«Item type text; Electronic Dissertation Authors Goertler, Senta Publisher The University of Arizona. Rights Copyright © is held by the author. ...»
As has been discussed in previous sections, teacher feedback was infrequent in all case study subjects’ chat transcripts. While often research discusses difference in feedback style as implicit and explicit styles (see for example Panova and Lyster, 2002), those differences were not found in this study. In the case study subjects’ transcripts, the MorningTeacher had a clear preference for one feedback form and the EveningTeacher did not. I will refer to this as a systematic and an unsystematic feedback style. However, again, it needs to be pointed out that this was only the feedback used in the case study subjects’ transcripts in response to errors made by any participant in those chat transcripts. Though no feedback count was established for all other transcripts, a readthrough of the other transcripts suggested that these feedback styles might be consistent across subjects.
First, I will present the word count and words per minute for all 46 subjects (see table 4.18.), and then discuss a possible difference between groups, or students taught by the two different instructors.
Table 4.18 Student Output
Table 4.18 illustrates the student output in the target language.
In the NSC students’ average words per minute ranged from 2.22 to 5.90 with an average of 3.64 and a median of 3.47. In the SSC students’ average words per minute ranged from 1.56 to
5.56 with an average of 3.32 and a median of 3.19. In the ESC students’ average words per minute ranged from 2.03 to 6.90 with an average of 4.01 and a median of 3.86.
Teacher – Student Output: Test of Between-Subject Effects
To establish significant differences between the two teachers, the data were analyzed using a one factor between subjects ANOVA, with teacher as the factor with the levels MorningTeacher and EveningTeacher. The dependent variable is the average words per minute produced by the students. As table 4.19 shows, the main effect of teacher is not significant for amount of student output (F(1,44)=2.42, p.05).
Since there are three classes, and there might be a difference between the three classes, the data were analyzed using a one factor between subjects ANOVA, with class as the factor and NSC, SSC, and ESC as the levels. The dependent variable is the average words per minute produced by the student. As table 4.20 illustrates, the main effect of class is not significant (F(1,43)=1.44, p.05).
Table 4.20 Class – Student Output: Test of Between-Subject Effects
In conclusion, the teacher did not have a significant effect on students’ average words per minute, despite the fact that in the case study subjects’ transcripts a difference in teacher feedback style was discovered. Furthermore, as was mentioned before, the teacher words per minute showed a significant difference between groups, which did not result in a significant difference between groups in students’ words per minute.
However, in reviewing these data, several problems have to be mentioned: (1) due to the partial (and occasionally complete) loss of transcript data in NSC and SSC, the words per minute measure may not reflect the actual true potential words per minute of each subject, and (2) in ESC, due to the group assignment done by the Lab Assistant, the groups were often larger, which may have influenced the words per minute produced by an individual student. In conclusion, while no effect of teacher feedback style on student output was found in this study, there may have been one present which could not be detected given the problems with human error in implementation of chatting and with the software. In this study no silencing effect of any feedback style could be established, in contrast to the results discussed in Ene et al (2005).
4.3.2 Research Question 2b The question is: What influence does corrective feedback style have on students’ learning, as perceived through learner uptake as measured by evidence of correction uptake within the same transcript? To respond to this research question, again the findings from research question 1 regarding the teachers’ feedback styles are discussed in relation to the findings of correction uptake in the transcripts of the six case study subjects. However, uptake is also possible as a result of a peer correction. Therefore, the dominant feedback style by the teacher may not play such an important role. Again, caution has to be expressed, since these findings only represent the findings from the six case study subjects and cannot be generalized.
Even though the transcripts reflected 1670 student and teacher errors, only 91 were followed by other-initiated feedback and 28 by self-correction. In response to the 91 other-initiated feedback moves, only 4 instances of correction uptake were found in the case study subjects’ transcripts. In each of MorningTeacher’s case study subjects’ transcript sets, one instance of error uptake was found, whereas none were found in the EveningTeacher’s case study subjects’ transcripts. Table 4.12. summarizes the uptake in each transcript set and provides the quotes from the transcripts.
Uptake was even more limited than corrective feedback moves. In EveningTeacher’s class there was no instance of uptake in either of the subjects’ transcript sets. In MorningTeacher’s classes, there was little evidence of uptake. There was one instance of uptake in each of the transcript sets investigated. NSCDanielle’s partners received 12 corrections of which 2 (16.66%) resulted in one instance of uptake following a clarification request made by the teacher and an explanation given by NSCDanielle. NSCJennifer received 10 corrections of which 1 (10%) resulted in uptake following a translation request by the teacher. SSCEmily’s partners received 15 corrections, of which 1 (6.67%) led to uptake following a clarification request by SSCEmily. SSCGina received 2 corrections of which 1 (50%) led to uptake following a model by the teacher. In the transcripts investigated uptake was low.
No clear patterns between feedback style and uptake can be established from these results, since there were only 4 instances. However, in conjunction with prior research, feedback forms that are more engaging to the learner may be more likely to lead to noticeable uptake (Panova & Lyster, 2002), such as the clarification and translation requests that led to uptake in this study.
It should also be mentioned that in addition to the four instances of uptake, there were also 2 instances of clear noticing of the feedback received from MorningTeacher.
NSCCarsten; Ich hatte Astronomy *Haushaltaufgaben.
MorningTeacher: Ja, du hattest Hausaufgaben.
Furthermore, one student also provided a description of MorningTeacher’s corrections in the self-report forms: “Said “Jawohl” and then wrote my question again. I wrote “habt sie” and MorningTeacher wrote “hat sie.” The three instances of uptake following feedback from MorningTeacher, in addition to the two instances of noticing, and the comments on self-report forms, may suggest that MorningTeacher’s systematic feedback style was easier to notice than EveningTeacher’s unsystematic style. It could, however, also be due to the higher frequency of feedback, or a chance finding.
One issue that has not yet been explored in the data may be another possible explanation for low uptake and feedback rates: pauses could potentially be considered a form of feedback. This was not investigated for this study, but should be investigated in the future. In fact, pauses could be considered a form of clarification request that may have lead to some of the self-corrections in this data. Such self-corrections could then be considered uptake. However, this has not been explored in this study, nor in any other study that I am aware of, making it a point of interest for future research.
Since the amount of uptake is so low no pattern between feedback form and uptake can be established. The infrequent uptake was surprising, since it was assumed that the repetitive nature of many of the tasks would encourage uptake by the participants. However, similarly to Fernandez-Garcia and Martinez-Arbelaiz’s (2002) discussion of the change in the negotiation of meaning sequence in chatting due the nature of the medium, uptake may also be considered an unnecessary tool in chatting by the participants. It could be that since participants do not face each other in the conversation, they may not feel the need to acknowledge the correction through uptake.
4.3.3 Research Question 2c The question is: What influence does corrective feedback style have on students’ learning, as perceived through improvement of the structures taught during third-semester German classes as measured by an achievement pre-/post-test? In addressing, this question the pre- and post-test results are considered in conjunction with the results from research question 1. However, caution needs to be expressed since (a) dominant feedback style is based on the case study subjects’ transcript sets, yet (b) all students’ pre- and post-test scores are considered.
In chapter three a detailed description was given about how the tests were scored.
When scoring the tests it was found that several subjects did not complete the writing section of the test, so it was decided for comparability purposes to discuss the test without the writing section.
To establish whether or not there was improvement from pre- to post-test and if there was a difference in improvement between teachers, a repeated measures ANOVA was used. Repeated measures ANOVA allows one to match up pre- and post-test scores of one subject, which limits the amount of variance. In this analysis the between-subject variable factor is teacher and the within subject factor is test, pre-test versus post-test.
The test factor was highly significant (F (1,42) = 162.75, p.001) whereas there was no significant effect for teacher (p =.73). This means that while students improved from preto post-test, the students taught by one teacher did not significantly improve more or less than the students taught by another teacher (see tables 4.22. and 4.23).
Table 4.22 Tests of Within-Subjects Contrasts of Pre- and Post-Test
The analysis was run again with the between-subject factor of class instead of teacher. The within-subject factor of test score was again highly significant (F (1,41) = 173.40, p.001). Once again, the between subject factor was not significant (p =.93) (see also tables 4. 24 and 4.25). Again, this confirmed that students did improve from pre- to post-test, however, the class they were in did not effect their improvement from pre to post-test. This suggests that all students in all three classes taught by either one of the two teachers improved statistically equally.
Tests of Within-Subjects Contrasts of Pre-and Post-Test – Class Difference
In conclusion, there was a significant improvement across all classes from pre- to post-test, however, there was no significant difference in improvement between the teachers or among the classes.
Conclusions drawn from these quantitative results have to be viewed with caution, since the information about dominant feedback style was based on the transcripts of only six case study subjects, and the chatting was only a minor portion of class. Furthermore, during pre-test administration, problems were encountered with the speaking portion which was initially part of the test. Due to these problems, students were not able to complete the test in the allotted 50 minutes. While in ESC the two hour class format allowed me to have students complete the test within the same class period, this was not possible in NSC and SSC.
Using the detailed score card, where I noted kinds of errors in detail and simple test scores, allowed me to observe the development of students’ errors from pre- to posttest. While no quantitative or statistical analysis was performed, some patterns were still suggested by the data, which will need further investigation in the future. Not all students showed dramatic improvements in terms of quantitative analysis; yet, in some cases, a decrease in score actually meant improvement, because while students had formed an incorrect hypothesis about the language, they had formed one. Occasionally such hypothesis formation resulted in more errors than the chance performance on the pre-test.
A general pattern appeared to be from either incorrect to correct, or from unsystematic error to systematic errors. This means that over the course of the semester, students either developed an incorrect or a correct hypothesis about the formation of the structures under investigation. For example, some students used only independent word order in the beginning, and then only dependent word order when following a conjunction or subjunction in the post-test. Since there were more independent clauses in the exercise than dependent clauses, this meant a decrease in score. In terms of case endings, several stages were suggested by the data: (1) no system, (2) overuse of nominative, (3) overuse of dative, and (4) overuse of genitive. These patterns of error development are qualitative observations only which were not analyzed quanitatively. However, they are a stepping stone for a more detailed quantitative analysis of error development in the future.
4.4.1 Research Question 3a The question is: What patterns occur in the data between error type and error treatment? Due to the limited number of instances of feedback received, no patterns could be established between error type and treatment type.