g., Long, Reference Long, Ritchie and Bhatia1996, Reference Long2007, Reference Long2015; Doughty, Reference Doughty and Robinson2001). Long (Reference Long2007) states that ‘recasts convey needed information about the target language in context, when interlocutors share a joint attentional focus, and when the learner already has prior comprehension of at least part of the message, thereby facilitating form-function mapping’ (p. 77). Two contributing factors to the potential efficacy of recasts on L2 learning are semantic transparency and an immediate juxtaposition of a learner's non-target-like utterance and a recast (Long, Reference Long, Ritchie and Bhatia1996, Reference Long2007, Reference Long2015). Semantic transparency allows L2 learners to utilize freed-up attentional resources to focus on linguistic forms and facilitate form-function mapping. An immediate juxtaposition of a learner's nontargetlike utterance and a recast, which likely contributes to enhancing the salience of the target form in the recast and thus precipitates learner noticing (e.g., Leeman, 2003*), renders highly likely a cognitive comparison of the two forms in contrast, which may lead the learner to perceive the corrective nature of the recast provided. Similarly, Doughty (Reference Doughty and Robinson2001) suggests that immediately contingent recasts enable L2 learners to perform effective cognitive comparisons during speech processing.

Early on, when researchers began to show interest in the potential benefits of recasts during interaction, recasts were examined with respect to whether they were in fact provided, and then used (see Gass, Mackey, & Ross-Feldman, Reference Gass, Mackey and Ross-Feldman2005 for classroom vs. laboratory contexts), and if used, how learners responded to recasts, that is, in terms of their distribution and relation to learner uptake in L2 learning contexts (e.g., Lyster & Ranta, 1997*; Lyster, 1998a*, 1998b*; Ellis, Basturkmen, & Loewen, 2001*; Panova & Lyster, 2002*; Sheen, 2004*). In his recent meta-analysis of classroom-based CF studies, Brown (2016)* observed that, as in Lyster and Ranta's (1997)* findings (and many other studies of different L2 learning contexts and task types, e.g., Sheen, 2004*; Yoshida, 2008*; Gurzynski-Weiss & Révész, Reference Gurzynski-Weiss and Révész2012), recasts were the most frequently used CF moves that teachers provided in L2 classrooms (57%), followed by prompts (30%), i.e., repetition, elicitation, clarification, and metalinguistic feedback. He also found that educational contexts (level of learners' education) influenced teachers’ CF choices with more recasts supplied to university/adult learners and elementary-level learners than prompts and with more prompts to high school learners than recasts. Furthermore, his meta-analysis revealed classroom teachers offered more recasts to more advanced learners, compared to beginners. The amount of learner uptake following recasts, however, has been argued to vary dependent on L2 learning contexts with more recasts tending to be taken up in form-oriented learning contexts than in meaning-focused contexts (e.g., Sheen, 2004*; Lyster & Mori, 2006*) and also on the appropriateness of modified output production in given discourse contexts (e.g., Oliver, 1995*; Mackey & Philp, 1998*; Braidi, 2002*).

Given the possibility of learner responses being related to learner noticing, and the significance of noticing in overall L2 learning (Schmidt, 2001), different characteristics of recasts and learner uptake following recasts have also been extensively researched in relation to noticing and L2 development (e.g., Mackey, Gass, & McDonough, 2000*; Philp, 2003*; Mackey, 2006*; Egi, 2007*, 2010*; Yoshida, 2010*; Bao, Egi, & Han, 2011*). Overall findings suggest that shorter recasts were more likely to be noticed (e.g., Philp, 2003*; Egi, 2007*) and led to more uptake (Sheen, 2006*; see also Sagarra & Abbuhl, 2013* for the positive role of orally enhanced recasts in the production of targetlike modified output), and that intonation, morpheme length, and number of changes were the most relevant factors mediating the extent of beneficial effects of recasts on L2 performance (e.g., Loewen & Philp, 2006*; but see Erlam & Loewen, 2010* for nonsignificant performance differences between implicit and explicit recasts) although intonation does not always appear to affect learner noticing (e.g., Al-Surmi, 2012*). Egi (2010)* reported that learner noticing of recasts was found to be positively related to the production of uptake, modified output, and repairs (correct modifications). However, it should be noted that research findings indicate immediate uptake following recasts and learner noticing may not necessarily be isomorphic (e.g., Mackey & Philp, 1998*; Yoshida, 2010*; Bao et al., 2011*), and learner responses may in fact be red herrings (Mackey & Philp, 1998*, p. 338). One reason may be, as Long (Reference Long2007) notes, ‘discourse constraints often preempt the possibility of repair in the immediately following turn, especially in classroom settings, and also because studies show that effects are sometimes delayed’ (p. 94). In other words, ‘absence of evidence is not the same thing as evidence of absence’ (Mackey, 2006*, p. 409). Furthermore, incorporation of recasts into new contexts may be more related to L2 learning than immediate uptake following recasts (e.g., McDonough & Mackey, 2006*; Nassaji, 2011*).

With these descriptive dimensions in mind, researchers have investigated and discussed the overall effectiveness of recasts in actual L2 development (e.g., Doughty & Varela, Reference Doughty, Varela, Doughty and Williams1998; Long, Inagaki, & Ortega, 1998*; Mackey & Philp, 1998*; Ishida, 2004*; Loewen & Philp, 2006*; McDonough & Mackey, 2006*; Erlam & Loewen, 2010*; Nassaji, 2017*), and their relative efficacy over other CF moves (e.g., recasts vs. prompts or explicit correction) on the learning of specific L2 target features (e.g., Lyster, 2004*; Ammar & Spada, 2006*; Ellis et al., 2006*; Ellis, Reference Ellis and Mackey2007; Sheen, Reference Sheen and Mackey2007, 2008*; Ammar, 2008*; Lyster & Izquierdo, 2009*; Nassaji, 2009*; Yang & Lyster, 2010*; Adams, Nuevo, & Egi, 2011*; Goo, 2012*, Reference Goo, Granena, Jackson and Yilmaz2016; Yilmaz, 2012*, 2013a*, 2013b*; S. Li, 2013*, 2014*; van de Guchte, Braaksma, Rijlaarsdam, & Bimmel, 2015*; Gooch, Saito, & Lyster, 2016*). Research findings to date in this regard show a range of outcomes, depending upon such variables as research contexts, target type, outcome measures, learner proficiency, and so on. For instance, although the positive effects of recasts on L2 learning have been evidenced in quite a number of empirical studies, especially in lab-based ones (see Mackey & Goo, Reference Mackey, Goo and Mackey2007; Goo & Mackey, 2013*; Nassaji, 2016* for relevant reviews), results on the relative efficacy of recasts when compared directly with other CF moves (e.g., prompts or explicit CF moves) have indicated a general, albeit not definitive, trend in favor of prompts or explicit feedback in many classroom-based studies (see Spada & Lightbown, Reference Spada, Lightbown, Mackey and Polio2009; Lyster & Saito, 2010*; Lyster et al., 2013*; Nassaji, 2016* for reviews). Earlier research also suggests that high-proficiency learners are more likely to benefit from recasts than low-proficiency learners (e.g., Mackey & Philp, 1998*; Ammar & Spada, 2006*; Ammar, 2008*; see also Saito & Wu, 2014* for a similar discussion on their findings), due probably to high-proficiency learners’ being better able to notice recasts (e.g., Panova & Lyster, 2002*; Yoshida, 2010*).

Recasts have also been examined in relation to other varied potential factors that may mediate the effects of recasts on L2 development, for example, recast features/types (e.g., Erlam & Loewen, 2010*; Al-Surmi, 2012*; Sagarra & Abbuhl, 2013*; Nassaji, 2017*; Wacha & Liu, 2017*), synchronous computer-mediated-communication (SCMC) contexts, often in comparison with face-to-face (FTF) contexts (e.g., Yilmaz & Yuksel, 2011*; Yilmaz, 2012*; Baralt, 2013*; Parlak & Ziegler, 2017*; Saito & Akiyama, 2017*), task features (e.g. Baralt, 2013*; Révész, Sachs, & Hama, 2014*; Y. Kim, Payant, & Pearson, 2015*; Kourtali & Révész, 2020*), individual differences in cognitive capacities (e.g., Mackey, Philp, Egi, Fujii, & Tatsumi, Reference Mackey, Philp, Egi, Fujii, Tatsumi and Robinson2002; Goo, 2012*, Reference Goo, Granena, Jackson and Yilmaz2016; Mackey & Sachs, 2012*; Révész, 2012*; S. Li, 2013*; Y. Kim et al., 2015*; Yilmaz & Granena, 2016*, 2019*; Granena & Yilmaz, 2019*; J. Kim, 2019*; S. Li, Ellis, & Zhu, 2019*; Kourtali & Révész, 2020*), and target structures (e.g., Long et al., 1998*; Mackey, 2006*; Varnosfadrani & Basturkmen, 2009*; Yilmaz, 2012*; S. Li, 2014*; van de Guchte et al., 2015*; J. Kim, 2019*; H. Li & Iwashita, 2019*). Other moderator variables have also been explored in terms of whether they relate to the effectiveness of recasts, such as language anxiety (e.g., Sheen, 2008*; Rassaei, 2015a*; H. Li, 2018*), cognitive styles, e.g., field-dependent vs. field-independent (e.g., Rassaei, 2015b*), and joint attention through eye gaze and gestures (e.g., McDonough, Crowther, Kielstra, & Trofimovich, 2015*; Nakatsukasa, 2016*, 2019*; McDonough, Trofimovich, Dao, & Dion, 2017*). Rassaei's (2014)* study is also another interesting extension of recast research where recasts were compared with scaffolded feedback (feedback from a sociocultural perspective). Overall, relevant research findings shed light on the roles of some of these potential variables in the extent to which recasts facilitate L2 learning. One important point to note is that as in many other areas of L2 research, there has been an overall lack of longitudinal research on recasts (but see Akiyama, 2017*; Saito & Akiyama, 2017*; Wacha & Liu, 2017* for recent efforts), and as such, recent empirical studies have mainly focused on immediate and short-term effects only (e.g., J. Kim, 2019*; Yilmaz & Granena, 2019*; Kourtali & Révész, 2020*). L2 research on recasts designed to investigate long-term effects, preferably in a longitudinal treatment setting, would bring to light more diverse and dynamic aspects of recasts in L2 learning. Much more attention needs to be drawn to this methodological issue in future research.

As illustrated above, research on recasts has recently been expanding in terms of experimental boundaries and diversifying in range, above and beyond what was formerly the epicenter of scholarly discussion and empirical investigation, by exploring varied mediating factors, for example, joint attention established through gaze in eye tracking research (e.g., McDonough et al., 2015*; McDonough et al., 2017*). Obviously, the burden of proof with respect to the extent to which these newly explored (and thus still underexplored) variables mediate the efficacy of recasts on L2 learning is on L2 CF researchers. Researchers' concerted efforts along with methodological rigor and innovations in research methodology will, for sure, further and deepen our understanding of the precise nature of recasts and their multifarious functions in the process of L2 learning and teaching, especially in relation to various learner-internal and -external factors.

Given the limited space allowed for this particular section, the present timeline includes articles published only in peer-reviewed second language acquisition journals from the 1990s, when Long's focus on form (Long, Reference Long, de Bot, Ginsberg and Kramsch1991: Long & Robinson, Reference Long, Robinson, Doughty and Williams1998) and his updated Interaction Hypothesis (Long, Reference Long, Ritchie and Bhatia1996) took center stage, and thus interaction researchers began to show an unequivocal and intense interest in, and focus much scholarly attention on, this multi-faceted feedback move and its potential for interaction-driven L2 development, through mid 2019. It should also be noted that Lyster and Ranta's (1997)* study has no doubt contributed to proliferating CF research involving the role of recasts in L2 learning. One hundred and five refereed journal articles are included in the present research timeline and coded according to the following scheme. In addition, considering research contexts may influence the extent to which recasts are provided and perceived as CF, and facilitate L2 learning (e.g., Sheen, 2004*; Lyster & Mori, 2006*; and see also Mackey & Goo, Reference Mackey, Goo and Mackey2007 for the role of research contexts in interaction-driven L2 learning), all empirical studies included are also coded as FL (foreign language context), SL (second language context), or IM (immersion).

1. A. Review articles on recasts or certain aspects of recasts (including meta-analyses)

2. B. Experimental studies of the effects of recasts (vs. other feedback moves)

   1. 1. Dyad

   2. 2. (Teacher-fronted) Classroom

3. C. Descriptive studies of recasts (vs. other feedback moves) and learner responses

   1. 1. Dyad

   2. 2. (Teacher-fronted) Classroom

4. D. Research designed to investigate learner perception/noticing of recasts

   1. 1. Dyad

   2. 2. (Teacher-fronted) Classroom

5. E. Mediating factors

   1. 1. Study on recasts in SCMC (vs. FTF) contexts

   2. 2. Study on recasts and task types/characteristics

   3. 3. Study on recasts and individual differences (IDs) in cognitive/affective factors

   4. 4. All other variables (e.g., proficiency, target type, outcome measures, feedback timing)