Identifying the Linguistic Features and Content of Teacher Electronic Feedback and Students’ Text Revisions in Writing

Despite the substantial work on teachers’ feedback on students’ writing, scant attention has been paid to electronic (e-) feedback. In order to fill up this gap, the current study on teacher e-feedback provided to ten pairs of Saudi learners via E-Blackboard identified seven categories of the linguistic features of e-feedback: directives, questions, combined comments, confirmation, error corrections, praise and single statements on linguistic accuracy, content, organization and appropriateness. The students made most text revisions in responding to directive feedback, questions and combined comments, while least text revisions in responding to error corrections, suggestions and statements, and no text revisions in responding to confirmation and praise. The quantitative counts of these categories of feedback and students’ responses demonstrate the prevalent categories of feedback and text revisions. Most of the teacher’s electronic feedback focused on issues related to students’ use of the language in writing, followed by content and idea development, appropriateness and organization. Similar results were obtained from the quantitative analysis of students’ text revisions. This could be due to students’ low language proficiency and inadequate language competence in English, which hindered them from using accurate language in their assignments. Cross-referencing of students’ text revisions to the various patterns of teacher e-feedback revealed that most of students’ text revisions resulted from directive feedback, questions and combined comments. The study offers pedagogical implications for teacher feedback practices and students’ responses to feedback.


INTRODUCTION
Teacher written feedback has attracted the attention of many researchers in higher education in both English as a second/foreign language (ESL/EFL) contexts. It is central to effective language teaching and learning, especially at university (Bacquet, 2019;Chen, 2009;Rassaei & Moinzadeh, 2011). It is a way to respond to students' issues and errors in their written tasks (e.g., Alvarez, Espasa & Guasch, 2012;Bader, Burner, Iversen & Varga, 2019;Ferris, 1997;Hyland & Hyland, 2001). Studies have identified the various patterns of teacher feedback, including questions, suggestions, clarifications, statements and other types that show how teachers compose their feedback (Alvarez et al., 2012;Ferris, 1997;Hyland & Hyland, 2006;Terglia, 2008). Studies have also looked at the foci of teacher written feedback and revealed that teachers' feedback varies in its foci from content, organization, language to other areas of writing such requirements or appropriateness. Although such research has provided better insight into teachers' patterns of written feedback, the question of how teachers should formulate their feedback to address errors/issues in their Published by Australian International Academic Centre PTY.LTD. Copyright (c) the author(s). This is an open access article under CC BY license (https://creativecommons.org/licenses/by/4.0/) http://dx.doi.org/10.7575/aiac.ijalel.v.9n.6p.82 students' writing needs to be further addressed in research (Russell, 2009).
The problem is related to how EFL students, in particular Saudi EFL learners, find writing in English challenging. Like many other EFL learners (Boubekeur,2015), Saudi learners encounter difficulties in writing in English, including inappropriate vocabulary, inaccurate grammar and disorganization of ideas (Al-Khairy, 2013;Javid, & Ume, 2014). As a result, teaching writing to EFL Saudi learners becomes challenging and complicated for teachers and instructors (Alhaisoni, 2012). This suggests the need to look for more effective ways to assist and help Saudi learners overcome these challenges and improve their writing in English. One of these effective instructional strategies to help learners in writing is using feedback (Kang & Han,2015;Langer, 2011).
Another aspect of the problem is relevant to the gap in early research. Despite the large volume of research on identifying the patterns of teacher feedback (e.g., Ferris, 1997;Hyland & Hyland, 2001;Zoghi & Nikoopour, 2014) and students' perception of or reactions to teacher feedback Identifying the Linguistic Features and Content of Teacher Electronic Feedback and Students' Text Revisions in Writing 83 (e.g., Vasu, Ling & Nimehchisalem, 2016;Zarifi, 2014), most of the previous research has focused on handwritten feedback, while only a few studies have explored teacher electronic (e-) feedback on students' writing (Ene & Upton, 2014;2018). Teacher e-feedback refers to the feedback provided by teachers via technological tools, such blogs, E-Blackboard, chats and others (Elola & Oskoz, 2017;Ene & Upton, 2018Tuzi, 2004. Findings of these studies on teacher e-feedback revealed the positive impact of e-feedback on students' text revisions. Yet, which linguistic patterns of e-feedback are effective at encouraging learners to substantially revise their texts have not been explored.
In other words, questions related to how teachers should linguistically formulate their feedback and what errors and flaws they should address need to be further explored.
Although much research has been devoted to teacher feedback, there is a need to further explore whether students productively use teacher feedback (Ma, 2018). Therefore, the current study aimed to identify the linguistic features of teacher e-feedback and students' text revisions in responding to the e-feedback. Being situated in an undergraduate class of 20 Saudi students, the study attempted to answer the following specific research questions: (1) What are the patterns of teacher e-feedback in relation to its linguistic functions and content provided on students' writing? (2) To what extent do the students revise their writing in responding to teacher written various linguistic functions of written feedback?

Teacher Written Feedback
Investigation of teacher feedback on ESL/EFL students' writing has focused on the various patterns of teacher's/ instructor's feedback by analyzing the pragmatic or linguistic functions and content/foci of feedback and come up with various categories of feedback. In this section, we discuss the literature review on these aspects of teacher's feedback accordingly. For the linguistic functions of feedback, Ferris (1997) reported that teacher written feedback falls into questions, statements and imperatives that request students to revise their written texts. In addition, question was found as the most frequently used type of feedback. Hyland and Hyland (2001) identified three categories of teacher's feedback: praise, criticism and suggestions with praise as the highest category, followed by criticism and suggestions. However, the number of the three categories differed between the two case students who received the feedback and between the first and final drafts. Teachers also provided feedback in the form of combined comments (e.g., critical remarks with praise or suggestions). Other patterns of feedback are hedges that play a role in softening the teacher's critical tone or mitigating the relational damage caused by criticism. In Hyland's and Hyland's (2006) case study, the teachers' feedback most frequently emerging patterns of feedback are praise, followed by criticism and suggestion, which indicates that both teachers were attempting to create a positive learning environment for students' improvement of writing and development of their confidence in writing. Terglia (2008) identified four categories of teacher feedback: praise, mitigated comments, directive and others (e.g., clarifications and funny remarks). The highest number of feedback was directive, followed by praise, mitigated comments and others. According to Terglia (2009), teacher feedback falls into suggestions, clarifications, praise, providing information and requests. The most frequently used type of feedback was request, whereas clarifications and suggestions were the least frequently used patterns of teacher feedback. As for Alvarez et al. (2012), suggestions and corrections were identified as the most dominating types of teacher feedback, while questions, clarifications and evaluations were found with the lowest rates of use.
Researchers have also identified patterns of teacher feedback on students' writing by looking at the content or foci of feedback in an attempt to understand the issues and flaws addressed by teachers in students' writing. There are some researchers who argue that teachers should highly focus on global issues of students' writing, including idea and content development and organization and coherence (e.g., Ferris, 1997;Hyland & Hyland, 2001). Their argument is based on the claim that teacher's feedback highly focusing on local issues or forms such as grammar "may no longer accurately describe the practices of modern composition instructors" (Bitchener & Ferris, 2012:79). On the other hand, some other researchers argue that teachers should make a balance in focusing their feedback on both global and local issues such as language and grammar (e.g., Ashwell, 2000). This argument is based on what issues the teacher actually identifies in students' writing.
There are also studies which conducted an empirical analysis of the foci of teacher feedback. For instance, Brice (1995) found that the teacher feedback addresses issues related to content, organization, grammar, vocabulary and conventions in both drafts of their writing. The number of teacher feedback varied among the three cases of students. In addition, Ferris (1997) reported that most of the teacher feedback focuses on content and ideas rather than grammatical issues. For Hyland and Hyland (2001), teacher feedback targets ideas, form, academic, process and general. According to Terglia (2009), most of teacher feedback addresses issues related to ideas followed grammar, form, generic, process and heading. Findings reported by Hyland and Hyland (2006) indicated that the majority of teacher feedback identifies flaws and errors relevant to content of writing, followed by feedback on form, academic aspects of writing, process and finally feedback that was described as general.
All the above-mentioned studies have focused on teacher handwritten feedback rather than e-feedback. There are only a few studies that have looked at teacher e-feedback (Elola & Oskoz, 2017;Ene & Upton, 2018Tuzi, 2004, the latter two of which have shown a positive impact of teacher e-feedback on students' subsequent drafts of writing. In addition, the majority of e-feedback serves comments, such as IJALEL 9(6):82-90 directives, explicit instruction and others that are based on students' issues in writing (Ene & Upton, 2014).

Students' Text Revisions
Researchers have showed a great interest in how students revise their written texts in responding to teacher feedback in attempt to gain a better understanding of the patterns of teacher feedback that effectively promote students' substantive and meaningful text revisions. Known as uptake of teacher feedback, it includes students' reactions to feedback (Zoghi & Nikoopour, 2017). According to Ferris (1997), feedback linguistic features, such as request for information and summary comments led to substantive text revisions among students, whereas questions and statements resulted into less text revisions and positive comments did not lead to any text revisions. Conrad and Goldstein (1999) found that declaratives result into more revisions than imperatives and questions. In a study by Terglia (2009), single requests and giving information generate most text revisions. Furthermore, Hyland (2003) found that only three students made a high proportion of the teacher's form-focused feedback in revising their texts, while the other three students' use of feedback was medium and low. The researcher attributed this to students' beliefs about the importance of teacher feedback on content as opposed to form-focused feedback on language and grammar. According to Mahfoodh (2017), the majority of students accepted teacher written feedback and integrated it into revising their writing. This study also showed variations in the rate of students' integration of teacher feedback according to the type of feedback as ordered from the highest to the lowest: giving information, grammar editing, direct coded, making requests and reflective statements, whereas praise and negative evaluation did not lead to text revisions.
Other recent studies (Han & Hyland, 2015;Hyland, 003;Zhang & Hyland, 2018;Zheng & Yu, 2018) focused on students' behavioral engagement as one dimension of their engagement with feedback by analyzing text revisions and operations made by students in their writing in responding to feedback. The findings of this study showed variations in the extent to which individual students engaged with text revisions owing to several factors, including their English proficiency, their metacognitive strategy use in revising their texts and the type of feedback and its nature: direct or indirect. In relation to teacher e-feedback, Tuzi, 2004) reported that most of the students' changes to their texts, especially additions of new information to the content resulted from e-feedback. Moreover, e-feedback had a positive effect on students' text revisions at the macro-level structure, such as clause, sentence, and paragraph levels. The researcher concluded that e-feedback is more efficient for it results into sentence and paragraph level-changes. Similarly, most of e-feedback was addressed by students and resulted into substantial text revisions of the contents of writing (Ene & Upton, 2018). On the other hand, e-feedback was effective in eliciting students' appropriate text revisions of surface-level issues, including grammar and sentence structures (Ene & Upton, 2014). Yet, these later studies on teacher e-feedback have not addressed the efficacy of each pattern of e-feedback on students' subsequent.

Study Setting and Participants
The present study was conducted among 20 EFL Saudi undergraduates joining the English Department at a public Saudi university over an academic semester. As third year-students at the university, the students need to write essays as part of a continuous assessment of their performance in the academic writing course. Specifically, the students joining this course were required to write a four paragraph essay. In order to accomplish this assignment, students were guided through the writing procedure that covers several steps starting from the planning to the revision and submission of the final drafts of the assigned essay (Week 13). The participants were all female students owing to the segregation of university students in the country. They were all willing to receive feedback from the instructor and revise their assignments accordingly. This was not the first time for them to work on pair writing and revise their text based on feedback as they had been engaged in similar activities in another writing course the previous academic semester. The course instructor is a a female Saudi instructor who is an experienced lecturer.

The E-Feedback Activities
Prior to the study procedure, the students were instructed on the pair assignment. They were also divided into 10 pairs, each pair had to work on one essay. They also had to choose their topics of interests for the essays relevant to the course and report it to the instructor earlier. The activities were also scheduled as in Table 1: During the 5th week, students started writing the first drafts, were taught how to use the course E-Blackboard for the e-feedback activities and were trained on how to reply to the instructor's comments and uploading their written texts. However, at this stage, the written texts uploaded by students were not yet related to their texts used as data in this study. Training through explicit instruction continued till the 8th week, when each pair of students had to upload their first written draft for the instructor to read. The first stage of feedback and revision of writing started from the 9th week till Table 1. The E-Feedback Activities

Time-line Activities
Week 1 Instruction on the assignments of the course Week 2-4 Topic selection and pair divisions Week 5-8 Writing the first drafts and uploading them on E-Blackboard Discussion Week 9-10 Provision of feedback on the first drafts and students revising their drafts Week 11-13 Provision of feedback on the second drafts and students revising their drafts

Week14
Students' submission of the final drafts for assessment

Identifying the Linguistic Features and Content of Teacher Electronic Feedback and Students' Text Revisions in Writing 85
the 10th week, a period during which each pair of students received feedback through written comments and had to start revising their first draft using highlights to track the changes made to their assignments. This stage ended with students' accomplishment of their second drafts. The second stage of feedback started from the 11th week till the 13th week, during which the instructor provided feedback on the second draft of each pair. Each pair of students had to revise their second drafts based on the feedback. This stage ended with students' completion and submission of the final drafts of assignments for assessment. The last week of the semester, just before the final exams, was dedicated for submission of the final drafts.

Data Collection and Analysis
The data was collected from teacher's written e-feedback and students' text revisions. For the written feedback, the instructor's written comments on students' assignments was imported from the E-Blackboard to word files, each of which consists of each pair's assignment and comments attached to it. Each file also consisted of each pair's first draft and revised drafts which would be later used for analyzing the text revisions made by the students.
The data analysis was an iterative process that involved reading and getting familiar with the written data, developing codes from previous research, coding and re-coding of the data, categorizing and coming up with themes and finally, comparing and checking inter-rater reliability of coding between the two independent coders, and reporting findings in the form of categories and themes with samples.
Once all data was saved and organized into word files, the data analysis was initiated by developing various codes. For the e-feedback, first, it was coded in relation to its linguistic functions (e.g., statement, questions, and so on) developed from several previous studies on feedback (Appendix 1). The second round of coding feedback was intended to code each written feedback in relation to its content or foci-the issue (s) in academic writing the feedback addresses. So based on the literature with some modifications of codes to suit the purpose of assessment in the course, each written feedback was coded in terms of its foci (e.g., insufficient details, inaccurate use of tense, missing punctuation, etc).
For students' text revisions, they were tracked through the track change of Microsoft word and each draft was compared to its previous draft using the comparison feature of the Microsoft file in order to find if there were more changes which were not highlighted by students. Then, the changes varying from word levels to paragraphs were coded according to the foci of each written feedback attached to them: content, organization, linguistic accuracy and appropriateness (Appendix 2). For the transcripts of interviews, the data was also read and a thematic analysis was used within previous research on students' reactions to feedback, and new themes emerged.
Following this was cross-referencing students' text revisions to teacher written feedback in order to see how students responded to the various linguistic patterns of feedback through text revisions and also identify the patterns of feedback resulting into learners' highest and lowest numbers of text revisions. The final stage was conducting a quantitative enquiry of the patterns of written feedback and text revisions as well as the cross-referenced text revision-feedback. Moreover, this involved counting the numbers of these categories among the 10 pairs of students. This stage ended with checking inter-rater reliability of coding by comparing the codes made by each coder. Differences were discussed and an agreement of 90.14% was achieved.

What are the Patterns of Teacher E-feedback in Relation to its Linguistic Functions and Content Provided on Students' Writing?
The findings revealed that the instructor provided an overall number of (786) written feedback on students' written assignments over the study period. As shown in Appendix (1), the teacher linguistically formulated his feedback in different linguistic functions: error corrections and comments combining various pragmatic functions (e.g., question and statement, suggestion and question and so forth). He also provided suggestions or comments carrying out some advice on certain issues in students' writing and used questions seeking students' clarifications, reasoning or even requesting them to provide the accurate remedy or revision and directives that ordered students to take actions or fix their issues in writing. Other linguistic functions of teacher written feedback are inclusive of comments acting as confirmation of students' text revisions, praising them for their successful attempts in revising their writing and even providing them with single statements such as explanations of specific aspects of their writing, clarifications and evaluations of writing.
Based on the quantitative enquiry of the various linguistic functions of teacher written feedback (Table 2), directive feedback was the most dominating pattern of teacher written feedback provided on students' writing (194), followed by questions (144), combined comments (135), confirmation (84) and praise (78). On the other hand, the least frequently used types of feedback are error corrections (53), suggestions (51) and statements (47). The results also show the varying distributions of these linguistic functions of teacher feedback among the 10 pairs of students. In brief, while pair 1 received the highest number of teacher feedback (152),  (47). In addition, the rate of each category received by the pairs of students also differed as for error corrections, combined comments, confirmation and praise, pair 1 received the highest numbers of these two patterns of feedback, whereas for other remaining patterns of feedback: suggestion, question, directive, and statement, pair 10, pair 4 and pair 5 (the same highest number), pair 5 and pair 9 (the same highest number) and pair 2 and pair 6 (the same highest number) received the highest numbers of these four types of feedback, respectively. Since the issues addressed by teachers through feedback are important, an analysis of the foci of teacher feedback was performed in this study. Various issues and flaws in students' writing addressed by the instructor through written feedback were identified. As shown in the Appendix, the teacher feedback targeted issues related to content and idea development (e.g., insufficient information or details), organization and structure (e.g., flow of ideas), linguistic accuracy that includes accurate grammar, appropriate vocabulary and correct sentence structures as well as spelling and punctuations and finally appropriateness (e.g., formatting).
Calculation or counting the various patterns of teacher feedback in relation to its foci (Table 3) demonstrates that the teacher paid most attention to issues and flaws related to students' linguistic accuracy (347) in academic writing of assignments, followed by those issues relevant to content and idea development (261), appropriateness (114) and finally organization (64). Again such varying distribution of the foci of teacher written feedback among the pairs of students is an indicator of the varying number of issues and flaws identified by the teacher in each pair's academic writing of assignments. For instance, while pair received the highest numbers of feedback focusing on content and idea development as well as linguistic accuracy, the highest numbers of feedback focusing on organization and appropriateness were received by pair 6 and pair 4, respectively. There are also variations of the extent to which each pair received the four categories of the foci of teacher feedback.

To What Extent do the Students revise their Writing in Responding to Teacher Written Various Linguistic Functions of Written Feedback?
For the second research question, the learners responded to teacher written feedback by revising their academic writing. Analysis of students' drafts of assignments showed that the students made various changes or text revisions at the paragraph, sentence, clause, phrase and word-levels. The samples of text revisions provided in Appendix (2) demonstrate that students' various text revisions fall into content and idea development, organization, linguistic accuracy and appropriateness. Moreover, our quantitative enquiry of students' text revisions (Table 4) indicates that most of text revisions made by students are local text revisions addressing issues and flaws related to the linguistic accuracy of their writing (267). The second highest category of text revisions made by students is content and idea development (193), followed by revisions of issues pertinent to appropriateness (92). How-ever, organization received the least attention from students through their text revisions. The order of the numbers of text revisions in relation to these four categories of issues in writing reflects the same order of the foci of teacher feedback. So organization, for instance, received the lowest number of text revisions made by students is owing to the teacher's lowest number of feedback focusing on organization.
The students' above text revisions were also cross-referenced to the various linguistic functions of teacher feedback in an attempt to better understand the patterns of feedback leading to more text revisions (Table 5). It is evident that as the teacher formulated his feedback in the forms of directives, questions and combined comments, he could attract students' highest attention to text revisions (190, 133 & 127), respectively. Error corrections led to 53 text revisions, while suggestions and statements resulted into 48 and 47 text revisions, respectively. On the other hand, feedback acting as confirmation and praise turned out to be non-revision-oriented feedback for both of them did not lead to any text revisions. This could be due to the nature of both patterns of feedback (such feedback was just intended to confirm the accurateness of students' certain text revisions and praise them for successful text revisions rather than targeting issues in their writing).

DISCUSSION
The present study attempted to address the question on the patterns of the linguistic features and content of teacher e-feedback. The findings of the present study shed lights into the various patterns of teacher written feedback on students' in terms of its linguistic functions and foci. The teacher could vary his feedback linguistically from single comments functioning as error corrections, suggestions, questions, directive, conformation, praise to statements. This particular finding is almost similar to findings of several earlier studies (e.g., Alvarez et al. 2012;Ferris, 1997;Hyland & Hyland, 2001;Hyland & Hyland, 2006;Terglia, 2008;). In line with Hyland's and Hyland's (2001) study, this study adds to previous knowledge on the linguistic functions of teacher feedback by revealing one more category, which is combined feedback. In formulating his feedback in the form of combined comments, the teacher could combine two (e.g., suggestion & question, statement & question, praise & suggestion) and even more than two linguistic functions (e.g., statement, question & suggestion). This category of feedback provides learners with further information or explanation of the issues or flaws detected in their written assessments. It also helps to mitigate the imperative tone of the teacher in his directives especially in cases when directives are combined with praise. Regarding the most frequently used type of feedback in terms of its linguistic functions, the findings show that directive is the most dominating type of feedback, followed by questions and combined comments. This result is in agreement with the findings of some previous studies (Ferris, 1997;Terglia,2008), which could be due to the topdown approach used by the instructor of this study, which reflects that the teacher has the power over his students. Our findings also think better revealed to be consistent with the sentences in the discussion that confirmation of students' text revisions and praise are used by the instructor with almost a medium level of frequency of occurrence. This suggests that the instructor' providence of these two types of positive feedback could be due to the need for encouraging students to be self-confident about the accurateness of their text revisions and praise them for their successful text revisions. According to Hyland and Hyland (2001), the use of praise in teacher feedback is influenced by the way or style teachers respond to issues in students' writing and their system beliefs. In addition, the finding of this study illustrated the lowest use of error corrections and statements. For error corrections, this particular finding contradicting the finding of Alvarez et al. (2012) that error corrections as the second most dominating category of feedback suggests that the instructor in our study did not highly offer students with spoon-feeding since error corrections are the most direct way of feedback which shows teachers' direct intervention in students' text revisions.
Part of the first research question is exploring the patterns of teacher written feedback in relation to its foci. This is an important aspect of feedback investigation since it reveals the content or information delivered to students by teachers. In this regard, the teacher paid most of his attention to students' linguistic accuracy, followed by content and idea development, appropriateness and organization. This finding is not consistent with findings of most previous research reviewed in this study. Studies reported that most of teacher feedback targeted global issues related to content and idea development (Ferris, 1997;Hyland & Hyland, 2001;Hyland & Hyland, 2006;Terglia, 2009). This could be due to the students' highest cases of failure to accurately use English, such as grammar, vocabulary choice, sentence structure and mechanics (spelling and punctuations), when writing the first drafts of their assignments. In other words, As EFL learners, the participants' language proficiency seemed low and their language competence in English could be inadequate, which prevented them from using an accurate language in their assignments. Similar results were reported by Zheng and Yu (2018) that students' lack or inadequate linguistic knowledge is a major factor behind their language errors or issues in writing and their unsuccessful text revisions.
The study also addressed another important question relevant to the extent to which students revise their texts based on teacher feedback. In this respect, students' responses to teacher feedback provide insights into whether students are responsive to feedback and the extent to which they are willing to revise their writing once receiving feedback from teachers (Alvarez et al. 2012). The way students respond to teacher feedback through text revisions corroborates behavioral engagement with teacher feedback cited in a few recent studies (e.g., Zhang & Hyland, 2018;Zheng & Yu, 2018). So in this study, most of students' text revisions are intended to fix issues related to linguistic accuracy, followed by content, appropriateness and organization. This indicates that the students engaged with teacher feedback in all aspects of writing and most highly engaged with teacher feedback focusing on the use of accurate language in their writing. This also supports the finding of Mahfoodh (2017) that the rate of students' acceptance of teacher feedback is high and that most of text revisions focused on surface-level issues, including grammar and sentence structures (Ene & Upton, 2014). Such finding is contradictory to findings on e-feedback (Ene & Upton, 2018;Tuzi, 2004) that most of students' text revisions focused on macro-level issues, including content. Another possible reason behind this particular finding could be students' willingness to revise their writing, which is consistent with Zheng and Yu (2018) that such factor affects students' responses to feedback and integration of it in revising their writing. They could also highly follow the instructor's feedback in revising their writing because they were concerned about the quality of their assignment that would be later graded as part of their performance in the course.
What is of interest in this study is exploring the patterns of feedback resulting into students' highest and lowest numbers of text revisions. Based on the results obtained from cross-referencing of students' texts revisions to the linguistic functions of teacher feedback, effective teacher feedback that facilitates students' text revisions falls into these patterns: directive feedback, questions and combined comments. Explaining this finding in light of the literature review, Terglia (2009) raises the questions related to whether teachers/instructors compose their feedback in the form of directives due to their feeling IJALEL 9(6):82-90 that students may fail to understood their intended messages or their use of such directives reflect the perceived literacy skills of their students and whether teachers' use of feedback in the form of directives is a mere spontaneous reaction to authentic readers/audiences. In this regard, the finding of the present study suggests that the teacher's overuse of directives could be due to its role in triggering students' engagement with text revisions, which is contradictory to the finding of Conrad and Goldstein (1999) that declaratives promote more revisions than imperatives. For the use of questions, such feedback results into students' second highest number of text revisions. This supports what was stated by Ferris (1997). According to the author, teachers are recommended to compose their feedback in the form of questions for its role in stimulating students' thinking processes and avoiding appropriating of students' written tasks. However, questions may not necessarily lead to text revisions rather than asking students to provide information such as clarifications or even justifications. Similarly, according to Ferris (1997), questions lead to less text revisions. Moreover, combined comments are conducive to a relatively high number of text revisions. Alvarez et al. (2012) found that, students highly respond to combined feedback as they discuss the changes made to their texts. In line with Ferris (1997) and Mahfoodh (2017), positive comments such praises do not result into any text revisions because they focus on the positive aspects of students' writing and text revisions rather than addressing issues and weaknesses in students' writing.

IMPLICATIONS AND CONCLUSION
The current study aimed to identify the linguistic features and content of teacher e-feedback. It also aimed to determine the extent to which students revise their essays in responding to the feedback. Results indicate that the instructor varied the way he linguistically formulates his feedback and addresses issues and flaws in students' written essays. Similarly, the extent to which students responded to feedback in revising their essays differs according to the linguistic features and content of feedback. In conclusion, although the above findings enrich our understanding of teacher feedback and its role in engaging learners in revising their written assignments and reflecting on their reactions to feedback since this is one of the rare studies devoted to teacher e-feedback, there are several limitations of this study that should be addressed for future research. First, the current study focused on a number of 20 Saudi undergraduates in a particular course, and therefore, the findings should be interpreted by researchers with caution when used in other contexts. Another limitation of this study is that the feedback explored was provided by one EFL instructor. However, inclusion of feedback provided by different instructors in future research will provide better insights into the varying styles of instructors and its impact on their formulation of feedback. Moreover, the assignments were written in 10 pairs of students rather than individually so this did not allow to find out individual differences that might have affected students' responses to feedback.
Finally, students' reactions to feedback should be also explored in future research as part of their responses to e-feedback. This can be achieved through recording of students' reactions to feedback once they receive it. Another way to achieve this is to allow and encourage students to respond to each e-feedback through a written comment in which they express their reactions and evaluations of e-feedback. They can be also encouraged to record what they feel and how they will deal with such feedback. By so doing, it will be possible to explore students' cognitive engagement with teacher feedback, which is defined as one dimension of learners' engagement that refers to the extent they cognitively engage with feedback, use cognitive and metacognitive strategies, plan, monitor and make decisions about teacher feedback (Han & Hyland, 2015;Hyland, 003;Zhang & Hyland, 2018;Zheng & Yu, 2018).