«Item type text; Electronic Dissertation Authors Goertler, Senta Publisher The University of Arizona. Rights Copyright © is held by the author. ...»
To summarize the role of the teacher as discovered in this study: (1) teachers did not follow the instructions provided to them in their Manual; (2) teachers were not able to and did not use the Lab Assistant in the same way; (3) the two teachers used the physical space differently; (4) while students’ descriptions of the teachers’ role in the virtual space matched the observed role, they did not describe the teachers’ role in the physical space;
(5) students in the ESC received more input than the other students; (6) there was no difference in student output; (7) teacher words per minute were lower than student words per minute; (8) teacher feedback was infrequent in general and even less frequent from EveningTeacher; (9) the two teachers used some similar and some differing teacher moves during chatting; (10) MorningTeacher played the role of facilitator and EveningTeacher the role of conversationalist; (11) despite the student-centered activities and the more equal language output between teachers and students, the teacher still holds a power role in the classroom.
5.3. Corrective Feedback in SCMC Several of the research questions answered in chapter three addressed the issue of corrective feedback in the transcript sets of the six case study subjects. The major finding was that corrective feedback was infrequent, regardless of who provided the feedback, or to whom the feedback was provided.
The percentage of errors that received feedback from the teacher ranged from 0% to 10.42% in the six transcript sets. Peer feedback ranged from 1.58% to 7.78%, bringing the total of other-initiated feedback to a range of 1.30% to 12.50% which is much lower than the F2F reports from (Lsyter & Ranta, 1997) as discussed in chapter two. In addition, self-corrections ranged from 0% to 16.67%. Looking at the other-initiated feedback, this means that even in the most ideal circumstances, no more than 25 of 200 errors received feedback.
The two different teachers used two different feedback styles in the transcripts analyzed. MorningTeacher used repetitions with corrections and models more often than any other form of corrective feedback, and in general used significantly more feedback than EveningTeacher. I referred to MorningTeacher’s feedback style as systematic.
EveningTeacher used a variety of feedback forms such as clarification requests, models and repetitions in approximately equal frequency, and hence her feedback style was referred to in the discussion as unsystematic. The two generally opposed feedback styles of implicit and explicit were not found as the opposing factors in this study.
When providing feedback to peers, students across all three classes most frequently used models, while when correcting their own mistakes they most frequently used repetitions. The repetitions were often partial repetitions and usually marked “*.” Furthermore, not all repetitions included a successful correction.
Due to the overall infrequent use of feedback, no patterns could be established between error type and feedback type. However, a correlation between teacher feedback and self-correction could be established for the six case study subjects. This may indicate that students who receive more opportunities to observe more correction may gain a greater awareness of their own errors.
In terms of the effectiveness of feedback styles, two forms of evidence were considered: improvement from pre- to post-test and instances of uptake. This argumentation faces several problems that should be mentioned before reviewing the results. First of all, since chatting only played a minor role in the overall curriculum, gains on the test are most likely due to learning that occurred in the 47 to 48 hours of other instruction, than in the two to three hours of chatting. Furthermore, the dominant feedback style was labeled based on findings from only the chat transcript sets of the six case study subjects. While these are severe limitations to this study, given the time constraints, an in-depth analysis of only a subset of students was possible. For pedagogical and administrative reasons, chatting for more than 20 minutes a week would not have been possible or advisable. However, the results could still suggest trends, which should be investigated further.
As mentioned before, the first measure of effectiveness of feedback was the improvement from pre to post-test. The good news is that there was a significant difference between pre and post-test scores across all groups and teachers according to the repeated measures ANOVA analysis. However, there was no significant difference in student improvement among groups or between teachers. Again, this suggests that the differing feedback styles may not have a negative or positive influence on language learning, in contrast to what other researchers have argued, which will be discussed later.
However, as mentioned above, chatting was only a minor portion of class, and in addition, the dominant feedback style was only quantitatively established from six case study subjects’ transcripts.
To review, in the transcripts analyzed, 1670 student and teacher errors were identified, of which 91 received feedback from others and 28 were followed by selfcorrection. However, the 91 other-initiated feedback moves only resulted in four instances of correction uptake (4.40%). Due to this low frequency of uptake, no patterns could be established between corrective feedback type and correction uptake. As already discussed Lyster (1998a) found that feedback forms that promote negotiation resulted in more uptake. A similar pattern may present itself in this study, though it can not be substantiated due to the limited data. Nevertheless, it could be argued that feedback forms that engage the learner more may lead to more uptake.
In regards to teacher feedback this study, showed the following research results:
(1) corrective feedback was infrequent from all sources; (2) the two teachers used different feedback styles and frequency; (3) peer-feedback and self-correction styles were the same across classes; (4) no patterns were found between error type and feedback type;
(5) effectiveness of one feedback style over another could not be established as measured by uptake frequencies and by gain on achievement.
5.4. Language Use in SCMC In discussing the language used during SCMC, I will again differentiate between the students and the teachers. First the language used by the teacher will be discussed and then that of the students.
It has already been mentioned that there were differences between the two teachers in regards to the input they exposed the students to, the amount of feedback and the kinds of teacher moves used. Next, the teachers’ target language use and error rate will be reviewed based on the findings from the transcripts of the six case study subjects.
According to a chi square analysis of the target language use in the six case study subjects’ transcripts there was a significant difference between the two teachers, with EveningTeacher using less target language (93.65% in contrast to 100%) during the chat sessions. However, as was also found in Ene et al. (2005), target language use was high.
In addition, in the transcript sets analyzed, EveningTeacher exposed the students to significantly more teacher errors (6.19% of her words in contrast to 1.13% of MorningTeacher’s words according to a chi squared analysis). Finally, it was argued that EveningTeacher’s errors were systematic while MorningTeacher’s were not.
Below, the students’ language will be discussed by addressing the following aspects: target language use of the six case study subjects, fluency of all students, comprehensibility of students’ language use as observed in the six case study subjects’ transcript sets, error rate in the six case study subjects’ transcript sets, improvement from pre- to post-test of all students, error uptake, and correction uptake. It needs to be reiterated at this point that data were only coded by the researcher and no inter-rater reliability was established.
As mentioned earlier, EveningTeacher used significantly more English words in the case study subjects’ transcripts than MorningTeacher and MorningTeacher’s case study subjects used significantly more target language than the EveningTeacher’s case study subjects. To reiterate, this may suggest that the teacher’s frequency of the use of the target language may be seen as a model by the students for their own language choices. A more focused study of this phenomenon would be of great interest.
As mentioned before, there was a significant difference between groups in regards to the input received by the students from the teacher. However, there was no significant difference in terms of students’ output by group or by teacher according to an ANOVA.
No silencing effect of any teacher participation style was found as was the case in Ene et al. (2005). The NSC on average produced 3.64 words per minute, the SSC 3.32, and the ESC 4.01. Furthermore, it should be mentioned that except in the case of ESCTiffany, the quantity of teacher words per minute observed was always lower over the course of the semester than students’ own words produced, which may confirm the democratization effect (Beauvois, 1998). It is interesting, though, that even though the feedback rate was different for each teacher, and the output rate was different for each group, there was no significant difference between groups or teachers in regard to students’ fluency.
Comprehensibility of the chat transcripts analyzed was generally high. More than 95% of the language was comprehensible, as I evaluated. However, one should mention that I am a sympathetic reader used to American language learners of German.
To summarize my findings concerning accuracy, errors by the case study subjects’ were counted. In relation to the total German words used, an error rate was established (see table 5.1. for data summary). All case study subjects had an error rate below 20%. The two case study subjects from the SSC had the lowest error rate (SSCEmily 4.75% and SSCGina 7.25%) and the students in ESC the highest error rate (ESCAmanda 14.84% and ESCVictoria 19.45%). These error rates, at a second-year leel of language study, are encouraging as counter-argument for those who might claim that a focus fluency and/or using CMC at this stage of interlanguage development may lead to severly reduced accuracy.
Case Study Subjects’ Error (E) Rates.
As mentioned in the previous section, there was significant improvement from pre- to post-test; however, there was no difference in improvement between classes or teachers. Students were observed moving from making unsystematic errors to systematic errors. Furthermore, there was a significant relationship between words per minute produced by the students and the post-test scores, i.e., between indicators of fluency and of accuracy. Students with higher fluency also had higher accuracy.
Since there was only a limited amount of feedback, one fear was that there would be a high frequency of error uptake. However, as in Ene et al. (2005), error uptake was low. Only 4.91% of all errors in the six case study subjects’ transcripts sets resulted in error uptake. Yet, it should be mentioned that the error uptake rate of 4.91% is in contrast to an overall other-initiated feedback rate of 5.45%. In addition, while 4.91% of errors resulted in error uptake, only 4.40% of other-initiated corrections (to only 5.45% of errors) resulted in correction uptake.
In regards to patterns between error type and error uptake, no clear patterns could be established. I am speculating that errors such as grammatical gender, plural forms, and stem vowel changes in verbs may be more likely to result in error uptake than other kinds of mistakes based on the reading of all transcripts. However, this will have to be substantiated at a later point with more data and a more precise definition of error types.
In regards to the source of errors, I found that MorningTeacher’s errors least frequently resulted in error uptake (0%), followed by the students’ errors (0.31%), and the errors that most frequently resulted in uptake were errors by EveningTeacher (16.67%). At this point, I am speculating that error uptake may be a result of error type and context. In addition, no correlation was found between teacher feedback rate and error uptake rate, i.e., providing more teacher feedback may not reduce error uptake rate.
To summarize 5.4, teacher and student language use in SCMC was presented. The teachers both used the target language almost all of the time, however, MorningTeacher used it more. Furthermore, EveningTeacher had a higher error rate than MorningTeacher.
The students also had a high level of target language use, with the students taught by MorningTeacher using even more target language than the EveningTeacher’s. The students in the three different classes received differing levels and quality of input, but, this did not effect their output frequency. In general, the teachers produced fewer words per minute than the students. Comprehensibility was above 95% for this researcher, and the error rate of case study subjects did not exceed 20%. All students improved from preto post-test; however, there was no significant difference between the groups. There was a positive correlation between the fluency and accuracy measures, with higher fluency matching up with higher accuracy. Error uptake (4.91%), other-initiated feedback (5.45%), and correction uptake (4.40%) were all infrequent. In addition, there were no clear patterns between error uptake and error source, or type of error and error uptake, suggesting the need for further research with more data to pursue the questions raised by the dearth of clear patterns.
5.5. Students’ Preferences and Actual Practices As mentioned in chapter four, there were hardly any differences among opinions of the students in the different classes, taught by the two different teachers, or across time. Students reported liking corrective feedback, though some of the students expressed some caution in regards to peer feedback. Furthermore, some students expressed their fear of embarrassment due to corrective feedback, which is probably the reason why some students said that feedback should be done discreetly. However, students thought that feedback was necessary for learning. This finding confirms Schulz’s (1996) survey study findings.