«Item type text; Electronic Dissertation Authors Goertler, Senta Publisher The University of Arizona. Rights Copyright © is held by the author. ...»
The most controversial form of feedback according to the literature appears to be recasts. Although recasts have been studied extensively, as will be shown later, and it may be that the inconclusive research surrounding this form of feedback can be attributed to varied definitions (for an overview of such definitions see Nicholas, Lightbown & Spada, 2001, p. 733). Morris (2002) summarizes features of recasts as follows: “… (1) they reformulate an ill-formed utterance; (2) they expand the utterance in some way; (3) the central meaning of the utterance is kept, and (4) recasts immediately follow the illformed utterance.” (pp. 396-397). A similar description was provided by Nicholas, Lightbown and Spada (2001) who summarized recast features as acknowledging content, providing positive affect, and modeling the correct form. Recasts are an implicit form of feedback, but also provide positive evidence in the reformulation and are claimed to promote noticing through the contrasting of the ill-formed utterance and the reformulated target language utterance (Saxton, 1997). The interpretation presented here claims that recasts are negative evidence (see also Doughty & Varela, 1998; Long & Robinson, 1998; Oliver, 1995), as they are used to indicate to the learner that a mistake was made.
In this study, investigating the use of corrective feedback by teachers, it is assumed that teachers use recasts intentionally to correct their students’ utterances. In contrast, Lyster (1998b), for example, considers recasts to be more like positive evidence than negative evidence because they do not reveal the error. Even if one disagrees with Lyster and argues that recasts are negative evidence, the further question becomes whether recasts are implicit feedback. Recasts can be marked with overt signals such as change in intonation, gestures, or, in the case of writing, with some written symbols, and then could arguably be considered explicit negative feedback (Nicholas, Lightbown, & Spada, 2001). Ene et al. (2005) found that teachers and students used a “*” symbol to mark any form of correction during online chat. From this example, the question then arises whether or not implicit feedback moves become explicit once they are marked with such a symbol, especially since this appears to be an established tradition in the messaging and chatting community. The challenge then is to determine, when a signal is overt enough to qualify a corrective feedback move as explicit.
When looking at corrective feedback one also has to consider what happens after the correction. Uptake is seen as one possible reaction to corrective feedback. Panova and Lyster (2002) define uptake as “…different types of student responses immediately following the feedback, including responses with repair of the non-target items as well as utterances still in need of repair” (p. 574). Lyster and Ranta (1997) also differentiated between two forms of uptake: uptake with repair and uptake which needs repair. Uptake with repair is an utterance following the correction that no longer contains errors. Uptake which needs repair is the unsuccessful attempt made after a correction to correct one’s own ill-formed sentence. In addition, in Lyster and Ranta’s, definition acknowledgment of the corrective feedback, repetition by the learner, and self-repair are considered uptake if they immediately follow the feedback. Mackey and Philp (1998) argue against this narrow definition of uptake, suggesting that change often does not occur until later in the conversation, suggesting a redefinition of uptake to include instances of repair or attempt to repair later within the same conversation. But even with this wider definition, it is still possible that the effect of a correction may occur even later, hence causing some methodological issues for using uptake as a measure of effectiveness. Thus, the challenge from a research analysis standpoint is to determine how long after received feedback acknowledgment of the corrective feedback, repetition by the learner, and self-repair may be considered as uptake. One must also acknowledge that while uptake can be seen as a successful corrective feedback move, it would be wrong to assume that a corrective feedback move that is lacking evidence of uptake was not successful, since some learners or some discourse contexts may make it unnecessary, or even linguistically awkward, to provide any proof of uptake, as it may interrupt the flow of the conversation. Therefore, more than one measure of the effectiveness of corrective feedback must be used in such research.
2.3.3 Research Results Panova and Lyster (2002) summarized corrective feedback applications and
research as follows:
(1) Teachers have at their disposal a wide variety of corrective strategies to focus on learner errors. (2) Choice of feedback type can be dependent on type of error. (3) Recasts are the most widely used type of feedback in the observed classrooms. (4) The discourse functions of recasts may lead classroom learners to confuse recasts with positive feedback moves. (5) Learner repair immediately following feedback can be either repetition or learner-generated repair, depending on the type of feedback used. (6) In comparison with other feedback types, recasts do not promote immediate learner repair, which, in the case of recasts, involves repetition. (7) Recasts that reduce the learner’s utterance and add stress to emphasize the corrective modification are more effective at eliciting repetition of the recast and are more likely to be identified by learners as corrective feedback. (8) The corrective techniques of clarification requests, elicitation, metalinguistic feedback, and repetition of error correlate more positively with learner uptake and immediate repair, and, in these cases, the repair is learner generated. (9) Learners claim to notice forms that they are pushed to self-repair more than forms that are implicitly provided by teachers (p.577-578).
Nicholas, Spada and Lightbown (2001) identified similar features of successful error correction: (1) there has to be consistency in feedback patterns, (2) feedback has to be recognizable as such, and (3) additional cues such as non-verbal cues may be necessary to better mark corrective feedback.
The effectiveness of corrective feedback has been measured in several ways.
Some studies have investigated improvement in language proficiency or developmental stages in relation to the corrective feedback received. Others have evaluated the effectiveness of corrective feedback through an analysis of the discourse in which they focused on self-repair. Still others have used noticing measures such as uptake, repetition, stimulated recall, or private speech as evidence for potential language development and development opportunity.
Lyster and Ranta (1997) did not consider uptake evidence for learning, but they did suggest that uptake with repair was beneficial for second language development.
Generally speaking, although repetition, self-repair, and uptake may be considered measures for noticing, or even evidence for second language development, it is possible that they may not be present, if the discourse may be made linguistically awkward should the utterance be repeated (Lyster, 1998a; Oliver, 1995, 2000), thus failing to show that the utterance was noticed, and failing to show sign of these measures. Hence, while uptake and evidences of noticing have been used as evidence for language learning opportunities, their effectiveness has not been established and additionally their absence may not mean a lack of learning opportunities.
In addition to understanding which evidence has been used to measure effectiveness of corrective feedback, it is also important to understand what feedback types are used by teachers and interlocutors. Early research on corrective feedback was descriptive in nature, and found that teachers employed varied forms of corrective feedback, some of which were not always easy to identify by learners. Doughty (1994, as cited for example in Nicholas, Lightbown & Spada, 2001) found that 40% of students’ utterances, whether correct or incorrect, received feedback from the teacher. The most common feedback forms were clarification requests, repetitions, and recasts. If the utterance was correct, it was likely repeated. In utterances in which there was a single error, 68% received a recast, and 23% received a clarification request. This frequency ranking was confirmed by Oliver (1995), and by Lyster and Ranta (1997), who found that 62% of the erroneous student utterances received feedback. Lyster and Ranta (1997) also found that recasts were the most commonly used feedback move with three of the four teachers considered, while the teacher instructing an advanced class used fewer recasts.
Oliver (1995) studied corrective feedback in NNS-NS dyads of teenagers, and found that NSs used recasts, repetition, clarification requests, and comprehension checks. Similar to the other studies mentioned here, 65% of errors received implicit negative feedback, although recasts were not the most frequent. Therefore, forms of feedback that impact the flow of the conversation minimally appear to be preferred by teachers and interlocutors.
This preference does not mean that implicit feedback is more effective.
The effectiveness of different feedback types have been investigated. Since recast appears to be the most frequently used feedback form, it is important to review the research on its effectiveness. Empirical research studies have also investigated the effectiveness of recasts for language learning, and they have been shown to be of benefit in L2 instruction (Doughty & Varela, 1998; Long, Inaki & Ortega, 1998; Mackey & Philp, 1998). Long, Inaki and Ortega found recasts to be more effective than models of the correct structure in the short term, although their study did show much individual difference. In a study of young adults, Doughty and Varela (1998) found that students receiving corrective recasts, these being repetitions followed by recasts, achieved better progress on accuracy and development than did a control group. The developmental stages were evaluated through an interlanguage analysis of pre- and post-tests design to force subjects into using the past tense. The analysis investigated participants’ abilities to produce the past tense, specifically to determine at which interlanguage stage the students were at for this specific structure. The problem with research studies focusing on developmental stages is that little is known about them, especially in the acquisition of languages other than English. In addition, some researchers caution that recognizing the recast as a correction may be necessary for language development, in that the difference in learner utterance and recast may not be noticed by the learner. Saxton (1997) claimed that first language learners would be able to notice corrective feedback because of the difference between the recast and the original sentence; he referred to this as the direct contrast hypothesis. Long and Robinson (1998) also stressed the importance of attention to form for adult language learning, and thought that recast would be ideal for such a focus. Nicholas, Lighbtown and Spada (2001), however, argue that recasts are difficult to identify in L2 contexts, even though recasts can be more effective if the focus of instruction is on form. Doughty & Varela (1998) found that repetitions and recasts did not interfere with a focus on meaning in the classroom, and thus they were an unintrusive, low affective filter form of corrective feedback. However, they were skeptical in regard to the effectiveness, and suggested that recasts are best utilized as corrective feedback when focusing on one form at a time, especially when the recast is emphasized with additional cues. Long, Inaki and Ortega (1998) further cautioned that effectiveness of recasts may be dependent on the language structure. In conclusion, while recasts appear to be the most frequently used form of feedback, research has not confirmed their effectiveness.
A partial explanation for the possible ineffectiveness of recasts is that recasts could be interpreted by learners as a conversational reply, confirming the content of the message rather than expressing a disapproval of the structure of the message (Schachter, 1981). In Lyster and Ranta’s (1997) study, only 5% of the recasts following a correct utterance led to uptake, versus 31% of the recasts following incorrect utterances, suggesting that learners may have perceived the difference between the recast with correction and recast without correction. This is based on the assumption that only recasts to incorrect sentences require uptake in the discourse. However, this suggested ability to differentiate between corrective moves and non-corrective moves does not necessarily result in improvement. This is true especially since Lyster and Ranta (1997) also found recasts to be the least effective of various forms of corrective feedback, and Panova and Lyster (2002) found little uptake following recasts. Lyster (1998b) agreed that recasts were hard to perceive by the learner, explaining that recasts to grammatically correct or incorrect utterances seemed to have the same frequency, which is different from L1 contexts. Hence, the learner may not understand the difference between a recast used as a form of negative evidence and a recast used as positive evidence, creating ineffective correction for the learner. Lyster also found that recasts were often accompanied with approvals of the meaning, further complicating the identification process of recasts as corrective feedback, especially since learners do not know if the meaning is being approved or if their language is being corrected. Accordingly, Lyster (1998b) also argued that recasts are hard to identify due to their ambiguity and similarity with repetitions of well-formed student utterances. In conclusion, the unintrusive nature of recasts, which may make it the preferred choice by teachers, may be the reason for the difficulties in proving their effectiveness.
Before discussing the effectiveness of all feedback forms, it is important to review relevant definitions. Lyster’s (1998b) definitions of corrective feedback types are often
used or referred to in related studies as follows: