Corrective Feedback in SLA : Classroom Practice and Future Directions

In the realm of language teaching, error correction has a long and contentious history. Some schools of thought like nativism refute error correction while others firmly adhere to error correction and regard error as a sin that should be avoided. This dilemma bewilders TEFL practitioners and teachers how to treat errors. Due to the controversial nature of this issue, whether and how to correct errors have spawned numerous celebrated publications in this area in the domains of first language acquisition (FLA) and second language acquisition (SLA). In this vein, lots of studies have probed the role of corrective feedbacks in language classrooms. This paper reviews the main surveys on corrective feedback, providing the theoretical rational for and against error correction, shedding light on different types of corrective feedbacks, and encapsulating the theoretical and empirical studies conducted to investigate corrective feedback and its impact on different aspects of language, offering issues for further directions to cast away all the doubts in this domain.


Introduction
For decades, questions about errors correction in second language acquisition (SLA) have been hotly debated, giving birth to a great deal of theoretical and empirical research.A big question mark on the top of TEFL practitioners' head is whether to provide learners with only positive evidence as nativists and rationalists believe or to expose them to negative evidence as well.Those working within nativist paradigm argue that offering learners the target-like language i.e. positive evidence is sufficient while interactionist scholars (e.g., Gass, 2003) devote a pivotal role to negative evidence as well.The term negative evidence is often used interchangeably with the terms negative feedback and corrective feedback to refer to any erroneous utterances of language learners (Gass, 1997;Schachter, 1991).Some schools of thought like Behaviorism considered errors as taboos in their discourse and believed that they should be immediately corrected by the teacher (Brown, 2007;Larsen-Freeman, 2000;Richards &Rodgers, 2001) while others claimed that error correction was not only unnecessary, but also harmful to language learning (Krashen, 1981a;1981b).With the emergence of communicative approaches, error correction underwent a radical shift (Nicholas, Lightbown, & Spada, 2001;Russell, 2009).CLT advocates created a balance between what Audiolinguists and Cognitistvists do and suggested that an error must be viewed as evidence of learners' linguistic development, not as a sin to be avoided.CLT advocates recognized the need for fluency and this allows teachers to leave some errors uncorrected.
The dilemma of error correction, i.e. to correct or not to correct errors has raised controversial issues in SLA.Hendrickson (1978) states that attention must be paid to global errors rather than local ones and the process of error correction must occur with consistency and systematicity.Lee (1990), along with Bailey and Celce-Murcia (1979) also believe that error correction is an indispensible part of mastery in language learning.On the other hand, some scholars opine that corrective feedback should be abandoned because it can have potential negative effects on learners' affect, and hence impeding the flow of communication (Krashen, 1981a;1981b;Truscott, 1999).Nevertheless, currently SLA researchers strongly believe in error correction and corrective feedback (Ellis, 2006).

Theoretical Rationale for Error Correction
In nativist paradigm, the application of corrective feedback has little impact on language learning since it merely affects performance and leaves the underlying competence untouched (Schwartz, 1993;as cited in Kim, 2004).This idea is rooted in the tenets of nativism.That is the formation and restructuring of grammars is attributed to innate human linguistic mechanism (Cook, 1991;Schwartz, 1993;as cited in Kim, 2004).Moreover, Krashen (1982Krashen ( , 1985; as cited in Kim, 2004), a nativist, also repudiates any discernable effects of corrective feedback in SLA.He affirms that any knowledge consciously learned through explicit instruction cannot have a significant impact on L2 acquisition.Swain's (1985Swain's ( , 1995) ) Output Hypothesis, shedding light on the significance of output opportunities in L2 development, argues that this can help learners to make and test hypotheses about linguistic correctness and to develop metalinguistic knowledge of how the L2 works (Kowail & Swain, 1994;as cited in Kim, 2004).Schmidt (1990Schmidt ( , 1995Schmidt ( , 2001)), in his Noticing Hypothesis, opines that noticing is a prerequisite of learning, continuing that conscious attention must be paid to input in order for L2 learning to proceed.The proponents of the Noticing Hypothesis (Schmidt, 1990(Schmidt, , 1995(Schmidt, , 2001) also ascertain the benefits of corrective feedback regarding the facilitative role it has in drawing learners' attention to form.From this theoretical front, corrective feedback acts as stimulus, triggering learners to identify the gap between their erroneous utterance and the target form.Thus in perceiving different types of feedback and enhancing their benefits for language learners, noticing and awareness is vital.
The updated interaction hypothesis proposed by Long (1996) also lend support to explicit error correction.According to Long (1996), corrective feedback provides direct and indirect information about the grammaticality of the utterances as well as additional positive evidence which may otherwise be absent in the input.He argues that "negotiation for meaning, and especially negotiation work that triggers interactional adjustments by the NS or more competent interlocutor" ease the process of language learning since it "connects input, internal learner capacities, particularly selective attention, and output in productive ways" (Long, 1996, pp. 451-452).
Similarly, the advocates of connectionist model of language learning, differentiating conscious and unconscious learning, underline the role of attention and consciousness in conscious learning and the role of connectionist learning in implicit learning and yield support to explicit error correction (Nick Ellis, 2005; as cited in Ellis, Loewen, & Erlam, 2006).

Different Types of Corrective Feedback
In addition to recast which is the most frequently used feedback, six different corrective strategies have been identified: explicit correction, clarification requests, metalinguistic information, elicitation, repetition, and translation (Lyster & Ranta, 1997;Panova & Lyster, 2002).All of these techniques are placed in an explicit-implicit continuum.The following section elaborates on each of these corrective feedback techniques.

Recast
The term recast was initially used in the literature of L1 acquisition to refer to responses by adults to children's utterances (Nelson, Carskaddon, & Bonvillian, 1973;as cited in Nicholas et al, 2001); afterward it merged into the domain of L2 acquisition in which different definitions were utilized for this term.Lyster and Ranta (1997, p. 46) define recast as 'teacher's reformulation of all or part of a student's utterance, minus the error'.According to Ellis and Sheen (2006, pp. 78-80), recasts are of various types including corrective recasts (Doughty & Varela, 1998), corrective/non-corrective recasts (Farrar, 1992), full/partial recasts, single/multiple recasts, single utterance/extended utterance recasts, and simple/complex recasts (Ellis & Sheen, 2006).Nelson, Denninger, Bonvillian, Kaplan, and Baker (1983) also propose two further classifications of recasts, i.e. simple and complex recasts; the former deals with minimal changes to the child's utterance while the latter is concerned with providing the child with substantial additions.It is also mentioned that in terms of their linguistic development, children benefit from simple recasts more than complex ones (Nelson et al., 1983) There is no general agreement among SLA practitioners regarding the effectiveness of recasts due to their limitations.Though some researchers (e.g., Long, 2006;Doughty, 2001) consider recast as an effective corrective feedback technique, others (Lyster, 1998a;Panova & Lyster, 2002) propose that learners usually pass recasts unnoticed and thus they regard them not as effective for interlangauge development.A number of interaction researchers (e.g., Braidi, 2002;Chaudron, 1977Chaudron, , 1986;;Fanselow, 1977;Long, 1996;Lyster, 1998aLyster, , 1998b;;Lyster & Ranta, 1997;Nicholas et al., 2001;Oliver & Mackey, 2003), referring to the ambiguity of recast, also argue that recast might be perceived as synonymous in function as mere repetition for language learners hence learners might fail in perceiving the corrective function of recasts (e.g.Long, 2006;Lyster & Ranta, 1997;Morris & Tarone, 2003;Nicholas et al, 2001).Learners might be simply provided with the correct form without being pushed to modify their interlanguage since recasts don't elicit repair (Loewen & Philp, 2006).In addition, Loewen and Philp (2006), based on previous studies (e.g.Ellis & Sheen, 2006;Long, 1996;Long, Inagaky, & Ortega, 1998), affirm that the effectiveness of recasts depends on the targeted form under study.Despite all these limitations, bulky researches on this issue yield evidence for the positive impact of recasts on L2 learning (Ayoun, 2001;Braidi, 2002;Doughty & Varela, 1998;Han, 2002;Havranek, 2002;Iwashita, 2003;Leeman, 2003;Long, Inagaki, & Ortega, 1998;Mackey & Philp, 1998;Oliver, 1995Oliver, , 2000;;Philp, 2003).Doughty and Varela (1998) conducted a study on the effectiveness of the corrective recasts and reported that learners who received corrective recasts outperformed the control group in both oral and written measure.

Explicit feedback
As the name suggests, explicit feedback falls at the explicit end of corrective feedback spectrum.This kind of error correction therefore, is characterized by an overt and clear indication of the existence of an error and the provision of the target-like reformulation and can take two forms, i.e. explicit correction and metalinguistic feedback (Ellis, Loewen, & Erlam, 2006).In explicit correction, the teacher provides both positive and negative evidence by clearly saying that what the learner has produced is erroneous, while in metalinguistic feedback he or she only provides students with "comments, information, or questions related to the well-formedness"(p.47) of their utterances (Lyster & Ranta, 1997) .
The communicatively intrusive nature of explicit feedback amplifies the provision of both negative and positive evidence, potentially aiding learners in noticing the gap between their interlanguage and the target-like form.However, in providing the target-like reformulation, explicit error correction reduces the need for the learner to produce a modified response.Thus, explicit error correction, because it supplies the learner with both positive and negative evidence, facilitates one type of processing, the noticing of an interlanguage/target language difference, but reduces another type of processing, the modified production of an interlanguage form to a more target-like form.

Clarification Requests
Feedback that carries questions indicating that the utterance has been ill-formed or misunderstood and that a reformulation or a repetition is required are identified as clarification requests.This kind of feedback encapsulates "problems in either comprehension, accuracy, or both" (Lyster &Ranta, 1997, p.47).Clarification requests, unlike explicit error correction, recasts, and translations, can be more consistently relied upon to generate modified output from learners since it might not supply the learners with any information concerning the type or location of the error.

Metalinguistic Feedback
Much like explicit error correction, metalinguistic feedback-because it diverts the focus of conversation towards rules or features of the target language-falls at the explicit end of the corrective feedback spectrum.Lyster and Ranta (1997) categorize metalinguistic feedback as "comments, information, or questions related to the well-formedness of the student's utterance, without explicitly providing the correct form".Unlike its name, the inclusion of metalanguage is not its deterministic characteristics; rather the encoding of evaluations or commentary regarding the non-target-like nature of the learner's utterance is considered as the defining feature.Metalinguistic feedback is divided into three subcategories: metalinguistic comments, metalinguistic information and metalinguistic questions (Lyster & Ranta, 1997).
The least informative one is metalinguistic comments which only indicate the occurrences of an error.But the next subcategory, i.e. metalinguistic information not only indicates the occurrences or location of the error but also offers some metalanguage that alludes to the nature of the error.Metalinguistic questions, the last identified subcategory of metalinguistic feedback, "point to the nature of the error but attempt to the information from the student" (Lyster & Ranta, 1997, p.47).This kind of metalinguistic feedback requires learner to reconsider their assumptions regarding the target language form while metalinguistic information applies metalanguage to mark the nature of the error.

Elicitation
Elicitation is a correction technique that prompts the learner to self-correct (Panova & Lyster, 2002) and may be accomplished in one of three following ways during face-to-face interaction, each of which vary in their degree of implicitness or explicitness.One of these strategies is request for reformulations of an ill-formed utterance.The second one is through the use of open questions.The last strategy which is the least communicatively intrusive and hence the most implicit is the use of strategic pauses to allow a learner to complete an utterance.Therefore, elicitation falls in the middle of explicit and implicit continuum of corrective feedback.This kind of corrective feedback is not usually accompanied by other feedback types.

Prompt
In the related literature two other terms are used interchangeably to refer to this kind of feedback, i.e. negotiation of form (Lyster, 2002;Lyster, 1998b;and Lyster & Ranta, 1997) and form-focused negotiation (Lyster, 2002b).Lyster and Mori (2006) introduce prompts as a range of feedback types, consisting of four prompting moves: elicitation, metalinguistic clue, clarification request, and repetition.All these moves offer learners a chance to self-repair by withholding the correct form.

Repetition
Another approach to provide corrective feedback is repetition which is less communicatively intrusive in comparison to explicit error correction or metalinguistic feedback and hence falls at the implicit extreme on the continuum of corrective feedback.This feedback is simply the teachers or interlocutors' repetition "of the ill-formed part of the student's utterance, usually with a change in intonation" (Panova & Lyster, 2002, p.584).

Translation
Translation was initially considered as a subcategory of recast (Lyster & Ranta, 1997), but what distinguishes it from recast is that the former is generated in response to a learner's ill-formed utterance in the target language while the latter is generated in response to a learner's well-formed utterance in a language other than the target language.What translation and recast have in common is that they both lack overt indicators that an error has been produced.This shared feature places both toward the implicit end of the corrective feedback spectrum, though the degree to which translations are communicatively obtrusive can also vary.Translations also have another feature in common with recast as well as explicit error correction that is they all contain the target-like reformulation of the learner's error and thus provide the learner with positive evidence.

Corrective Feedback and Uptake
In studies on corrective feedback, uptake is " ..a student's utterance that immediately follows the teacher's feedback and that constitutes a reaction in some way to the teacher's intention to draw attention to some aspect of the student's initial utterance (Lyster & Ranta, 1997;p.49)."Uptake in this sense is used as a way of evaluating the effectiveness of feedback types which can be divided into two categories: "repair" and "needs repair" (Lyster & Ranta, 1997, p. 49)."Uptake in this sense is used as a way of evaluating the effectiveness of feedback types which can be divided into two categories: "repair" and "needs repair" (Lyster & Ranta, 1997, p. 49).However, several arguments question the utility of uptake, claiming that considering it as an indication of learning is not reliable (Long, 2006;Ellis, Basturkmen & Loewen, 2001).Besides, it is mentioned that a lack of response to recasts might be attributed to conversational constraints (Oliver, 1995(Oliver, , 2000;;Nabei & Swain, 2002).The third argument against the utility of uptake is that that a lack of immediate uptake does not preclude the possibility that recasts are in fact useful as the results of some surveys indicate (e.g., Mackey and Philp, 1998).The effectiveness of uptake pivots on a number of characteristics of feedbacks including: complexity, timing, and type of feedback (Loewen, 2004).

A Review of Major Studies on Error Correction in L1
Corrective feedback was initially a fertile area of research in L1 studies in which researchers probed how parents and caretakers interact with children to explore the potential effect of corrective feedback.A number of studies in this vein reveled that explicit corrective feedback on children's ill-formed utterances is rarely observed (e.g., Demetras, Post, & Snow, 1986) and that neither explicit positive reinforcement nor explicit negative comments could account for why children were so successful at learning their L1 (Brown & Hanlon, 1970;as cited in Nicholas et al, 2001).
Unlike the aforementioned studies, other studies on caretakers-child conversations (e.g., Bohannon & Stanowicz, 1988;Farrar, 1992) perceived traces of some types of corrective feedbacks like repetitions and clarification questions in adults' responses to children utterances which uphold the existence of corrective feedbacks in L1 input.Moreover, in an experimental study, Saxton (1997; as cited in Kim, 2004) found that children discerned the feedback as a comment on their ungrammatical utterances.Several other researches (e.g., Nelson, Denninger, Bonvillian, Kaplan, & Baker, 1984) also revealed that children who received negative feedback following their incorrect forms outperformed those who received the same amount of input without any kind of negative evidence.
Although these studies on error correction bear the facilitative role of corrective feedback in L1 acquisition, there are still several untouched issues in this domain that should be thoroughly investigated.
A study conducted by Chun et al. (1982) revealed that corrective feedback rarely occurred.Further investigations of teachers' corrective feedback in classrooms also suggested that feedbacks were provided erratically and often went unnoticed by students (e.g.Chaudron, 1988Chaudron, , 1986;;Fanselow, 1977).While Oliver (1995), investigating the effect of negative feedback in child NS/NNS conversations, reported NS's used 'implicit negative feedback' to 61% of the NNS's erroneous utterances and held the effectiveness of recasts.What these surveys shed light on is the existence of corrective feedback as well as their momentous role they have in the development of interlanguage.
However it should be noted here that the weakness of the most of these studies (e.g., Briadi, 2002;Ellis et al., 2001;Morris, 2002;Oliver, 1995Oliver, , 2000) ) is that the efficacy of corrective feedback has been only assessed in terms of the learners' immediate responses to the feedback.
Other studies on error correction (e.g.Allwright, 1975;Hendrickson, 1978) claim that pushing learners in their output rather than providing them with correct forms could benefit their interlanguage development.Other researchers (Allwright & Bailey, 1992) also argue that the use of corrective feedback should be delayed to trigger learners' self repair.While some experimental studies (e.g.Carroll, Roberge, & Swain, 1992) found that feedback group outperformed the no-feedback group in acquiring the targeted form under the study.
Lots of studies have been done to investigate the effectiveness of different types of corrective feedback.In this vein, lots of studies have been devoted to recast, as the most frequently used corrective feedback.The results of these surveys (Carroll & Swain ,1993;Lyster & Ranta, 1997;Panova & Lyster, 2002) revealed that although recast is the most frequent used corrective feedback, it resulted in the lowest rate of uptake whereas elicitation, metalinguistic clues, clarification requests, and repetition of error led to higher rates of uptake.However, application of recasts has its own benefits as Mackey and Philp (1998) reported a positive effect of recasts on the acquisition of question formation in English and suggested that in order for a recast to be effective, learners must have reached a stage of developmental readiness.Han (2002b) also identified four conditions that may affect the utility of recasts: (1) individualized attention, (2) consistent linguistic focus, (3) learners' developmental readiness, and (4) intensity of the treatment.
Few studies (Kepner, 1991;Robb, Ross, & Shortreed, 1986;Sheppard, 1992) attempted to investigate whether receiving written corrective feedback on the errors improve the accuracy of students' writing and proposed that there was no significant difference in the writing accuracy of the students, though Fathman and Whalley (1990) recognized that fewer grammatical errors were made by students who received error feedback.
In a quasi-experimental study, Ammar and Spada (2006) investigated the effects of recasts and prompts on L2 learners' written and oral ability across different proficiency levels and found that prompts were more effective than recasts and that the effectiveness of recasts was sensitive to the learners' proficiency level.In particular, high-proficiency learners benefited equally from both prompts and recasts, whereas low-proficiency learners benefited significantly more from prompts than recasts.In a more recent study, Lyster & Izquierdo (2009) probed the impact of recast and prompts on the acquisition of grammatical gender among adult French learners and claimed that both types of feedback are effective.Learners receiving recasts benefited from repeated exposure to positive exemplars as well as from opportunities to infer negative evidence, whereas learners receiving prompts or clarification requests benefited from repeated exposure to negative evidence as well as from opportunities to produce modified output.
Both immediate and delayed effects of two other types of interactional feedback, i.e. recasts vs. elicitations were also examined by Nassaji (2009) which signified that recasts were more effective than elicitations in immediate effects.Also the results of this study indicated that in both corrective feedback types, the more explicit form was more effective than its implicit form.Therefore the degree of explicitness was reported to be very crucial in the effectiveness of these two types of corrective feedback.

Concluding Remarks and Future Directions
This paper provided an overview of corrective feedback as one of the major issues in the domain of error correction.It elucidated different types of corrective feedback, reviewed and recapitulated the theoretical and experimental surveys on this area of language teaching in order to illuminate the significant role it has in triggering learners to notice the gap that exist between their non-target like speech and the target forms.In response to the dilemma of error correction, it can be stated that leaving students' errors untouched might lead to the fossilization of ill-formed structures.Therefore, corrective feedback can be used as an effective way in eliminating possible non-target-like utterances in the learners' interlanguage.However, it should be noted that the bulky research on corrective feedback has overlooked the fact that "learners' noticing of gaps is not a static phenomenon" (Kim, 2004, p. 19).This phenomenon has a dynamic nature and is affected by internal factors like age of leaner's as well as the external ones such as task effects.The related literature also failed to embrace the process of language learning and focused on the product hence assumed that if the trace of some erroneous structures can be identified in learners' utterances no improvement has occurred in the their interlanguage.In addition, the effectiveness of corrective feedback has been usually assessed in terms of uptake while as mentioned above, the application of uptake as a yardstick has its own limitations.Most of the studies on this issue examined the efficacy of corrective feedback in a very short period of time ignoring the fact that "[a] learner may require a certain amount of time to make use of negative input, and in the interim will continue to operate with old, as-yet-unmodified hypotheses" (Brock et al, 1986; as cited in Kim, 2004, p. 19).It can indicate that the long-term effects of corrective feedback should also be investigated.
Meta-analysis of corrective feedback revealed that despite the bulky literature on recast, there is a dearth of published discussion by ESL/EFL practitioners on explicit correction, metalinguistic feedback, and negotiation moves such as clarification and elicitation (Shaofeng, 2010).Due to this unbalanced representation, comparing the effect size of the corrective feedback types is rather difficult.Some specific moderators should also be taken into account while examining the impact of corrective feedback.To name a few of these variables, one can mention culture, age, gender, and proficiency.Investigating the facilitative impact of learner factors and corrective feedbacks can also be addressed in further research.Such learner factors include developmental readiness, language aptitude, personality factors, motivation, and attitude toward correction.Another fertile area of research that has not been comprehensively explored is the significance of alphabetic literacy level in processing oral L2 input and corrective feedback recall.
Feedback can be provided in face-to-face communication or through the computer.Though the first mode of delivery has been deeply explored, few studies have touched the second one or delved into comparing the two modes of delivery.Finally, corrective feedback through metapragmatic feedbacks and comments is another area for exploration rarely touched by SLA researchers.