Investigating novice EFL writing teachers’ beliefs and practices concerning written corrective feedback across contexts: a case study from a complexity theory perspective

While previous research has identified misalignment of L2 teachers’ beliefs and practices on written corrective feedback (WCF), much remains to be known about whether and how writing teachers’ beliefs may vary across contexts and evolve over time, as well as the extent to which their beliefs are manifested in their practice. To address this issue, this study draws upon the perspective of complexity theory to explore the data collected in a case study of two novice Chinese EFL teachers’ changes in beliefs and practices about WCF across student-centred contexts and teacher-centred contexts. Data were collected from mul- tiple resources, including interviews, drafts of students’ writing, and teacher on-script WCF. The data revealed a complex process of change in teachers’ beliefs about WCF that underwent distinct stages across contexts. Misalignments between teachers’ WCF beliefs and practices and possible reasons for these differences (e.g. teacher professional identity, affective factors, and individual student differences) are also discussed. This study deepens our understanding of writing teachers’ complex beliefs about feedback practices and sheds new light on feed- back pedagogy in L2 writing classrooms.

2023, VOL 32, nO 3, 465–486

Investigating novice EFL writing teachers’ beliefs and

practices concerning written corrective feedback across contexts: a case study from a complexity theory

perspective

Wenting Chen

College english Department, Capital normal university, Beijing, China

ABSTRACT

While previous research has identified misalignment of L2 teachers’

beliefs and practices on written corrective feedback (WCF), much

remains to be known about whether and how writing teachers’ beliefs

may vary across contexts and evolve over time, as well as the extent to

which their beliefs are manifested in their practice To address this issue,

this study draws upon the perspective of complexity theory to explore

the data collected in a case study of two novice Chinese EFL teachers’

changes in beliefs and practices about WCF across student-centred

contexts and teacher-centred contexts Data were collected from

mul-tiple resources, including interviews, drafts of students’ writing, and

teacher on-script WCF The data revealed a complex process of change

in teachers’ beliefs about WCF that underwent distinct stages across

contexts Misalignments between teachers’ WCF beliefs and practices

and possible reasons for these differences (e.g teacher professional

identity, affective factors, and individual student differences) are also

discussed This study deepens our understanding of writing teachers’

complex beliefs about feedback practices and sheds new light on

feed-back pedagogy in L2 writing classrooms.

KEYWORDS

Written corrective feedback; EFL teachers; beliefs; practices; complexity theory

© 2022 Informa uK Limited, trading as Taylor & Francis group

CONTACT wenting Chen 6501@cnu.edu.cn

This article has been corrected with minor changes These changes do not impact the academic content of the article.

关键词

书面纠正性反馈；英语教 师；信念；实践；复杂性 理论

Plain language summary

L2 writing teachers always provide responses to linguistic errors in learners’ written work This is a common way teachers use to help students to improve accuracy in their revised writing However, there may be a mismatch between their understanding of what kind of feedback is most useful for learners and what kind of feedback they most frequently tend

to provide in practice As L2 writing teachers gain experience in teaching different writing classes, they may change their understanding about what kind of feedback is most useful for learners and their feedback practice may change as well For this reason, this study explores two novice Chinese EFL writing teachers’ changes in beliefs and practices in pro-viding feedback across student-centred contexts and teacher-centred contexts The two teachers taught different writing classes (e.g college composition classes, reading-and-writ-ing classes, academic writing classes) at different times in China and in the US This study found complex mismatches between their beliefs and practices in providing feedback across different contexts over time This study also found three possible reasons that may explain the complex mismatches: what role writing teachers see themselves play in feedback pro-vision, their personalities, and individual student differences The findings of this study encourage L2 writing teachers to enhance feedback practice by reflecting on their feedback beliefs and to raise their awareness when confronting mismatches

Introduction

Written corrective feedback (WCF) has been widely adopted in second language (L2) or English as a foreign language (EFL) writing classrooms as an important pedagogical practice While previous research has shown that providing students with WCF contributes to writing accuracy in their subsequent drafts (e.g Bitchener, 2008; Ferris, 2003), there has been ongo-ing debate regarding the potential of WCF for L2 development, especially after Truscott’s (1996) call for the abandonment of the practice In line with the debate, some factors have been noted as influencing the effectiveness of WCF in facilitating L2 writing development, such as the type of feedback provided (e.g Bitchener & Knoch, 2009; Ferris, 2004; Hyland & Hyland, 2006), students’ emotional reactions (e.g Han & Hyland, 2019), and learner beliefs (e.g Han, 2017) However, most attention has been given to learner factors and contextual factors (e.g curricular goals and instructions, school policies, learning strategies), and research on teacher beliefs about WCF is still limited (Mao & Crosthwaite, 2019), particularly concerning how beliefs may vary across different contexts It is worth noting that ‘Context’ discussed in this study refers to micro-context (e.g IEP classes, college composition classes, reading-and-writing classes, academic writing classes) rather than macro-context (e.g L2 or EFL classrooms)

Teacher beliefs about WCF may have a direct bearing on L2 writers’ learning processes and outcomes by shaping students’ learning beliefs (e.g Wenden, 1999) and influencing their own teaching practices (e.g Borg, 2001) Also, given that classroom teachers’ beliefs about language teaching may change alongside changes in their motivation, attitudes, and emotions (e.g Aragao, 2011), it is important to uncover whether and how teachers’ beliefs about WCF may change across contexts and evolve over time Such developments, in turn, could influence teachers’ practices, as well as mediating factors like motivation, attitudes, and emotions According to previous research (Borg, 2015; Zheng, 2015), the dynamic nature

of teacher beliefs relates to other facets of teacher cognition (e.g teacher knowledge, rience, and motivation), and studies have examined interactions between teacher beliefs, behaviours, and contexts within a complex language learning system Thus, drawing on a complexity theory perspective (Larsen-Freeman & Cameron, 2008, Zheng, 2015), the present study investigated two novice Chinese EFL teachers’ changes in beliefs about WCF across student-centred contexts and teacher-centred contexts, as well as the extent to which their beliefs were manifested in their practice The intent of this study is to contribute new knowl-edge to the literature on writing teacher education and feedback Such knowledge not only enhances writing assessment practices in L2 or EFL classrooms but also provides insights into the complex process of how writing teachers beliefs are shaped, as well as how they interpret their roles and ultimately act as evaluators of student writing.

on learners’ belief formation (Wenden, 1999) While teachers’ belief development can exert

a positive impact on their teaching practices, there is still a gap between teachers’ beliefs and practices, especially in L2 or EFL contexts (Yuan & Lee, 2014) given that teachers’ beliefs may influence their practice and student learning, studies in WCF have investigated teachers’ pedagogical choices (e.g direct/indirect feedback, focussed/unfocused feedback, selective/comprehensive feedback) when responding to students’ writing (Bitchener, 2008; Ellis, 2008; Lee, 2004, 2008) However, existing research on L2 or EFL writing teachers’ beliefs and prac-tices concerning WCF has raised the problematic issue that teachers’ beliefs about WCF are not always aligned with their actual correction practices (e.g Mao & Crosthwaite, 2019; Min,

2013; Lee, 2009) For example, Lee (2009) uncovered ten mismatches between in-service EFL high school teachers’ WCF beliefs and practices According to her study, while teachers believed that giving indirect and selective feedback, requiring multiple revisions, providing both encouragement and suggestions, and leaving students more room for independent thinking are practices that are conducive to learning, they still tended to pay much attention

to language form, give comprehensive feedback, locate all the errors, respond mainly to weaknesses in student writing, and assign single draft assignments in practice

In a similar vein, several studies (e.g Min, 2013; Junqueira & Payant, 2015) have taken longitudinal approaches to investigating teacher feedback beliefs and practices since beliefs are not static and change as teachers gain more experience For example, Junqueira and Payant (2015) study conducted with a pre-service L2 writing teacher found that although teachers believed in the value of providing feedback on global concerns, their local WCF still outnumbered their global WCF in actual practice More importantly, this study revealed the complex beliefs L2 teachers hold about WCF and their frustration concerning how to translate

their beliefs into practice In a recent study conducted in an EFL context, Mao and Crosthwaite (2019) discovered misalignments between college teachers’ beliefs regarding WCF and their actual practice The results indicated teachers tended to provide direct feedback while believ-ing indirect feedback is more useful Also, teachers paid more attention to local issues even they believed that indicating global errors is more conducive to learning.

given these misalignments among L2 writing teachers, researchers have explored crucial factors that can influence teachers’ beliefs about WCF, including a lack of procedural knowl-edge about how to give feedback in practice, previous teaching and learning experiences, and contextual experiences (Ellis, 1998; Lee, 2010; Mao & Crosthwaite, 2019) However, while the literature has demonstrated a non-linear and reciprocal relationship between teachers’ WCF beliefs and teaching practices due to the complexity of their beliefs (e.g Junqueira & Payant, 2015; Lee, 2009), limited attention has been paid to whether and how writing teach-ers’ beliefs may vary across contexts and the extent to which their beliefs may affect their practice, particularly from the perspective of complexity theory Since complexity theory addresses how complex dynamic systems composed of different components form, develop, and adapt in response to changes in environment (Larsen-Freeman & Cameron, 2008), it is meaningful to adopt complexity theory to examine how teachers’ WCF beliefs and practices connect and interact in different and changing contexts over time

A Complexity theory perspective on teachers’ beliefs and practices concerning written corrective feedback

This study is anchored to complexity theory, which is concerned with dynamic interactions among components or agents in producing some overall state or form at a particular point

in time within or beyond complex systems (Haggis, 2008; Larsen-Freeman & Cameron, 2008)

In this theory, systems mean units or different types of elements closely related to each other and interact in different and changing ways (Zheng, 2015) Those different elements or units simultaneously interact with each other and change over time in a nonlinear manner (Clarke

& Collins, 2007) The developmental phases that the complex systems undergo are governed

by different attractors, or factors (e.g teacher factor; learner factor or contextual factor) that can be sufficient to govern a system by itself (Dörnyei, 2014) A powerful attractor or a set

of attractors determines the development of the systems As an effective tool to understand relationships among components or agents in complex, dynamic, and contextualised sys-tems, complexity theory enables us to monitor nonlinear changes between language teach-ers’ beliefs, practice, and contexts within complex systems—and as complex systems themselves (Zheng, 2015) While a system may have degrees of freedom for changing as the elements within interact with each other, it can undergo self-organisation (Davis & Sumara,

2005), meaning the dynamic process in which ‘an organism is responding to the environment

by configuring itself and metamorphosing in order to survive and transcend itself’ (Morrison,

2008, p 20) In other words, the connectedness of each element affects the evolution of the others within the system and reduces the number of possible outcomes as the components are interlinked (Dörnyei, 2014) In this study, I refer to L2 teachers’ beliefs about WCF as a complex system, within which teachers’ beliefs about WCF, their actual practice, and various contexts are sets of interacting variables while, in the meantime, existing as complex and dynamic systems themselves (see Fig 1)

First, L2 teachers’ belief systems are complex (Larsen-Freeman & Cameron, 2008; Zheng,

2015), because they are multifaceted systems that interact and engage with their ments and other variables (e.g WCF practice) Specifically, teacher beliefs include views about learners and learning, teaching, learning to teach, subject matter, and one’s self and the teaching role (Calderhead, 1996) In this study, teacher beliefs refer to L2 teachers’ beliefs about WCF in general (e.g whether or not WFC is useful), how to give WCF (e.g what types

environ-of feedback are helpful), how learners learn through WCF, and L2 teachers and learners These components are connected and influence each other within teachers’ WCF belief systems Second, teachers’ beliefs are dynamic and unpredictable because both teacher factors (e.g knowledge and language learning experience) and contextual factors (e.g soci-olinguistic and educational contexts of writing instruction) could affect and change their beliefs (Borg, 2003; 2015; Yang & gao, 2013) Thus, the change of teachers’ WCF beliefs could appear in different forms of WCF practice Third, teachers’ belief systems are adaptive and flexible because they could encounter continuous changes in response to variations in their environment (Larsen-Freeman & Cameron, 2008; Cochran-Smith et al., 2014) For example, teachers’ beliefs and philosophies may adapt to the changes from their enhanced sense of commitment to different types of classroom instruction, school policy, and cultures of learn-ing (Yuan et al., 2018; Zheng, 2015)

Complexity theory with a focus on non-linear change can shed light on the transformation

of teacher beliefs through complex interactions between teachers as individuals and the teaching environment; however, use of complexity theory as a theoretical lens in understand-ing teacher beliefs remains scant In light of complexity theory, this study looks at two novice Chinese EFL teachers’ changes in WCF beliefs across contexts, and the (mis)alignments between their WCF beliefs and their actual practice The present study contributes to our understanding of the interplay between internal (e.g cognitive knowledge) and external factors (e.g stakeholders and systems) in shaping teachers’ beliefs about WCF given this, it

Figure 1 an L2 teachers’ complex belief system (based on Zheng, 2015, p 36).

may also generate useful implications on how to promote reforms in writing classrooms and teacher training for the development of new entrants To this aim, this study addressed the following research questions:

1) What specific beliefs did the novice EFL teachers hold about WCF across contexts?2) What were their WCF practices across contexts? Did their beliefs match their practice?

If not, why?

Methodology

The aims of this study were to understand what Chinese EFL teachers’ beliefs are across contexts and the extent to which their beliefs were manifested in their WCF practice The case study approach was well suited, as it allows an investigator to explore ‘a real-life, con-temporary system or multiple systems through detailed, in-depth data collection and mul-tiple sources of information’ (Creswell, 2007, p 176) Also, since complexity theory is relevant

to how individuals (writing teachers) and contexts constantly interplay within a complex system (Morrison, 2008), the case study approach enabled the researcher to closely examine the extent of change in teachers’ beliefs and practice in the different contexts This study used a longitudinal case study approach, which enabled us to closely examine the extent

of change in learner attitude and to explore the ways in which it influenced patterns of interaction and language learning opportunities during multiple collaborative writing tasks (Creswell, 2013) It should also be noted that longitudinal in this study means the complex process of changes in teachers’ WCF beliefs and behaviours were collected at different times across contexts, echoing a retrospective qualitative research method, which tracks back the reasons why the complex system has ended up with a particular outcome option (Dörnyei, 2014)

Research context and participants

This research adopted a case study approach and looked into two novice EFL Chinese ers’ WCF beliefs and practices across contexts This study is based in four research universities across two countries, China and the United States of America (US) In the two universities in mainland China, teacher-centred and exam-oriented educational instruction predominates

teach-in language classes even though the reform of college English towards a student-centred approach has been encouraged (Chen & Yu, 2019) In the two US universities, teachers mainly apply student-centred educational policies Two novice EFL teachers (Wendy and Fiona, both pseudonyms) were selected to participate in this study because they had contrasting teaching experiences as EFL teachers across contexts Wendy (age 30, female) taught L2 writing as an international teaching assistant for an Intensive English Program (IEP) at a large research university in the US for four years After that, she returned to China and she had taught EFL writing at a research university in mainland China for two years In contrast, having

a Master’s Degree in Applied Linguistics, Fiona (age 31, female) taught EFL writing at a Chinese university Following that experience, she began her doctoral study at a research university in the US for four years, during which she taught EFL writing classes for both international students at the IEP program and international freshmen Both participants held

Ph.D in Applied Linguistics, and both are near-native speakers of English (see details about profiles of participants in Table 1).

Most notably, Wendy and Fiona have both taught similar writing courses at their tive Chinese universities One typical writing course they have taught was college compo-sition, which was imbedded in a reading and writing course in which the writing component focussed on essay production Another writing course they taught in China besides college composition was academic writing, which aimed at developing undergraduates’ academic English skills in writing These two courses were designed for non-English major undergrad-uates who had to pass the College English Test Band 4 (CET-4, a standardised national college English test for non-English majors) to achieve the required English level for their degrees Wendy and Fiona were also writing instructors or tutors for three types of writing classes in the two universities in the US First, they both taught different levels of non-credit IEP writing courses Students in the IEP had to pass the Test of English as a Foreign Language (TOEFL) examination in order to enter American universities or professional schools Second, they both had experiences teaching Composition I for international freshmen This course was

respec-an introduction to the rhetorical process, emphasising audience, purpose, respec-and occasion of academic writing, in which students received extensive training in planning, drafting, and revising Third, they both worked at their university Writing Centres as tutors to help inter-national students with their writing in terms of task fulfilment, language use, APA format, etc

Data collection procedure

Research data were collected from multiple complementary sources, such as drafts of student texts, teacher on-script WCF, and semi-structured interviews with the two teachers The student texts and teacher on-script WCF included multiple drafts from different teaching contexts, such as student-centred IEP classes (20 copies), college composition classes (6 copies), teacher-centred reading-and-writing classes (20 copies), academic writing classes (10 copies), and writing centres (4 copies) Altogether, 60 writing samples entailing a variety

of genres (e.g argumentative, compare-contrast, cause-effect, and letters) were provided (30 for each participant)

Most of these samples were formative writing and very few were formative writing with summative component imbedded For example, in the response paper, students were directed to begin with a summary of the reading assigned, devoting less than 30% of the paper, and then they were told to react to the reading and support their arguments in the rest of the paper Those student texts and teacher on-script WCF comments are self-selected texts from the two teachers The researcher has asked the two teachers to consider mixed

Table 1 Profiles of the two participants.

name gender (age) Years of teaching english writing L2 proficiency backgroundeducational academic position wendy Female (30) 4 years in an american

university and 2 years in a Chinese university

near-native (118 out of 120 TOeFL)

PhD in applied Linguistics assistant Professor

(China) Fiona Female (31) 3 years in a Chinese

university and 4 years in

an american university

near-native (114 out of 120 TOeFL)

PhD in applied Linguistics esL Program Instructor (us)

language proficiencies of students while selecting writing samples After the writing samples were collected, the researcher and the two teachers had a meeting to discuss why the selected writing samples are most representative The 60 writing samples were consented

by the researcher and the teachers Also, the two teachers emailed students whose texts are used and get their consents before using those 60 texts as writing samples It should be noted that each teacher only provided two copies of the writing samples from writing centre tutoring session because they were not supposed to keep students’ written texts unless students required special needs These copies were therefore from returning students who asked the tutor revisited their errors from the first tutoring session The tutor therefore scanned students’ texts with feedback in the first session and saved in the portfolio for the returning session Both the drafts and teacher WCF were copies of handwritten texts (not original texts) It should also be noted that all the writing samples were from single-draft assignments considering the following factors: First, due to the large class size, drafts col-lected in teacher-centred academic writing class and teacher-centred reading-and-writing classes from Chinese universities were assessed commented only one time Second, if the writing samples the two participants provided were a mixed collection of first, second, and final draft for different assignments, it may be questionable to look at one selected draft without knowing the other drafts as teachers may provide feedback on higher-order con-cerns in the first draft and low-order concerns on the subsequent drafts Thus, it is more comprehensive and holistic to understand teacher feedback from selected single-draft assignments

The researcher and the two participants had a conference before the study during which the researcher introduced the purpose and procedure of the study and explained the textual data (i.e drafts of student texts, teacher on-script WCF) that participants would need to provide Ethical approval was obtained from the author’s university and consent was sought from the two participants Face-to-face, semi-structured interviews were conducted three times with each of the participants The pre-task interview (Interview 1) was conducted at the beginning of the study Each participant was interviewed (see Appendix A for pre-task interview questions) for 30 minutes to elicit their personal details, language teaching and learning background, educational training, and work experiences Then, the two teachers were asked to prepare a collection of student texts and teacher on-script WCF from different teaching contexts (e.g student-centred IEP classes; college composition classes; academic writing classes, writing centres) After that, the researcher conducted the second interview (Interview 2) with each participant for 70 minutes the following week The interview ques-tions (see Appendix B for interview questions) were related to their perceptions and prac-tices concerning L2 or EFL writing in the US and in China (student-centred classrooms and teacher-centred classrooms) The researcher actively encouraged the participants to use student texts and teacher on-script WCF to illustrate their practices in different stages of teaching across contexts Not only did the researcher encourage the participants to provide examples to illustrate their WCF practices as they explained their beliefs, but the participants also assisted the researcher in analysing possible reasons for the (mis)alignment between beliefs and practice After the second interview, the researcher compared the two partici-pants’ self-reported WCF practices with their actual practices in the student texts they had provided Then, a follow-up interview (Interview 3) with each participant was conducted

to recheck and confirm the reasons for any misalignments All the interviews were audio-recorded

Data analysis

Student texts and teacher WCF

The examination of textual data aimed to explore the teachers’ actual WCF practice in response to their beliefs Also, these data illustrated teachers’ actual practice in comparison

to what they had self-reported in the interview about their WCF practice

All the in-text comments, marginal comments, and endnote comments were identified for feedback points (Hyland, 2003) Specific attention was given to identifying multiple single-sug-gestions within a long written comment In such cases, each individual suggestion was con-sidered as one feedback point Following Mao and Crosthwaite (2019), the data were analysed

in terms of how WCF was provided and what errors the WCF aimed to address Specifically, the feedback points were first identified as positive or negative feedback Comments that addressed the correctness of content, learner ideas, or the linguistic veracity of the utterance were all coded as ‘positive’ Comments that were corrective in intent were coded as ‘negative’ Then, the feedback points were classified into the use of direct correction, indirect indicating (e.g circling, underlining or highlighting) errors, and teachers’ indirect comments While pre-vious studies (e.g Shintani & Ellis, 2013) divided feedback points into direct, indirect, or met-alinguistic, this study only applies direct and indirect forms because the ambiguity of the term

of metalinguistic feedback (e.g error or correction codes) used as both metalinguistic feedback (e.g Robb et al., 1986) and indirect feedback (e.g Ferris & Roberts, 2001) After that, each feed-back point was further categorised according to whether it dealt with local (e.g grammar: morphological and syntactic errors, vocabulary: lexical errors, mechanics: spelling, punctuation and capitalization) or global (e.g content: feedback on the conveyed information, or errors altering the intended meaning of the writing ideas: feedback on the intention, organisation: feedback on the structure of sentences, paragraphs and passages) aspects of writing (Lundstrom

& Baker, 2009) Finally, each writing sample was coded as to whether it provided primarily focussed or unfocused feedback In this study, feedback was treated as focussed if the WCF primarily targets to specific error types or patterns (Ferris et al., 2013) whereas texts containing

a range of errors where teachers correct all of them or a range of them in written work were considered as unfocused (Ellis et al., 2008) Although there is an ambiguity in defining focus and unfocused feedback in previous research, where several error categories can be referred

to as focussed WCF (e.g Ferris et al., 2013) or unfocused WCF because of containing more than one type of error (e.g Ellis et al., 2008) or even mid-focussed WCF in some recent research (e.g Mao & Lee, 2020), it is more feasible to consider focussed WCF as selected number (decided

by the teacher) of error types, which includes only one error type and several error types for the current study, because none of the texts collected in this study was one error type focussed

To ensure reliability, a colleague of the author coded all the feedback points independently and compared with the author’s coding The second coder was an experienced L2 writing teacher with a doctoral degree in Applied Linguistics The inter-coder agreement for the iden-tification of the feedback points as 89.3% The inter-coder agreement for categorizations of positive/negative, direct/indirect, local/global and focussed/unfocused feedback was 100%, 92.5%, 98.9%, and 96.5%, respectively Any differences were resolved through discussions until all the inter-coder agreements reached 100%

Interviews

The interviews were conducted in Chinese (the researcher and participants’ first language) and verified through emails by all the participants for accuracy They were then transcribed

by a research assistant to the author, and proofread by the author All the data were analysed according to qualitative data thematic analysis, and the researcher and the colleague who helped with coding were informed to be unbiased and avoid contaminating the data to achieve reliability (Miles & Huberman, 1994).

Data analysis was conducted based on the conceptual framework of teacher cognition (e.g Borg, 2015) The researcher carefully read all the data and identified important codes relating to teachers’ beliefs about WCF Then, the researcher compared teachers’ beliefs with their reported practice to distinguish the differences and gaps between their WCF beliefs and practice in different stages After that, the researcher compared and grouped relevant codes into main themes To identify the reasons for gaps between teachers’ WCF beliefs and practice, the researcher performed an iterative and recursive process of continuously com-paring one piece of data to another and then combing similar categories based on cross-case synthesis (Yin, 2009) Specifically, the researcher coded where the data explained gaps between beliefs and practice and categorised them into themes to summarise the reasons that had influenced participants’ WCF Sample initial codes emerged included educational and teaching background, past learning and teaching experiences, diverse school contexts, classroom settings in relation to their beliefs and practice of WCF Then the codes yielded from coding were compared and merged For example, codes about the changes in feedback approaches in exam-oriented classrooms had been categorised as constrains from complex school contexts In addition to the initial coding, the second coder (the same colleague who helped code feedback points) coded the data independently Discrepancies were resolved through rounds of discussion

Wendy’s WCF beliefs, according to her own description, have undergone three distinct stages:

a stage of having a lack of knowledge about WCF, a stage of awakening and learning, and a stage of confusion and reshaping beliefs Wendy experienced the first stage (Stage I) of having little knowledge about WCF as an L2 learner before she went to the US for her doc-torate Before her doctoral studies, the only experience Wendy had with WCF was to help her graduate advisor grade student essays At that time, she believed her job as a graduate assistant was to help mark as many errors as possible, and most of her comments were direct and primarily on local errors, which can be attributed to her lack of assessment knowledge and instructional training She explained,

The professor just told me to correct errors and give some general comments for the student papers She did not give me specific criteria So, I gave each paper the same consideration that

I would want one of my papers to receive I tried my best to correct all the grammatical errors

as the professor usually did on my paper (Wendy, Interview 2)

Wendy’s WCF beliefs had not been systematically developed and shaped until the second stage (Stage II) during her four-year doctoral studies as an international teaching assistant (ITA) at a US research university In Stage II, Wendy was not only taught how to give WCF to international students taking Composition I and visiting the Writing Centre, but also gained

a greater understanding of writing assessment literacy (e.g purpose, process, and practice

of assessing L2 writing) in her doctoral courses Influenced by her experiences as a language learner and teaching assistant during this time, she formed strong beliefs about how to teach L2 writing and give WCF in US schools For example, Wendy believed that ‘teacher feedback should be more selective and indirect’ (Interview 2) and L2 teachers need to involve students in a problem-solving process to encourage them to rethink their errors critically instead of providing immediate error correction:

I think it is a critical thinking and learning process when students make revisions based on teacher comments If writing teachers offer direct error correction, students would not be able

to reflect and think hard on their errors Also, I only indicate major errors for each writing ment I give comments on paragraph coherence and clarity of ideas For example, I underline sentences that did not get the ideas across (Wendy, Interview 2)

assign-Wendy also mentioned that indirect feedback was more effective in student-centred writing classrooms; this was because ‘pair work was always implemented, then students had opportunities to analyse the indirect teacher feedback and help each other in the revisions’ (Interview 2) In line with her beliefs, indirect feedback accounted for higher proportion of the feedback points on the 12 submissions Wendy provided from Composition I and Writing Centre tutoring sessions

In terms of the balance between focussing on local or global errors, Wendy clearly explained in the interview that ‘comments on idea generation, paragraph coherence, and organisation will help improve the quality of the writing’ Meanwhile, she believed too much error correction was less helpful ‘because errors will still recur no matter how hard the teacher works’ (Interview 2) However, the feedback analysis showed that 82.1% (feedback points for local errors of drafts from Stage II/all WCF feedback points of drafts from Stage II) of Wendy’s feedback gave attention to local errors In this instance, Wendy did not seem to practice what she believed since the interview data suggested that she was fully aware of the importance of commenting on global errors

The reasons for this misalignment are complex On the one hand, Wendy noted that she had experienced conflict about her multiple identities (e.g accent, nationality) as a non-En-glish L2 teacher, which resulted in her emphasis on local errors in WCF practice She explained,

I do believe comments on bigger picture is more important But I was not confident enough to teach writing in the first two years Students would question how come a Chinese teacher teaches English writing Then, I intended to give more WCF no matter what kind I assume stu- dents will consider me as a more qualified teacher if I help them correct more mistakes in writing (Wendy, Interview 3)

On the other hand, such misalignment was correlated with her limited professional petence in teaching L2 writing due to insufficient training and in-class practice For example, although Wendy thought error correction should be selective, she felt her understanding

com-of ‘selective feedback’ was on the theoretical level and, thus, was doubtful about her ability

to implement it effectively She said,