Exploring the Effects of Genre-based Portfolio Assessment on EFL Writing With Focus on Learner Engagement


 With controversial implications in language learning context, there is not enough evidence for learners’ engagement in genre-based portfolio assessment (PA) and its impacts on their writing improvement. To fill the gap, this case study examined 46 EFL undergraduate students’ performance on descriptive and narrative writing tasks. In 12-week PA design, feedback points were collected from teacher formative assessment of the students’ descriptive and narrative writing, with reference to the genre-specific indicators in West Virginia Department of Education descriptive writing rubric and Smarter Balanced narrative writing rubric, respectively. Statistical results reported the significant impact of PA on improving components of word choice and grammar, development and organization of ideas in student descriptive writing, with no sign of improvement in their performance on descriptive writing post-test. Statistics also supported the positive impact of PA on improving the components of elaboration of narrative, language and vocabulary, organization and convention in student narrative writing, and their performance on narrative writing post-test. The qualitative data on students’ engagement in PA was collected from their reflective journals. After inductive content analysis, the students’ self-reports were schematized, and their level of engagement was rendered in terms of their approval of usefulness and novelty of PA, frequent mismatch between student self-assessment and teacher written feedback both in quality and quantity, ‘sensitivity’ or focus of teacher feedback to some writing features over others, applicability of teacher feedback to revision process, and overall perception of writing improvement. The paper provided teaching implications for EFL practitioners and suggestions for future research.


Introduction
Since the early 1990s, assessment for learning (AfL) has reached worldwide audience for improving teaching and learning in educational context (Cowie, 2005;Klenowski, 2009; O'Shea, 2019). Several studies have examined how AfL might bene t second/foreign language (L2) teaching pedagogy, language learning, and L2 learner performance (Brown & Abeywickrama, 2010;Dann, 2014;Darling-Hammond & McCloskey, 2008;Earl, 2013). As Ramaprasad (1983) conceptualized, AfL in language classroom needs the L2 learners' perception of a gap between a long-term goal and their status quo, as well as their commitment to bridge the gap to attain the goal. Ideally, either language learners will engage in self-assessment to generate the information about the gap, or the teachers themselves will explore the gap and to provide feedback about it to the students. Ultimately, the action to close the gap will be taken by the fully engaged students in the process of learning (Sadler, 2010). But in reality, L2 teachers and learners have more critical steps to take. In AfL practice, the teacher needs to reinforce the capacity in the students to engage, to diligently discover the gap, and to take full responsibility for carrying out remedial actions. Thus, L2 learner self-engagement is not an option; it is a survival kit. However, the focus on learner engagement is not a common practice in most L2 classrooms, as the majority of language teachers do not welcome such shared responsibilities with students (Ecclestone, 2007;Hargreaves, 2005). In essence, as Black and Wiliam (2009) rightfully disputed, the practicality of AfL at language learning classroom level has remained insu cient and more evidence is needed to support the real bene ts of various types of AfL, including portfolio assessment.
Portfolio assessment (PA), as a common platform of AfL, largely demands L2 learners to actively engage in self-assessment and self-re ection (Lam, 2013; for reaching a closure in the learning gaps. This is advised through redrafting and writing re ective journals (Hamp-Lyons, 2016; Lam & Lee, 2010). Yet, the full practice of PA in L2 setting has faced massive problems, such as teachers' AfL malpractice (Harris & Brown, 2009;Willis, 2011) or lack of learner self-engagement (Lee & Coniam, 2013;Li, 2010). Therefore, Hyland and Hyland (2006) called for more research on the aspects of PA impact on learning of writing skills in English as a Foreign Language (EFL) settings. Some research examined the teachers' experience with various models of PA (i.e., progress, workshop, and showcase) (Lam, 2017;2018a;Lee, 2016;Lam & Lee, 2010), the effects of PA on boosting L2 learner autonomy and metacognitive awareness (Aydin, 2010;Carless, 2011;Hirvela & Sweetland, 2005), and learner text revision strategies (Hamp-Lyon & Condon, 2000). However, research ndings on how teacher portfolio assessment might impact the L2 learner engagement in genre-based writing have remained unclear with limited empirical evidence (Hamp-Lyons, 2007). In the same vein, while the use of portfolio approach to collect student performance on different genres of writing has been well-reported in L1 writing (Hyland, 2004;), yet the contribution of portfolio assessment into EFL learners' genre-based writing performance is largely under-documented (Lam, 2019;Wiliam, 2006). To void the gap in the literature, this study aimed to set a genre-based PA platform to investigate the role of teacher formative assessment in EFL learners' degree of engagement in descriptive and narrative writing progress.

Literature Review
A writing portfolio is a collection from a large body of students' written works, often endorsed with re ection pieces of writing by the students. As a popular platform for self-regulated learning and evaluation, portfolio assessment (PA) is often assumed a better-quality alternative to traditional, productoriented assessment for improving the student writing performance and long-term learner engagement (Benson, 2006a;Hirvela & Sweetland, 2005;Li, 2010;Mohamadi, 2018). However, as Condon and Hamp-Lyons (1994) argued, "the portfolio has simply been accepted on faith, on writing specialists' feeling that the portfolio is better" (p. 277). Despite reported educational bene ts, PA has remained controversial when utilized in classroom situations, namely due to L2 teacher in exibility (Xu & Brown, 2016), insu cient and unwilling student engagement (Li, 2010), complicated and holistic grading system (Song & August, 2002), and lack of school support (Lam, 2018a;Lam & Lee, 2010).
As Lam (2018b) indicated, since writing portfolios are reported to sustain students' close attention to their own progress of writing, their active engagement in teacher feedback is central (Burner, 2014).
Furthermore, for a powerful PA experience, L2 writing teachers should prime such student selfre ectiveness. On the other side of the isle, however; evidence of how well students comprehend and engage in working portfolios in L2 context is still anecdotal and under-researched (Hamp-Lyons, 2007; Hamp-Lyons & Condon, 2000). To reach con dence in the student engagement and self-assessment in PA, L2 writing teachers may need 'scaffolding' the students in terms of tutorials on the entire portfolio process (Carless & Boud, 2018), using examples and prompts (Gregory, Cameron, & Davies, 2001), extending deadlines to sustain their engagement (Lam, 2014), and training them to writing assessment rubrics (Panadero & Romero, 2014). Romova and Andrew (2011) emphasized the critical role of selfassessment practice in PA as it warps student persistence, academic engagement (Finn & Zimmer, 2012), and ultimate achievement in PA. Successful engagement depends on how well L2 learners understand the goals of PA, how soon they picture the distance between their own status and the goals (Caner, 2010), and what they actually do to reach the goals. In other words, learner engagement is the bread and butter for effective learning, yet the topic has been overlooked in the mainstream research on PA (Price et al., 2010; Steen-Utheima & Hopfenbeck, 2018).
Apart from controversies over its aftermath in language learning context, PA is still assumed as a powerful pedagogical and assessment alternative, mostly because it reinforces the L2 learners' "understanding of writing as a socially-situated process in academic discourse communities" (Duff, 2010, p. 169). In doing so, genre-based writing PA can assess both microscopic (i.e, mechanical, formal) and macroscopic (i.e., textural, discursive) aspects in L2 learner writing progress (Borg, 2003). Learning writing genres such as narrative, descriptive or expository is one of the critical issues of all times in SLA research (De Fina & Georgakopoulou, 2015;Shober, 1996). Hyland (2003) celebrated engagement in the genre-based writing process as empowering, dialogic, and systematic metacognitive awareness in language learners. Adopting a genre-based approach, Hinkel (2002) also suggested that to develop effective written discourse, EFL students should master "the mechanical aspects of composing sentences, paragraphs, and larger units of discourse that correspond to the dominant genres of the academy, a speci c eld, or both" (p. 57). While the process approach has an eye on the L2 writers over ow of ideas, the genre-based approach has switched its focus to the socio-literacy of the L2 writers in generating real texts that properly address the target discourse community (Hyland, 2003). While Badger and White (2000) believed that the product, process, and genre-based approaches to writing interplay, Romova and Andrew (2011) argued that the genre-based approach only integrates with the process approach, by "adding focus on text/context, and emphasizing the role of language in written communication" (p. 114). Hamp-Lyons and Condon (2000) argued that writing PA can best t with such 'genre-process nexus' approach.
Genre-based approach to analysis of written narrative is no longer the sole responsibility of literary studies. Narrative analysis has entered the realm of human sciences and professional practice, including psychology and learning L2 writing (Roohani & Taheri, 2015). Assumed as an art or gift of storytelling, narration is made through every minute of every day in our life, so that we make narration plenty of times (Abbott, 2002). Lou-Conlin (1998) de ned written narrative as a system of gradual development through which the writer entertains with the logical sequence of ideas and events. Narration is mostly done with the purpose of maintaining the readers' interest in a given event or personal experience narrative (PEN) (Labov, 2001). In the same vein, descriptive genre of writing gives certain attributes to a person, place or chain of events in detail. Such entities should be described in such a way that the reader can capture the topic and enter the writer's experience. Descriptive writing is considered as a means to improve other genres of writing such as narrative and expository or perhaps as a dominant strategy in writing academic texts (Birjandi & Hadidi Tamjid, 2012).
In line with genre-based approach to PA, EFL learners may have a chance to engage in gaining control over a variety of genre-based writing such as narrative and descriptive in the target discourse. However, literature on PA mostly pertained to general writing in L1 (Hamp-Lyons, 2006) or in L2 (Gottlieb, 2000), with marginal focus on the L2 learners' genre-based writing performance, their weaknesses and their goals. Therefore, an urge to further research on this topic was strongly felt, particularly in EFL context.

This Study
To bridge the gap in the research literature of portfolio assessment, this study investigated the impacts of genre-based portfolio assessment on 46 EFL undergraduate students' engagement in descriptive and narrative writing. The academic goal for choosing descriptive genre of writing was to examine EFL students' ability in describing tables, gures, owcharts and other course-related writings at university level. Likewise, the academic goal for choosing narrative genre of writing was to observe EFL students' performance on reporting the stepwise experimental procedures or scienti c processes in their term projects or scienti c reports. Moreover, EFL students' engagement was conceptualized as a 'metaconstruct' or a framework in which the student regular self-assessment, critical thinking, motivation, self-e cacy and enthusiasm were integrated to achieve the learning goals (Fredricks, Blumenfeld, & Paris, 2004).
The research design was a 12-week genre-based PA, in which the distribution of teacher formative assessment was obtained once-a-week on the students' descriptive and narrative written scripts. The student engagement was assessed through the students' submitted re ective journals. To this end, the following research questions were raised: RQ1: What is the impact of portfolio assessment on EFL learners' descriptive writing? RQ2: What is the impact of portfolio assessment on EFL learners' narrative writing? RQ3: What are the EFL learners' perceptions of the portfolio assessment, teacher feedback, and their writing improvement? 4. Method

Participants
A sample of 46 EFL undergraduate students at different university majors (architecture, economy, mathematics and MBA) took part in this study. They were sophomore students at Karaj Azad University, Iran. A non-random purposive sampling method was adopted in this study (Ames et al., 2019), because selecting a representative group of university students with adequate experience in genre-based writing in English was the researcher's plan. The selected participants in this study had already performed on at least 10 genre-based writing tasks, such as writing an invitation card, a shopping list, a letter of application or travel journals as partial requirements in previous English writing courses at university. Their participation was voluntary and no payment was granted to them. Their experience in learning English was between 3 to 6 years (m = 5) and their ages ranged from 20 to 31 (m = 25.5). Their general English pro ciency level was measured by administering Oxford Placement Test, and determined as intermediate.
Additionally, four MA graduates of English language teaching (ELT) took part in this research as the EFL teacher (n = 1) and assistant researchers (n = 3) who provided feedback to the students' descriptive and narrative writing, co-rated them and analyzed the content of their re ective journals. Table 1 summarized the demographic information of the participants.

West Virginia Department of Education (WVDE) Descriptive Writing Rubric
The WVDE rubric meets the criteria for assessing descriptive writing in EFL context, by de ning reasonable cut-off scores to ensure a reliable impression of student performance in English. WVDE analytic rubric consists of ve components of organization, development, sentence structure, word choice and grammar, and mechanics, within the 1-6 band scores, ranging from 1 (Minimal) to 6 (Exemplary) spectrum. 'Organization' entails clear and logical progression of ideas in a descriptive writing. . In this study, WVDE descriptive writing rubric was selected for both teacher feedback and students' selfassessment of their descriptive writing performance. The reason behind adopting this rubric was its userfriendliness, clarity of rubric indicators, and creditability (Appendix A).

Smarter Balanced Narrative Writing Rubric
In 2014, Smarter Balanced Assessment Consortium (SBAC) and Tulare County O ce of Education created a digital library of formative assessment tools to support students and teachers in. Smarter Balanced narrative writing rubric has since been used to assess language learners' narrative writing. This analytic rubric consists of ve categories of narrative focus, organization, elaboration of narrative, language and vocabulary and conventions. 'Narrative focus' refers to the writer's effective establishment of a setting, narrator and/or characters. 'Organization' entails creating an effective plot which demands unity and completeness. 'Elaboration of narrative' refers to the narrator's thorough and effective elaboration of a narration by using details and dialogues. 'Language and vocabulary' refer to the writer's deliberate choice of words and structures that express personal experience or events. Finally, 'convention' indicates the effective and consistent use of punctuation, capitalization, and spelling in narration (Smarter Balanced Assessment Consortium, 2012). The Smarter Balanced narrative rubric has a 1-5 band scores, ranging from 0 (No evidence of the ability to write a narrative) to 5 (Meeting all the criteria of writing a real or imagined narrative). In this study, teacher feedback and students' self-assessment of narrative writing were conducted with reference to Smarter Balanced narrative writing rubric. The logic behind selecting this rubric was its clear-cut band scores and user-friendliness which made it more accessible to students' self-assessment (Appendix B).

Descriptive and Narrative Writing Elicitation Tasks
For 12 weeks, the participants were required to write and revise their descriptive and narrative drafts through their self-assessment, and regular teacher feedback. To select and incorporate 12 most favorable topics into descriptive and narrative writing elicitation tasks, a topic familiarity checklist was prepared by the researcher and distributed among the participants. To maintain the variety, the researcher split the descriptive and narrative writing tasks and assigned them into odd and even weeks. For summative assessment purposes, the initial descriptive and narrative tasks in Weeks 1 and 2 were labeled as the descriptive and narrative pretests, and the nal descriptive and narrative tasks in Weeks 11 and 12 were named as the descriptive and narrative post-tests. The topics and arrangement of the tasks were summarized in Table 2.

Student Re ective Journals
Re ective journals were written by the students after receiving the teacher feedback on their every written script and revising it accordingly. They were required to respond to four re ective prompts in their re ective journal about the writing task of the week. The prompts were prepared by the researcher and distributed in hard copies in order to collect data about how the participants engage in their (1) experience with writing portfolio system, (2) understanding of received teacher feedback, (3) ability to use teacher feedback in their revised drafts, and (4) perception of writing improvement. The students were free to write in English or Persian (students' L1). As a result, in the submitted re ective journals, around 700 words were collaboratively translated into English by the assistant researchers. The word limit in re ective journals was 500. Therefore, a corpus of around 276,000 words (46*500*12) was submitted to document analysis by the assistant researchers (Cohen's kappa (κ) = .830, interpreted as strong inter-rater reliability (Cohen, 1960)). From the written responses to every prompt, frequent themes were extracted and counted every time a similar word or concept was encountered during the content analysis.

Data Collection Procedure
The logistics of writing PA comprises four steps of collection, selection, re ection, and teacher delayed evaluation (Lam, 2013). However, the researchers are allowed to modify this framework to make it compatible to the purpose of the research or to cope with other limiting contextual factors (Hamp-Lyons & Condon, 2000). In a typical portfolio, 'collection' is the gradual compilation of students' multiple written drafts. 'Selection' is the student self-collection of best pieces of work for teacher's nal grading. Usually, in terms of re ective essays, 're ection' is the student self-assessment and self-re ection of their own personal and learning experience. 'Delayed evaluation' is assigning grades on the nal written drafts by the teacher. In this study, 'selection' was deliberately omitted to collect as much data on re ective journals as possible.
Two days before the study began, an OPT was administered as the placement test in order to normalize the selected participants for their English pro ciency level. A day before the research commenced, the researcher provided all the participants (including the EFL teacher and assistant researchers) with a 8hour tutorial on (1) the frameworks of descriptive and narrative writing by presenting two anchor essays, (2) the two selected rubrics for descriptive and narrative writing assessment (i.e., WVDE descriptive writing rubric and Smarter Balanced narrative writing rubric), and (3) the process of writing re ective journals by responding to the four prompts set for all the writing tasks in the portfolio system. The EFL teacher, assistant researchers and students and were presented with a brief discussion to the nature of teacher feedback in terms of comments, evaluation or suggestions they could give/receive without assigning grades to the writings.
The 12-week writing course was divided into six odd sessions devoted to descriptive writing tasks and six even sessions to narrative writing tasks, every second week. The participants were required to write a 300word essay on the assigned topic of the week, followed by their self-assessment before they submitted their written draft to the teacher. They were allowed to consult the selected rubrics during their writing and self-assessment. Drafting and self-assessment lasted for 60 minutes. At the end of every session, the teacher collected the papers and provided her handwritten comments, corrections or suggestions with reference to the selected rubrics, and in collaboration with assistant researchers. Every comment, correction or suggestion made on the student written drafts was considered as one 'feedback point' and the total feedback points for every draft of writing were calculated. The commented papers were returned to the students in the following week. The students were required to revise their rst draft according to the received feedback, and to write a 500-word re ective journal on their personal and learning experience, in terms of responses to the prompts. The revised drafts and re ective journals were stored by the participants for their portfolio compilation and teacher delayed evaluation. By the end of the course, the teacher evaluated the portfolios in holistic approach by assigning them the letters A, B or C, based on the overall quality of the revised nal drafts and completeness of the submitted portfolios.
To summatively assess the student writing performance, the assistant researchers scored the descriptive and narrative pre-and post-tests by counting the feedback points, with reference to the writing rubrics. The weekly teacher feedback, summative assessment of the student genre-based writing, and content analysis of the re ective journals were carried out collaboratively by the EFL teacher and assistant researchers. The whole process was supervised by the researcher in the study. In cases of rating or coding disagreement, ongoing negotiation was carried out until agreement was reached upon every occasion. (See Appendix C and D for selected samples of descriptive and narrative writing, along with teacher feedback). The inter-rater reliability indices were calculated for a variety of ratings in this study (Cronbach's α = 0.800, reliability index for descriptive writing tasks; Cronbach's α = 0.981, reliability index for narrative writing tasks; Cronbach's α = 0.881, reliability index for descriptive writing pre and post-test scores; Cronbach's α = 0.931, reliability index for narrative writing pre and post-test scores). All values of Cronbach's alpha represented strong agreement and statistical signi cant (p < .05).
The counted feedback points on students writing were keyed into Statistical Package for Social Sciences (SPSS) Version 25. The researcher conducted descriptive statistics to explore the normality of the data, two separate paired-samples t-tests to examine the student performance on pre-and post-test of descriptive and narrative writing, and repeated measures ANOVA to identify signi cant trends in teacher feedback points on the students' descriptive and narrative writing in the 12-week portfolio assessment. It should be noted that, the observed decrease in teacher feedback points was interpreted as the student progress in their writing. The signi cance level was set at 0.05 in this study. To analyze and interpret the collected qualitative data, the EFL teacher and assistant researchers carried out the inductive content analysis of the complied re ective journals over the course of 12 weeks. They schematized and reported the extracted themes out of the responses to the prompts (n = 4) set for every descriptive and narrative writing task.

Impact of PA on Students Descriptive Writing
Before running statistical tests, descriptive statistics and the assumption of normality were examined for the feedback points on six tasks of descriptive writing (Table 3). The dispersion of the feedback points was also narrowed from Task 1 (SD = 8.167) to Task 11 (SD = 6.271), which suggested an increasing uniformity in the students descriptive writing performance along the course. The assumption of the normality of the data was not violated, since the measures of skewness and kurtosis fell within the range of ± 2.00 (George & Mallery, 2010). To statistically examine the differences between the students' performance on descriptive pretest in Week 1 and post-test in Week 11, the researcher calculated a paired-samples t-test (Table 4).  Lenhard, 2016)). The ndings could be interpreted as the low effect of PA on students descriptive writing improvement. In order to graphically display the weekly trend of the feedback points on student descriptive writings, Fig. 1 was drawn with reference to the ve components in the WVDE descriptive writing rubric.
As Fig. 1 shows, the jam of feedback points were split up. At the top, two components of 'mechanics' and 'sentence structures' had a highly uctuating trend, and at the bottom, the other three components of 'word choice and grammar', 'organization', and 'development' showed a similar unsystematic trend. On the other hand, while the pattern of feedback points on the three latter components had a gradual downfall, the two latter components had a rise-up, by the end of the six-week period. As mentioned before, the decrease in the feedback points was interpreted as the sign of student writing improvement.
Since it was not clear whether the uctuations of the feedback points on the ve components of descriptive writing were meaningful over the six-week period, a set of one-way repeated measures (RM) ANOVA was conducted (Table 5). Before running RM ANOVA, the researcher calculated Mauchly's Test of Sphericity for the data which indicated that the assumption of sphericity was not violated, χ 2 (44) = .142, p = .627 > .05. and 'mechanics' (F (1, 5), 52.037, p = .000 < .05, η2 = .776). All the measures of partial η2 were interpreted as strong effect sizes (Lenhard, & Lenhard, 2016

Impact of PA on the Students Narrative Writing
Once again, descriptive statistics and the assumption of normality for the feedback points on six tasks of narrative writings were tested (Table 6). As seen in Table 6, the mean of feedback points largely decreased in number, from Task 2 (m = 28.25) to Task 12 (m = 18.06), which was interpreted as notable progress in students' performance on narrative writing tasks. The dispersion of the feedback points was also largely shrunk from Task 2 (SD = 9.581) to Task 12 (SD = 5.859), to show an increasing homogeneity in the students' narrative writing. In terms of normal distribution of the data, the measures of skewness and kurtosis were within the range of ± 2.00 (George & Mallery, 2010), which maintained the normality of the data. To examine the progress in the students narrative writing, the researcher conducted a paired-samples t-test between the students' performance on narrative pretest in Week 2 and post-test in Week 12. In Table 7, the results of paired-samples t-test were signi cant (t (45) = 3.687, p = .002 > .05., 95% CI [4.298, 16.077], r = 1.281, representing a large effect size (Lenhard & Lenhard, 2016)). They were interpreted as the positive impact of PA on student narrative writing improvement in six-week period. Figure 2 illustrates the weekly trend of the feedback points on student narrative writing, with reference to ve components in the Smarter Balanced narrative writing rubric.

Analysis of the Student Re ective Journals
An inductive content analysis was carried out with the 46 EFL students' self-reported re ective journals over the 12 week of portfolio assessment. As Table 9 summarizes, the inductive analysis of the responses to the given prompts resulted in eleven themes. The themes were extracted and categorized according to the responded prompts. The majority of responses to the rst prompt, which asked for the students' re ection on their learning and personal experience with PA, agreed upon the successful and positive impact of the PA (62.5%). 33.68% of the responses pointed to the novelty of their experience with PA, and only 3.81% of them expressed their frustration with working in portfolio system.
The encoded themes for the second prompt summarized a large proportion of the student critical engagement in the teacher feedback. 54.12% of the responses expressed an unexpected mismatch between the student self-assessment and the received teacher feedback, both in number, in type and in feedback focus. More speci cally, the students observed a sensitivity or bias in teacher assessment towards certain form-focused components in their writing such as mechanics or choice of words, at the expense of feedback to macro-components of organization or development of ideas (45.87%). Students believed that such tendency in teacher feedback narrowed down their focus to prioritize improving certain types of committed errors.
To the third prompt which asked for the student re ections on the applicability of teacher feedback in their revision process, almost half of the responses agreed on its usefulness (50.13%). The students also were satis ed with their experience of learning discourse features such as cohesive devices, generating 'good ideas' in writing, and their gradual alignment to genre conventions. Yet a large number of responses pointed to the teacher comments as incomprehensible, di cult to apply (26.38%), or inadequate (23.48%). The last prompt required the students' self-evaluation of their writing progress, to which the majority of the students positively responded (74.16%). 21.03% of the responses pointed to partial improvement of their writing, and only 4.79% of them found portfolio assessment as ineffective to their writing betterment.

Discussion
In line with the purpose of the study, three research questions were raised and explored. In research question 1, the researcher's purpose was to gain insight into the impact of portfolio assessment on student descriptive writing. The analysis of teacher summative assessment indicated no evidence for positive impact of PA on students' descriptive performance. However, further statistical results of one-way repeated measures ANOVA indicated the meaningful impact of PA on the student descriptive writing process, despite no observable improvement in their written product on the posttest of descriptive writing. The ndings implied that despite the insu cient skills and knowledge of genre-based writing, the students could engage in self-assessment, apply the received teacher feedback, and critically evaluate the quality of their writing from a descriptive-genre perspective. In other words, in the process of genre-based portfolio assessment, the students had a chance to receive feedback on both microscopic (such as 'mechanics' of writing or, 'sentence structure') and macroscopic (such as 'organization' or 'development' of ideas) aspects of descriptive writing (Borg, 2003).
In a case study of the challenges an Indonesia EFL teacher might face in portfolio assessment, Halim and Lestari (2019) also reported improvement in students' descriptive writing despite their low rate of engagement and the teacher's di culties in supervising the student peer and self-assessment. The ndings were in contradiction with Roohani and Taheri (2015) who supported the positive impact of PA on the subskills in student descriptive and expository writing achievement. However, their reported impact on the students' choice of words and conventions of writing was interpreted as weak and temporary.
In response to research question 2 which explored the impact of portfolio assessment on student narrative writing, the analysis of data from teacher formative and summative assessment indicated the positive double impacts of PA on improving the student narrative writing process, and on nal products in terms of their writing performance on the posttest of narrative writing. In other words, the students' writing progress and nal achievement suggested their constant reference to the multiple components in the selected writing rubric (i.e., Smarter Balanced narrative writing rubric), close observation of the received teacher commentaries, and successful and systematic application of them to their revised scripts.
The research literature on portfolio assessment and narrative writing dated back to the 1990s. In a case study with 22 EFL students, Shober (1996) conducted 12-week portfolio assessment and reported contrary results that only 68 percent of the students demonstrated improvement in narrative writing. Twenty-seven percent of the students' scores remained unchanged, and a single student had 5 percent decrease in her nal score. Shober (1996) concluded that portfolio assessment was de cient and ineffective as an evaluation tool. In another case study, Gearhart et al. (1992) adopted a methodological approach to portfolio assessment of 35 English-speaking elementary students' narrative writing. They reported critical issues regarding e ciency of portfolio assessment as an approach to evaluate students' narrative writing, such as controversial 'scorability' of the portfolio, disagreements over the 'domains for portfolio assessment', and its 'utility for large-scale assessment'.
Research question 3 explored the degree of student engagement in portfolio assessment of their descriptive and narrative writing performance. Regarding their general perception of writing portfolio assessment, the majority of the students agreed upon the merits and novelty of their experience. In the same vein, ndings by Herberg (2005), Pollari (2000), and Song and August (2002) showed the positive attitudes of students to PA as a learning tool for EFL writing. The ndings in student re ective journals were also in line with several studies that reported the impacts of writing PA on the students' con dence, Regarding the student engagement in the effectiveness of teacher feedback, the students reported an incompatibility between the teacher feedback and their self-assessment both in number (e.g., teacher feedback was outnumbered) and in nature (e.g., teacher feedback was more detailed, confusing and repetitive). The ndings were in line with several studies that reported the student failure in making sense of teacher feedback and, their subsequent low attention and required action (Carless, 2011;Pierce et al., 2010). The students also brought up the issue of teacher feedback sensitivity. They reported teacher sensitivity towards certain writing features, such as 'punctuation', 'description of setting in narrative', 'choice of words', and systematic leniency towards others, such as 'development of supporting ideas' or 'bringing details or examples'. The self-reports in student re ective journal were closely in line with the concept of 'didactical contract' (Brousseau, 1984). Regarding the assessment practice as a social process, the students and the teacher seem like actors interacting inside a network of mutual expectations where the students tend to 'legitimize' learning and laser-focus those areas in the new language which are bolded in teacher feedback. For example, in a classroom where the teacher's comments often center around the production of correct sentence structures or cosmetic features of language, students may interpret future feedback on the development of coherent ideas as 'illegitimate' or

Conclusion
The current study explored how genre-based portfolio assessment could impact the EFL undergraduate students descriptive and narrative writing performance and how they could engage, perceive, and act upon teacher feedback in the 12-week portfolio assessment platform. It was a case study on 46 EFL learners with some limitations in results and implications for future researchers and L2/EFL teachers. In this study, the participants were selected with non-random purposive sampling to participate in a case study. Consequently, the generalizability of the ndings in this research will be limited; yet the insight of how genre-based PA might affect the EFL learners writing at university level can provoke further research in educational settings of colleges and universities. Secondly, the collected data were limited to teacher formative assessment of student writing and the student re ective journals in 12-week portfolio assessment. The researcher strongly believes that extending the period of data collection could have yielded richer data on the student genre-based writing progress. Thirdly, more critical data could be collected from setting elicitation recall or interviewing the EFL teacher and assistant researchers, who were in charge of giving weekly feedback on the student writings. Fourthly, the students' engagement in PA was restricted to their self-reported perceptions in terms of re ective journals. Further research may add peer assessment or active collaboration of students in drafting and revising their texts, as further important sources of data on language learner engagement. Last but not least, no analytical analysis was conducted on the student revision process and their effective application of received teacher feedback, which can be a demanding topic for future research.

Funding
This research did not receive any speci c grant from funding agencies in the public, commercial, or notfor-pro t sectors.

Competing interests
Not applicable.
Author's information Natasha Pourdana is a PhD in applied linguistics and an assistant professor at KIAU, Iran. Her elds of interest are language assessment, translation quality assessment and computer-assisted language Appendixes.docx