«Item type text; Electronic Dissertation Authors Goertler, Senta Publisher The University of Arizona. Rights Copyright © is held by the author. ...»
One more definitional issue to be taken into consideration when connecting this study to other research on SLA is the issue of attempted correction. Panova and Lyster (2002) define uptake to include both attempted and successful uptake. Just as Panova and Lyster (2002) considered both attempted and successful uptake, this study took note of corrective feedback both when it included an attempted and a successful correction, for it is argued that both are an indication of providing feedback. In the discussion of the use of corrective feedback in chapter four, attempted and successful corrections in corrective feedback moves were listed separately so that the corrective feedback moves found can be more easily compared with the existing research.
Another issue in the definition of the corrective feedback literature is the definition of uptake, as has been discussed in chapter two. In one sense the definition of uptake in this study was narrower, yet in another sense wider than that of other researchers. This study was narrower with regard to the possible functional categories that could be categorized as uptake. For example, Lyster and Ranta (1997) included in their definition acknowledgment, repetition, and unsolicited self-repair, which were not considered as uptake here. On the other hand, Lyster and Ranta only consider uptake if it is immediately following the corrective feedback. This narrow definition was not adopted here, but rather in accordance with Mackey and Philp (1998), uptake was considered if it appeared at some point in the same conversation, i.e., the same chat transcript. As mentioned in chapter four, there were only four instances of uptake as defined for this study.
Moving on from the definitional differences and similarities between this study and the studies reviewed in chapter two, I will now compare the research results. Panova and Lyster (2002) summarized feedback applications and research as quoted in chapter two. In accordance with their statement, the teachers used various forms of corrective feedback; however, MorningTeacher had a clear preference, whereas EveningTeacher did not. In this study a relationship between error type and feedback type was not found due to the low number of feedback moves. Most research reports that recasts are the most common type of feedback. However, as mentioned before, the definitions of recasts differ. This study’s ‘repetitions with corrections’ match some other researchers’ definitions of recasts. Considering this, the present study can partially confirm the findings of other researchers (see for example Panova & Lyster, 2002). For example, MorningTeacher clearly used repetitions with correction, otherwise known as recasts, most frequently. However, EveningTeacher did not exhibit such a pattern. Panova and Lyster also discussed the problematic nature of recasts, as it can be difficult for learners to recognize them as corrective feedback rather than positive evidence. It is argued here that since several learners were able to describe the corrective feedback style of MorningTeacher, some learners over the course of the semester were able to recognize repetitions with corrections as corrective feedback moves, if they were presented consistently. Panova and Lyster (2002) also argued that recasts that reduce the utterance and add stress are more likely to be recognized. In this study those feedback forms were referred to as marked partial repetitions, which weren’t used by the teacher that frequently; however, the students used it for self-corrections. Furthermore, Panova and Lyster argued that feedback forms that involve the learner such as clarification requests are more likely to result in repair. This finding was somewhat confirmed by this study, but with only four instances of uptake present in the data set, the finding has to be seen in conjunction with the findings of others such as Panova and Lyster.
Nicholas, Spada, and Lightbown (2001) also discussed several features of effective corrective feedback as presented in chapter two. While quantitatively a difference cannot be proven, it appeared that the consistent feedback by the MorningTeacher was easier to be recognized by students based on a reading of all transcripts, which will have to be proven or disproven with a quantitative analysis at a later point. Students marked their self-corrections with an asterisk “*”, potentially to make them identifiable as feedback. Looking at the description of effective feedback put forth by Nicholas et al. (2001) and Panova and Lyster (2002), it appears that the preferred self-correction feedback form used by all students in all classes may be the most effective form of feedback. Since this form, marked partial repetition with (attempted) correction, was not used as frequently by peers or the teacher as it was for self-correction, no relationship to its effectiveness could be established in this study. However, for teacher training, one could suggest to teachers to adopt this feedback style used for selfcorrection since students are assumed to be more likely to recognize it. Not only is it a good idea for teachers to adopt that feedback style, but it would be even better if teachers would explicitly tell the students they are doing so.
As discussed in chapter four, uptake in this study was infrequent. Since uptake is seen as one measure of the effectiveness of corrective feedback, this may mean that the corrective feedback used was not effective. However, as Lyster (1998a) and Oliver (1995, 2000) point out, uptake can appear awkward in the discourse and may not always be present, even when recognition (noticing) and potential learning occurred. Instead, other measures of effectiveness should also be considered, such as noticing and language development. While Doughty and Varela (1998) found that students who receive repetitions with correction (also called recasts) showed more language progress on accuracy than a control group, the same could not be confirmed in this study. While it cannot be excluded that the feedback received by the students led to an improvement in language development, it cannot be proven either. Furthermore, since there was no significant difference between the three groups, claims cannot be made that the kind of feedback, or even the feedback in general received during chat made a difference in the students’ language development.
Overall, research on recasts seems to suggest that recasts result in a low repair or uptake rate in comparison with other feedback forms (see for example Lyster & Ranta, 1997; Morris, 2002). While it can be confirmed here that repetitions with correction had a low uptake rate, it cannot be confirmed that other feedback forms were more effective due to the infrequent uptake in general.
Looking at the amount of feedback reported in other studies, it has to be stated that feedback was less frequent in these groups and pairs comprised solely of non-native speakers, than in some of the research reported. For example, Lyster and Ranta (1997) found that 62% of errors received feedback, which is in contrast with 5.45% in this study.
Patterns between error type and feedback type as discussed in other research were not found here. The finding by Morris (2002) that morphosyntatic errors were responded to with recasts and lexical errors with negotiation moves can not be confirmed. Neither can this study confirm Lyster’s (1998a) finding that grammatical and phonological errors were corrected with recasts while lexical errors were followed by negotiation moves.
In this section some of the previous research within the interactionist framework and on corrective feedback was discussed in relation to this study. While definitional issues, as well as research findings on feedback frequency and uptake, have already been discussed, error uptake has not been mentioned yet, and will be discussed as part of the question of language use in SCMC in the following section.
5.10 Connections to CMC Research As discussed in chapter two, much of the research on CMC has compared face-toface (F2F) conversations with those in a CMC context. Several of the differences in medium have been discussed in chapter two. One of those features was the absence of paralinguistic features, discussed for example by Donaldson and Kötter (1998). Since the chatting in this study was done in the same physical environment, in contrast to the study by Donaldson and Kötter, students did not solely have to rely on the written chat medium to communicate with each other. They could see each other, though due to the screen names, they may not have known who they were chatting with. In addition, students were able to and did speak across pods to communicate with each other, especially in the ESC.
Again, this is one of the reasons why thinking about modifications of chat activities in distance learning contexts, for example to include community building, is important.
Additionally students also used emoticons, all capital letters, and extensive punctuation as other ways to compensate for lack of paralinguistic signs, as has been reported by others such as Kern (1995).
As discussed in chapter two, research methodology is problematic for CMC. As Smith (presentation at CALICO, 2005) had indicated, observations of the physical environment are crucial. While the attempt was made to follow his recommendation in this research study through the use of classroom observations and the self-report forms, there were still several methodological issues which will be discussed in more detail in a later section of this chapter and have been mentioned throughout chapter four. Since labeling of subjects was different in the self-report forms, the classroom observation notes, and in the transcripts, it was hard to triangulate these data sets. The observed chat sessions may not have been representative of all chat sessions. For example, MorningTeacher never played music when she was observed; however, in informal conversations she mentioned playing music during chatting. Additionally, while EveningTeacher explained the activities in German with a translation of key words and grammar explanations in English during the classroom observations, on another occasion she was observed introducing the entire activity in English. In addition, since the chat server did not record key strokes, not everything that students typed was recorded. Also, students were only partially found to be a reliable source of information in their selfreport forms. For example, their reports about the amount of time spent chatting were mostly incorrect.
One of the findings from early research has been that CMC encourages target language use (see for example Beauvois, 1992, 1997, 1998; Chun, 1994; Kelm, 1992;
Ene, Goertler, & McBride, 2005). Extensive use of the target language was also confirmed in this study. While in a tandem learning context, Donaldson and Kötter (1999) reported frequent code-switching, evidence for the phenomenon was very limited in this study.
Furthermore, CMC can enhance language development (Beauvois, 1992; Chun, 1994; Kelm, 1992; Kern, 1995; Warschauer, 1996). Due to the limited time spent chatting in comparison to the time spent in the traditional classroom context, such conclusions cannot be drawn from this study. However, the majority of students reported that they felt that chatting improved their language ability.
One of the concerns about CMC has been the issue of accuracy (see for example Beauvois, 1992; Kern, 1995). Research results so far have been contradictory. In this study the error rate was 4.40% and the error uptake rate was 4.40%. Ene et al. (2005) found little evidence of error uptake.
One of the underexplored issues in CMC is the use of feedback in CMC. Sotillo (2005) reported that when learners received feedback, NNSs provided explicit and NS implicit feedback. Obviously such a comparison was not made in this study, and as mentioned before, the terms explicit and implicit were not adopted here. However, for peer feedback students mainly used models, which are considered an implicit form of feedback, and the peers were NNSs. Learners received corrective feedback, but it was infrequent, and there was a difference in style and frequency between the two teachers.
No clear patterns were found between uptake and feedback type, unlike Ene et al. (2005).
However, it was confirmed that error uptake was rare and that self- and peer correction, while present, were also infrequent. Yet, as Smith (2001) argued, chatting exhibits disjointed negotiation moves, which may make it harder to identify uptake.
One of the main benefits discussed, especially in early CMC research, is the democratization (Beauvois, 1998; Kern 1995) or equalization effect (Beauvois, 1992, 1998; Bump, 1990; Kelm, 1992; Kern, 1995; Warschauer, 1996) for the teacher and students introducing topics. As discussed already, since students exhibited more words per minute than the teacher, one can argue for such an effect in this study. However, the teacher-only functions used such as the “invisible” and the “to all” functions may arguably not contribute to such a democratization. While in MorningTeacher’s case it was confirmed that most of the topic initiation came from the students and not the teacher, this was not the case with EveningTeacher.
Donaldson & Kötter (1999) reported that the teacher was seen as an intruder, which was not the case in this study. Students actually reported preferring active participation from their teacher. No silencing effect due to feedback style was found either, in contrast to the results from Ene et al. (2005).
An affective benefit of CMC has been the increase of student motivation (Batson, 1998; Beauvois, 1992, 1998; Bump, 1990; Donaldson & Kötter, 1999; Kelm, 1992). In this study no direct measures of motivation were made. Nevertheless, I can conjecture that, since the majority of students reported enjoying chatting and also saw it as beneficial, SCMC may be a factor contributing to an increase in motivation. Furthermore, the MorningTeacher reported that while in the past the students always disliked the lab sessions, now that they included the chat activities, students liked lab day, also giving credence to the idea that SCMC may contribute positively to student motivation.
One of the logistical issues and differences between asynchronous and synchronous CMC discussed was the use of resources. Abrams (2003) argued that in comparison to ACMC, the use of resources was cumbersome in an SCMC environment.
This cannot be confirmed by this study. The students used online and physical resources while chatting, according to my classroom observation notes, and in their own reports.