THE ROLE OF RECAST AND UPTAKE IN ESL TEACHER-LEARNER CONVERSATIONS

The study examined the role of recast and learner uptake following corrective recast in a group setting of teacher-learner conversations. Three female Universiti Teknologi Malaysia students in their twenties participated in the study. The relationship among types of uptake, linguistic focus of recast and the nature of recast were examined. The interactions between the female teacher and the group of female learners in a semi-formal interaction setting were recorded using an audio recording device. The recorded interactions were transcribed and coded for the types of learner uptake (Correct Uptake Episode (CUE), No Uptake Episode (NUE) or Incorrect Uptake Episode (IUE)), linguistic focus of recast (Morphosyntactic / Lexical) and the nature of recast (length of recast / degree of change / saliency of recast). Findings indicate that learners provided CUE and NUE in response to the teacher recast. Lexical recast has appeared to elicit more correct uptake by the learners. Recasts that are shorter, and involved one or two changes in the form of Repetition and Reformulation were found to be more assessable by the learners and more effective in generating successful repair.


Introduction
English teachers encounter quite many students making mistakes in their language production on daily basis, especially in the second language classroom.Almost all of the teachers would treat these errors by providing corrective feedback to the students.Thus, a response provided by a teacher or another interlocutor that attempts to signal to a speaker (usually non-native, NNS) of the incorrectness/ungrammaticality of his/her utterance is called corrective feedback.
Due to its importance, there is an increasing number of studies within the field of second language research (SLA) that focus on corrective feedback.A number of descriptive studies focusing on the role of corrective feedback in SLA have emerged based on data collected in classrooms, such as Panova & Lyster, 2002;Sheen, 2004, and also on data collected in a quasi-experimental setting (e.g., Iwashita, 2003;Mackey, Oliver & Leeman, 2003;Philip, 2003).
The stated studies have investigated the different types of corrective feedback received by learners and the extent to which corrective feedback is noticed, and/or up taken, by the learners.Further studies by Han (2002), Leeman (2003) and Lyster (2004) have been carried out to examine the function of corrective feedback in second language acquisition, addressing the relative efficacy of implicit and explicit corrective feedback.
One of the reasons for the growing interest in corrective feedback, as stated by Ammar and Spada (2006), is related to the observation of the difficulties experienced by Second Language (L2) learners in terms of accuracy, particularly in syntax (Schmidt, 1990) in communicative classrooms although they have acquired comprehension ability and some fluency in producing oral communication.According to Doughty & William (1998) and Long & Robinson (1998), these low levels of accurate grammar production show some evidence of insufficiency of comprehensible input and exclusively meaning-based instruction.Schmidt (2001) echoes this view in his noticing hypothesis by pointing out the need of getting learners to pay special attention to the formal properties of language in order to notice L2 forms.Consequently, form-focused instruction has been proposed as a method to draw students' attention to language form within communicative classrooms (Ammar & Spada, 2006).

Background of the Study
Second Language Acquisition (SLA) researchers have been investigating the role of corrective feedback in second language (L2) classrooms since the 1970s, based on the premise that learners benefit from information about the communicative success of their target language use (Long 1977).Long has also stated that when learners are not able to discover errors made through exposure to positive evidence, they might require some explicit feedback on the errors.This feedback, which normally occurs in interaction, plays a role in driving L2 development forward because learners rely on semantically contingent speech as a primary source of positive and negative L2 data (Long, 1996).Long (1996) suggested that the learner must not only notice the negative evidence but must also perceive it as negative evidence in order for the negative evidence to be useful.Long (1996) has also pointed out other issues regarding usefulness and usability of recasts.For instance, the learner must be able to accurately identify the error that is being corrected and hold the negative evidence in memory and compare it to the original error for it to be useful.
Farrar's study offers some evidence that children were not only able to notice the corrective feedback, but also the response to it.For example, Farrar (1992) has reported some proof of the usage of recasts, in which children were more likely to copy parental responses after corrective recasts.Moreover, Farrar (1990) found that recasts were found to be associated with children's acquisition of some morphological forms, such as plurals and progressives.Therefore, Farrar's study has exemplified Pinker's criterion of use; "in order to play a role in language acquisition, corrective recasts must meet several criteria: it has to exist, be useful, used by learners and is necessary for acquisition to occur" (Grimshaw & Pinker, 1989 pp. 19-20).Given the myriad of issues of implicit negative feedback, the usability of recasts in interactions is still open to exploration.

Statement of the Problem
There are several reasons why recasts have received so much attention from researchers.First, recasts have been found to occur more frequently than any other type of corrective feedback in a range of natural L2 classroom settings, such as elementary immersion classroom (Lyster & Ranta, 1997;Mori, 2002), university-level foreign language classrooms (Doughty, 1994;Roberts, 1995), high school English as a foreign language (EFL) classrooms (Tsang, 2004), and adult ESL classrooms (Ellis, Basturkmen, & Leowen, 2001;Panova & Lyster, 2002).
However, these researches have been producing varying results.While some have shown tremendous benefits of recasts to learners, others have suggested that other types of corrective feedback might be more effective than recasts.The different gap of the results somewhat shows that researchers and teachers are still unsure about the extent to which recasts are beneficial to L2 learners.In addition, the methodological aspects of previous studies are also in question, whether uptake and other measures of noticing can be used as valid measures of the beneficial role of recasts (Loewen, 2005;Loewen & Philp, 2006;Mackey & Philp, 1998;Nicholas et al., 2001).
There is also concern that studies that use pre-selected target linguistic forms and/or intensive recasts may increase learners' attention to recasts and the forms targeted by the recasts more than would naturally occur in L2 classrooms (Ellis & Sheen, 2006;Nicholas et al., 2001).Finally, many researchers have questioned the applicability of results from dyadic interaction studies to real L2 classroom situations (Ellis & Sheen, 2006;Loewen, 2005;Loewen & Philp, 2006;Lyster, 2004).
Furthermore, in the context of ESL learners, there is a need to obtain a more precise understanding of how these learners respond to input in the form of recast during semi-formal interaction.It is also important to examine the nature of recast as well as its linguistic focus and its effect on learners' successful uptake.This is because correct uptake may be important due to the fact that they provide opportunities for learners to practice using the language and to produce target-like output (Swain, 1995), which could create ideal conditions which would foster language acquisition.In light of these real concerns, there is a need for studies that examine recast while addressing these issues.

Purpose of the Study
There are many interaction-based studies which aim to provide explanations on L2 development and its connection with conversational interaction.However, according to a review by Lyster & Mori (2006), the cognitive mechanism which is perceived to be instrumental in second language development might be caused by a specific type of interactional feedback more effectively in different settings.
In response to these claims, some studies (Oliver & Mackey, 2003;Pica, 2002;Sheen, 2004) have investigated the role of interactional context, as a factor that affects the L2 development and acquisition opportunities provided.For instance, some classroom studies done by Ellis et al (2001) and Mori (2002) have produced results showing a tendency for learners to repeat recasts; meanwhile, in other researches, recasts were not found to be effective at producing uptake (Lyster & Ranta, 1997;Panova & Lyster, 2002;Tsang, 2004).In light of the different findings, this study attempts to discover patterns of corrective feedback, namely recasts, uptake and learner repair in a semi-formal conversation outside the classroom setting.
As noted by Long (1996), the occurrence of overt error correction is rare in conversation settings, due to the nature of conversation itself.Therefore, the status of negative feedback in this kind of interaction, whereby it lacks pedagogical focus, would offer insights into its use and role.Specifically, the study investigates the relationship between recast and learner uptake in semi-formal teacher-student interaction settings.

Review of literature
There are six forms of feedback, such as recasts, explicit correction, metalinguistic feedback, clarification request, elicitation, and repetition, as identified by Lyster and Ranta (1997).It was found that the most predominant feedback type was recasts and the most recurring error types were grammatical errors.Similar results were gained in a study done by Seung and Shaofang (2012), where the most recurring errors and also the most feedback given were regarding grammatical errors.
Uptake is used to measure the effectiveness of feedback because they may serve as evidence of the noticing and integrating of the corrective feedback as stated by Egi (2010) and Lyster & Ranta (1997).
There was also a study that investigated the relationship between types and distribution of corrective feedback and their effect on learners' uptake in Iranian adult EFL classrooms.The study that was conducted by Mohammad Rahimi & Arezou Sobhani (2015) used a framework that was adopted from Lyster and Ranta's (1997) analytic model.The results from the study showed recast as the most frequent error feedback type provided to learners in all proficiency levels.They also added that the uptake by students was led by corrective feedback in the form of elicitation and request for clarification (Rahimi & Sobhani, 2015).
There are a lot of studies which focus on the role of recasts and uptake in L2 learning.For example, Braidi (2002) examined the availability and usage of recasts in adult native-speaker/non-native-speaker two-way interactions in a non-classroom setting.There were three types of negotiations that were the focus of the study, which are one-single negotiated interaction, extended negotiated interactions and non-negotiated interactions.The other focus of the study was the provision of recasts following the different levels of ungrammaticality of the non-native speakers' utterance, which were divided into two; single and multiple errors.
The results from four sets of interactions outside the classroom showed that NS responses to NNS utterances were higher in non-recasts than recasts in all different types of negotiation and grammatical errors.Nevertheless, there were highly significant differences among the occurrences of recasts in the different types of negotiation.
Specifically, the provision of recasts is higher in extended negotiations, in which the interlocutor signals lack of comprehension repeatedly.Based on these findings, it shows that recasts that occur in NS-NNS interactions are affected by types of negotiation and by levels of grammaticality.For instance, recasts received more attention in the input when the interaction becomes more difficult by the interlocutor.The results also show some significant differences to other studies that have been done; recasts occur in different patterns from what others have reported.Nevertheless, the rate of the usage of recasts remained similar to previous studies.
From the results of the study, Braidi (2002) has concluded that explicit negative feedback in the form of recasts is available by the instructor when the learners possibly need them, as in the extended negotiation where the communication has broken down.This conclusion was supported by Long (1996), claiming that negative feedback such as recasts might be useful because they occur "when the NNS is likely to be attending to see if a message gotten across, and to assess its effect on the interlocutor" (p.429).Even though recast has been upheld as one of the corrective feedback forms that could offer beneficial effects to L2 learning, it has to be salient to the learner in order for it to be of positive effect.Therefore, Leeman (2003) has carried out a research to discover the effectiveness of recasts through saliency.She examined the effect of multiple variables combined in recasts for L2 Spanish learners engaged in communicative interaction.There were four groups involved in the study, which were negative evidence group, recasts group, enhanced salience with positive evidence group (with no negative evidence), and un-enhanced positive evidence group, which represents the control group of the study.
Leeman found that learners who had received either recasts (negative evidence and enhanced-saliency) or enhanced-saliency treatments outperformed those who had received either negative evidence alone or no increased saliency (control group).Leeman concluded that recasts can indeed be beneficial to students, even on forms with low communicative value.She further concluded that recasts may mainly be beneficial due to their increasing the saliency of the target form.This is because; recasts can also provide positive evidence besides being implicit negative feedback (Ammar & Spada, 2006).They suggest that recasts, as enhanced salience of positive evidence, can be provided by means of stress or emphasis.It is clear that recasts are beneficial as they do not only reformulate the non-target-like utterance, but also provide the correct model of use for language learners.
An experimental research was conducted to compare the effectiveness of intensive and extensive recasts on the acquisition of a planned target structure by Nobuhiro Kamiya (2014).One of the results showed that the groups receiving recasts demonstrated higher accuracy on both implicit and explicit knowledge tests as compared to the control group.The research also found the accuracy effect and it was even apparent in cases of a structure that the students were unfamiliar with (Kamiya, 2014).
In order to extend the knowledge on the effects of recasts, some research studies focus on pre-selected target items such as English question formation.Mackey and Philp (1998) conducted a study on the effect of intensive recasts on question development in dyadic interaction between NS-NNS of English.The students were divided into three groups: negotiated interaction, recast, and control.The two treatment groups were then further divided into "readies" (higher level students) and "unreadies" (lower level students) based on their performance in the first session (pre-test).After eight sessions (five treatments and three tests), results showed that the "readies" in the recast group outperformed the "readies" in the negotiated interaction group.The same was not true for the "unreadies".Mackey and Philp concluded from this that recasts may be effective if and only if the students are developmentally ready to receive the information in the recasts.
Besides studies in dyadic interaction setting, a number of researchers have chosen to undertake recast studies in small-group and classroom contexts.Research projects conducted within small groups of students are often done in an attempt to mimic the way in which group activities are conducted in L2 classrooms.For example, Han (2002) conducted a small-scale study with eight adult ESL students who were divided into two groups: recasts and no feedback to participate in eight small-group treatment sessions with the researcher.The pre-selected linguistic target was tense consistency.The instruments of the study consisted of written and oral narration tasks.
Recasts were provided to the recast group during oral interaction when they were not being tenseconsistent.The results of the oral tests showed that overall, both groups tended to use more past tense as the experiment progressed.On both written post-tests, the recast group noticeably increased their tense consistency, enlarging the gap between the two groups.Han concluded from this that the recasts impacted the students in a positive way and led to the students' better control over tense consistency.Han also proposed that recasts may have heightened learners' awareness of the need for tense consistency.

Methodology
Data collection for the study took place approximately 8 to 10 weeks in one semester.The data collection was carried out over 3 discussion sessions for approximately one hour of interaction per discussion.Recording discourse was the principle data gathering technique for the study.The semiformal interaction sessions were audio recorded using an audio-recorder.Recorder was kept on desk while the interaction was ongoing.In order to allow students to be familiar with the recording, the recording started while they had informal conversation before the actual data collection procedures.All the recorded data were independently transcribed by the researcher and checked by another teacher to ensure the reliability of the data.
The audio-recorded data was transcribed following conventions of conversation analysis (e.g., Ten Have 1999).Data sources from recordings of communicative interactions and transcriptions of recordings were analysed according to the following process.The analysis process commenced after the first recording of the semi-formal conversational interactions.All audio-recorded conversational discourse was transcribed immediately after conducting them.The researcher listened to the audiorecorded conversations repeatedly and checked each transcription several times before coding and categorizing.This method was followed by another individual coder to ensure the reliability of the data.
The data were analysed by identifying the recast episodes occurring during the conversational interaction.This was done on the basis of transcriptions which were checked by the researcher and rechecked by another teacher of English in identification of recasts and uptake episodes.Coding and categorization of the data were carried out initially by the researcher, followed by validation by another English teacher.Data coding for the present study employed several stages, which were coding for recasts, coding for the characteristics of recasts and coding for learner's uptake.
Data were categorized and coded in terms of recasts in response to non-target-like utterance and learner uptake.Non-target-like utterance was classified into two categories: morphosyntactic (grammatical features) and lexical errors.The characteristics of recast (degree of change, saliency and length) were also taken into considerations in order to see the relationship of these characteristics with learner's uptake.Learner's uptake was analysed in terms of three main categories: Correct Uptake Episode (CUE), Incorrect Uptake Episode (IUE), and No Uptake Episode (NUE).

Types of uptake
As it was mentioned earlier, one of the purposes of the study is to examine the different types of uptake move provided by English learners in respond to teacher recast.The findings of the study revealed that there were two types of uptake move made by the learners.The learners successfully incorporated teacher recast (Correct Uptake Episode) or they might not notice the corrective recast, therefore there was no successful uptake (No Uptake Episode) in the segment of interaction.From the analysis of data, there was no inaccurate uptake made by the learners.Hence, the analysis presented in this section is comparing the nature of recast and its relationship between learner's Correct Uptake Episode and No Uptake Episode.The analysis of three group interaction sessions showed that a number of 107 uptake moves were provided by the learners in response to teacher recast.Table 4.1 presents the distribution of learner uptake in each session of interaction.The findings revealed that the interaction elicits two types of responses, Correct Uptake Episode (CUE) (57%) and No Uptake Episode (NUE) (43%) when provided with recasts.The analysis revealed that there was no unsuccessful attempt of incorporating recast or inaccurate uptake episode occurred in the study.Successful immediate incorporation or Correct Uptake Episode (CUE) and unsuccessful incorporation or No Uptake Episode (NUE) occurred due to the nature of recast, which will be discussed further in this section.
Other than factors such as the nature of recasts, there are several possibilities that could be considered to explain the higher rate of CUE in the interaction.Firstly, learners' limited linguistic ability might have resulted in more successful immediate incorporation of the teacher's recast.That is, learners might tend to rely more on the teacher to reformulate their original utterances because they are less able to access the target-like forms themselves (Mackey et al., 2003).Thus, the learners in the present study could have used the teacher's recasting as a source to formulate their utterance into target-like form.It might also been due to the fact that these students are adults studying at tertiary level.According to Loewen (2004), adult learners might be more likely to produce uptake and successful repair than school-aged children (Lyster and Ranta, 1997;Oliver, 1995).

Learner uptake and linguistic focus of recast
Another objective of the study was to examine the relationship between learner uptake and the linguistic focus of recast provided by the teacher.The study is focusing on two types of linguistic focus of recast which are morphosyntactic and lexical focus.The findings showed that the interaction elicits successful immediate incorporation when provided with recast that focused on lexical item in comparison to morphosyntactic item.These results are in support of other studies such as Mackey et al. (2000), where learners recognized lexical feedback with relative accuracy whereas they were less likely to recognize the targets of morphosyntactic recasts.Lyster (1998) and Sheen (2006) discovered that lexical recasts lead to a higher frequency of uptake than morphosyntactic-focused recasts.According to a study by Carpenter et al. (2006), learners might be more aware of lexical recasts whereas morphosyntactic recasts might not be noticed by the learners at the same rate of noticing lexical recasts.In his data, lexical recasts were generally shorter than morphosyntactic recasts.Therefore, lexical recasts are associated with the factors that contributed to enhanced saliency of recasts which are recasts that are generally short and involved one change (Sheen, 2006).Due to these factors, lexical recasts elicit more successful immediate incorporation or Correct Uptake Episode than morphosyntactic recasts could do, which might also explain the results of the present study.
The nature of recast affected the nature of uptake Accurate uptake of recast was constrained by the nature of recast itself.In the study, there are three kinds of the nature of recast that are focused on.The effect of the length of recast, degree of change and saliency of recast on learner successful incorporation of corrective recast were studied.The influence of these variables on learner uptake is discussed in terms of Correct Uptake Episode following the respective nature of recast.As the graph shows, almost all recasts at word level (95%) were accurately repeated by the learners in comparison to only 67% of CUE occurred for recasts at phrase level and 17% for recasts at sentence level.
The results showed that recasts at a shorter level tend to foster more noticing and successful learner repair than recasts at a longer level in T-ESL interaction.It is clear that shorter level recasts were more likely to benefit learners to generate more learner uptake whereas recasts incorporated into longer sentences rarely lead to repair (Lyster, 1998).This result is corresponding to the argument from previous studies that when recasts have been shortened, they led to substantially more frequent repair (Sheen, 2006).
Other than that, the results could also be argued in term of the limitation of attentional capacity.Cowen (1988) suggests that units of information may be held in working memory for about 15 to 20 seconds.
Therefore, what remains in working memory is a factor which influences the rate of recalling the recasts provided, which might explain that the longer the recasts, the less ability to elicit correct uptake by the learners.Overall, the present study gives evidence to support the argument from previous studies that proposed that recasts that have fewer words might be more easily noticed by the learner.
The other factor that may influence learner uptake is the degree of difference between recast and the learner's non-target-like utterance.Research by Long (1996) suggested the level of difference between recast and learner's original utterance has some effects on learner's immediate incorporation.If recasts are too different from learner's non-target-like utterance, they may be unlikely to be incorporated in the utterance, as they are too far removed from the triggered utterance.
From the analysis of the data, it appeared that the teacher provided recast with three kinds of changes.
Recast provided by the teacher in the study either involves one change, two changes or three or more changes.The degree of difference between the learner's non-target-like utterance and the recast provided in response was another potential constraint on learner uptake.The results of the interaction indicated that learners performed better the fewer the changes to the trigger utterance.As shown in figure 4.5, the lesser the changes of recast from the original utterance, the higher the successful immediate repair by the learners.There was a significant difference in Correct Uptake Episode (CUE) with only one change (73%) compared to recasts with two changes (47%) and recasts with three or more changes (0%).

Percentage of CUE
According to Alan (1997) and Hatch (1983), it is apparent that the saliency of recast relates to the nature of learner uptake.In this section, the saliency of recast will be discussed in terms of two different types of recasts, Reformulation "R" and Repetition+Reformulation "R and R", since each recast form used and provided by the teacher in the present study has a characteristic saliency which influence learner uptake.
It is evident that more successful immediate repair by learners occurred when recast in the form of Repetition and Reformulation were provided.From the graph above, a significant number of "R and R" recasts (82%) was correctly repaired by the learners in comparison to only 39% of "R" recasts which were successfully noticed and repaired.This may have been due to the fact that recasts in the form of Reformulation were likely to be perceived as agreement with and confirmation of information about previous utterance or mere repetition.

Conclusion
The present study investigated the relationship between recast and learner uptake in semi-formal teacher-students interaction setting.Three issues were examined: the types of uptake move made by the learners when provided with recast, the relationship between learner uptake and linguistic target of recast, as well as the effects the nature of recast have on the nature of learner uptake.
The analysis of the data demonstrates that learners in semi-formal group interaction were able to notice the corrective recast provided by the teacher and most of the time were able to successfully incorporate them in their utterances.The corrective feedback in the form of recast is more beneficial when it is made saliently to the learners.
Thus, it can be concluded that recast can be of benefit and offers a less obtrusive way of modelling target-like form to adult learners at tertiary level who are strongly motivated to learn and practice English by social and personal demands, especially in communicative-based setting.The results of the study could be used to expand the knowledge of relevant contextual variables that influence English learners, especially in interaction setting.The variables could be some of the elements that are vital and crucial for second language development.

Percentage of CUE
Reformulation ("R") Repetition and Reformulation ("R and R") The study also provides some insightful analysis for English learners and instructors to discover the potential of semi-formal conversation outside the classroom setting towards language advancement.Moreover, the findings of the study could provide a better understanding of the relevance of classroom interactions between teacher-students and among students.
Figure1: Percentage of CUE in relation to linguistic focus of recast

Figure 2 :
Figure 2: Percentage of CUE in relation to the length of recast

Figure 3 :
Figure 3: Percentage of CUE in relation to degree of change of recast

Figure 4 :
Figure 4: Percentage of CUE in relation to the saliency of recast (T-ESL interaction)

Table 1 :
Distribution of types of uptake move made by the learners