Friday, November 9, 2007

Weekly Reflection (W12) by Myong Hee

On Thursday (L2 writing & error correction)
Covered Bitchener, Young, & Cameron (2005) and Ferris (2004)

About Design
• Three groups were different; not equivalent (Yao)
• In reality, we are likely to take intact classes and this can create very different contexts; so problematic.
• Another methodological dilemma in error correction: Longitudinal studies (hard to control invariable) vs. One-shot experimental studies (not being longitudinal)
• Pre-test seems missing (Yun Deuk)
• 1st essay can be used as a pre-test or administering a relevant grammar test can be another option.
• Conference for each individual seems very short and not clearly described (Sang-ki)
Probably it is done for a fast check; however 20 min. conference may not be doable.
• Number of errors in Table 1 is just row numbers; so, it is not based on the length of individual essay (Kevin).
• Keeping (internal) ethical requirement for control groups should be considered.

Possible ways for a Better Design
• In Table 2, stating N size may be a good idea
• MANCOVA may be a better tool to analyze since students may have different starting points at the 1st essay, and we have to check individual’s progress in the four essays.
• The design seems complicated, so make the design simpler (one feedback group vs no feedback group)

About Conclusion
• Students showed ups and downs in improving their grammar. Do we need a longer longitudinal study in order to have better ideas?
• Try to look at different categories of errors (rather than all the categories) in order to check students’ improvement
• Conferencing is working, so it should be implemented.
• Treatable vs. not treatable definitions are problematic (e.g., articles in English grammar). These terms are based on intuition.

MyongHee’s thought: In most studies in this area, NS teachers provide error correction in NNS students’ L2 writing. However, in reality, most English teachers in EFL contexts (e.g., Korea, Japan, & China) are NNSs whose English language skills including grammar are not perfect. I wonder to what extent the findings/suggestions and studies of these NS teachers-NNS students studies in written error correction are applicable/relevant to NNS teachers-NNS students contexts.

Weekly reflection (Nov 5th & 8th) by Yuki

Commentary for November 5th and 8th (by Yukiko)

Sorry for the late reflection notes. Please feel free to comment or modify the text below.


The reading topics of the week were on socio-cultural perspectives to collaborative learning process. Socio-cultural theorists view that learning is socially constructed and is mediated by symbolic artifacts (e.g., language). The following articles were discussed in class:

Tuesday:

  • Aljaafreh, A., & Lantolf, J. P. (1994). Negative feedback as regulation and second language learning in the zone of proximal development. Modern Language Journal, 78(4), 465-483

  • Nassaji, H. & Swain, M. (2000). A Vygotskian perspective on corrective feedback in L2; The effect of random versus negotiated help on the learning of English articles. Language Awareness, 9, 34-51.

  • Nabei, T., & Swain, M. (2002). Learner awareness of recasts in classroom interaction: A case study of an adult EFL student’s second language learning. Language Awareness, 11(1), 43-63.

Thursday:

  • de Guerrero, M. C. M., & Villamil, O. S. (1994). Social-cognitive dimensions of interaction in L2 peer revision. Modern Language Journal, 78(4), 484-496.

  • de Guerrero, MCM and Villamil, OS (2000) Activating the ZPD: Mutual scaffolding in L2 peer revision. Modern Language Journal, 84(1), 51-68.


1. Summary and commentary on Tuesday discussion

Researcher’s background:

  • M. Swain: She shifted her perspective towards L2 learning from cognitive-interactionist to Vygotskian socio-cultural perspective. Nabei and Nassaji were her PhD student.

  • J. Lantolf: His recent research interest includes linguistic typology and gestures. Aljaafreh was his PhD student.

The fist two studies (Aljaafre & Lantolf, 1994; Nassaji & Swain, 2000) were guided by Vygotsky’s notion of zone of proximal development (ZPD) in analyzing the learning process during tutor-learner dialogic collaborative activity. In this framework, it is considered that the feedback prompt teachers/tutors provide mediates learning (microgenetic learning: learning within a short period of time). Aljaafre and Lantolf provided a very useful framework for identifying twelve levels of implicit to explicit regulatory strategies for feedback prompts (Most implicit prompt: A learner identify error independently; Most explicit prompt: A tutor provides examples of the correct pattern after explicit explanation). Their framework seems very practical and can also be used as a guideline for “graduated” feedback in instruction. David commented that it is natural that many teachers adjust to learners’ ZPD in order to provide effective feedback. In classroom, within peers, different people fill in and contribute to reach an understanding from different ZPD starting points. It’s impossible to adjust to different individual ZPDs in a teacher fronted classroom.

Based on Aljaafre and Lantolf’s (1994) ZPD scale, Nassaji and Swain (2000) examined the difference in learning articles between a student who received corrective feedback with graduated contingent help within ZPD (ZPD condition) and a student who received a randomly gauged help (non ZPD condition). In the random ZPD condition, Nassaji and Swain also explored the relationship between the quality of help provided on each article error and the performance of accuracy in the final test for each article error. They concluded that (a) graduated help was more effective than random help, and (b) that more explicit help produced more accurate results than the implicit help. We have to take these results with caution because of the following reasons:

  • The two learners seemed to have different proficiency levels in terms of article use from the beginning. The ZPD student made 28 article errors across four tests, while the non-ZPD student made 20 errors (in one wring, the non-ZPD student had perfect article use).

  • The final test involved an article cloze test of their original writing. The indicator of learning was the accuracy ratio of the article use in the final test for each passage (the number of items ranged from 1 to 12). Since the number of items was so small that concluding the results with proportion score may be misleading. While it seems like a good idea to construct exactly how students learn based on their own writing, from a psychometric viewpoint, the tailored test may not be sufficient to provide trustworthy quantitative evidence.

Some suggestions were made to improve the design of the study. It would be more convincing if there was another article cloze post-test with a passage other than their original writing. Using different passage will potentially show the transfer of learning.

Despite above remarks, the study showed insights into the dynamic and contextualized nature of corrective feedback. Lourdes commented that we can assess necessary teacher/tutor/expert engagement and help the learner needs using the ZPD scale. It will be interesting to see how some students may require more mediation to self regulate and appropriate one type of information.

Nabei and Swain (2002) does not mention social-cultural framework, but looks at the learner as an agent. The learner in the case study “chose when to make use of the learning opportunities presented to her” (p. 59), willingly tuned in and out of the learning context, and was more engaged and cared more during group environment. Nabei and Swain concluded that the reaction to recast a student received was affected by the discourse contexts (e.g., teachers’ intention, group vs. teacher fronted) and learners’ orientation. A close attention to learner agency and orientation, and contextual factors is called for.

Lourdes pointed out that Nabei and Swain’s study is a good example providing thick description of the learner and the interactional context.

2. Summary and commentary on Thursday discussion

A. On IRB issues

We had extensive discussion on IRB issues. Here are some tips getting IRB approvals. Please check with IRB office directly for accurate procedures and information.

(a) Which category do we submit?

  • If it's a regular educational intervention/instruction and it has no potential harm, usually, our research falls under "exemption" category. You need to go through IRB's criteria to see if your study meets their eligibility criteria for exemption.

(b) Research involves your own students and the intervention is part of your regular teaching practice.

  • Get approval to use classroom data (e.g., writing, grades, homework, etc.) from your students after providing them course grades.
  • Ask IRB officer to come to your class at the beginning of the semester to collect information on consent/non-consent to participate in research. IRB will keep the information until the grading is done. Later, you can use the permitted data for your research.

(c) Is it ethical to provide extra credit to those who participated in research?

  • Minimal compensation is fine. You cannot withhold the compensation you promised to provide, even if your participant decided to withdraw in the end.


The rule of thumb is to ask for IRB approval before you start conducting your study!


B. Summary and commentary on the article discussions

Both 1994 study and 2000 study by de Guerrero and Villamil are part of a large scale project involving 40 dyadic (one reader and one writer) interactions during peer revision of their writing.

One of de Guerrero and Villamil’s contributions in 1994 study was their coding system specific to peer interaction for revising writing. They created an analytic framework for categorizing episode type (on-task, about-task, and off-task), interactional types within on-task interaction (e.g., reader writer interactive revision, writer teacher interaction), cognitive stages of regulation (object regulated, other-regulated, and subject-regulated), and social relationships (symmetrical vs. asymmetrical).

Peer review sessions of 27 pairs of ESL writers revealed that most students remained on-task, engaging in complex and productive interaction, and self-regulating themselves depending on tasks and roles (reader or writer). It was interesting to see the fluidity of regulation types, as well as the effect of social relationships on cognitive stages of regulation.

In 2000 study, one dyadic pair uncovered how the reader mediated the revision by flagging problematic phrases and linguistic errors, providing instructions and models, and giving supportive comments. The revision process revealed how social interactions shape the revision of the text and how personal and affective exchanges provide lubrication for equal self-regulations and commitment to task accomplishment.

In both studies, L1 use was considered as a mediating tool and resource for the learners to achieve higher level mental operation. The use of L1 will depend on the dynamic and the purpose of the class; however, it can be a facilitation tool for learners.

In collaborative work, regression is natural in dynamic learning process, as students may not come up with the right answer and/or solution. However, learning does not happen in a linear fashion, thus de Guerrero and Villamil conclude and I concur that peer collaborative work can provide learners with opportunities to appropriate learning strategies and tools which learners can eventually use on their own problem solving.